Foresight & ethics, run on one case ● 27 methods · 1 product · 0 predictions

Fokus 2040
The attention company

Fokus Labs makes a headband that reads schoolchildren's brainwaves and sells teachers a live attention map. In September 2026 the board asked one question: what are we actually building toward? This journal answers it with a full fdbx foresight chain (scan, deepen, map, experience, plan, work back) and all twenty-one edbx ethical-design methods, pointed at the same object.

6 fdbx methods 21 edbx methods 20 sourced signals, each linked to its source 2040 horizon 3 red lines the case cannot survive without

Vol. 2. Vol. 1, Sofra 2040, ran the same toolkit on a restaurant Every Fokus number is a labelled fiction Every signal link was live when this was built

Fokus · Room 2B · Period 3Live

Fig. 0 · What the teacher's tablet shows: thirty seats, one light each. Illustration; the room and readings are fictional.

01 · THE CASE INPUT DOSSIER

A company, a headband,
and a brief with one future in it

Everything downstream works on this object. Fokus is a fictional composite; the industry around it is real. Where a number belongs to Fokus, it is typed ASSUMPTION; where a claim touches the real world, it carries a source or waits in a leads to verify list.

The company

LegalFokus Labs B.V., Eindhoven (NL) — fictional composite

Founded2023 · 142 staff · Series B €40M, March 2026 A

ProductFokus Band 2 — three dry EEG electrodes (forehead + mastoids), on-device scoring

What it computesA per-minute “attention score” from alpha/beta ratio, blink & motion artefacts A

BuyerSchools, €18 per pupil per month A

Footprint today12 pilot schools, ≈4,200 pupils, NL & Flanders A

SurfacesTeacher dashboard (live class heatmap) · pupil app (streaks) · parent app (weekly Focus Report) · “Mentor” AI recommendations

The people in the room

CEOMarit Verhoeven

CPOJonas de Wit (owns the roadmap)

Learning scienceDr. Amira Bakkali

Privacy / DPOLotte Donders

Teacher partnerSanne, vmbo school De Brug, Tilburg

Pupil personaMilan, 14, second-form vmbo

The brief, verbatim — read as the litany in §03

Design the road to 2040: grow Fokus from 12 pilot schools to 2,500 schools across Northwest Europe. By 2040 every teacher should see attention in real time, every parent should receive a weekly focus report, and attention data should power personalised learning paths for every child.

Success = ≥90% attention-score coverage of lesson minutes · ARR €220M by 2035 · school churn <2% A

Six design decisions are already made and the ethics wall (§07–10) will hold them up one by one:

D1 — Live heatmap. The teacher sees each named pupil's attention, per minute, in real time.
D2 — Streaks & leaderboard. Pupils earn a personal “focus streak”; classes compete on a visible leaderboard.
D3 — Parent Focus Report. Weekly report to parents with the pupil's percentile versus the school year.
D4 — Retention & reuse. Attention data is kept 24 months and reused to improve scoring models; opt-out exists but is three menus deep.
D5 — Mentor & the medical edge. The AI suggests interventions and “ADHD-adjacent insights” to schools — untested as a medical device.
D6 — Research & insurer pilot. “Anonymised” lesson data is shared with two university labs and a health insurer's wellness programme.

2040

Horizon year

12 → 2,500

Schools, brief's ambition

Decisions already made

Future in the brief (Continued Growth)

How to read this journal

§02–06 (fdbx) widen the time horizon: what's changing, what the brief is really about, which forces are in play, what four very different 2040s feel like, how to get from here to there, and what to do Monday morning.

§07–10 (edbx) audit the present in four bays: whose brain is read, what could go wrong, how the team should decide, and what it promises in writing. §11 walks home from a preferred 2040 (and back from a feared one); §12 shows the receipts — the evidence ledger, method provenance, and what each method actually changed.

02 · THE SIGNAL SHELF FDBX · HORIZON SCANNING

What is already changing
around this product

Run as the UK Government Office for Science Futures Toolkit defines it: a scoping question first, every signal carried with its source, breadth checked against PESTLE, then clusters, ratings and drivers. Mode: scan. Signals were gathered by live search in the session that wrote this file; nothing comes from model memory unmarked.

fdbx-horizon-scanningGO-Science Futures Toolkit (2024), pp. 34–41 · scan mode · searches run this session

Scoping question: What changes in the next fifteen years could reshape how 12-to-18-year-olds learn to pay attention — and what would each change mean for a company selling attention measurement to their schools?

Signals from: user-supplied (0) · sourced (20). PESTLE balance: Legislative 7 · Societal 4 · Technological 6 · Economic 3 · Political 2 · Environmental 0 — a real gap; no environmental signal for classroom neurotech was found in this session's searches, so nothing is invented to fill it (see Leads to Verify).

#	Signal	Source	Date	Type	PESTLE	Why it matters
1	UNESCO member states adopt the Recommendation on the Ethics of Neurotechnology — the first global normative framework for neurotech, naming mental privacy and freedom of thought.	unesco.org — adoption announcement	Nov 2025	Emerging issue	L	Gives regulators, schools and parents a citable standard before Fokus scales. “Mental privacy” becomes a procurement question.
2	Full text of the UNESCO Recommendation published — covers brain–computer interfaces and consumer neurotech, mandates ethical guardrails for brain-activity access and manipulation.	unesco.org — recommendation text	Nov 2025	Emerging issue	L	The text schools' legal teams will quote. Fokus can pre-align or be forced to retrofit.
3	Chile becomes the first country to write neurorights into its constitution (Art. 19, Law 21.383 of 2021: protection of brain activity and the information derived from it); in 2023 the Supreme Court ruled against Emotiv over brain data. The implementing neuroprotection bill is still before Congress.	courier.unesco.org — “Chile: Pioneering the protection of neurorights” · law.stanford.edu — on the Emotiv ruling	2021–26	Emerging issue	P	Constitutional-level precedent: neurodata as a rights question, not just a privacy setting. Blueprints for EU debates.
4	Colorado HB24-1058 makes Colorado the first US state to classify neural data as sensitive data under consumer privacy law.	leg.colorado.gov — bill page	Apr 2024	Emerging issue	L	The “neural data is just biometric data” defence is dead in at least one major market.
5	US neural-data law wave: California SB 1223 (in force 1 Jan 2025), Montana (in force Oct 2025), Connecticut (2025): four states by mid-2025.	fpf.org — four-state count · kffhealthnews.org — state laws overview	2024–25	Emerging issue	L	A compliance patchwork is forming at state level; multi-state edtech can't ship one privacy posture.
6	The EU AI Act's Article 5 prohibitions take effect 2 Feb 2025: AI that infers emotions is banned in workplaces and education except for medical or safety reasons. Fines of up to €35M or 7% of global turnover (Article 99) apply from 2 Aug 2025.	artificialintelligenceact.eu — Article 5 · Article 99	Feb 2025	Risk / opportunity	L	Directly aimed at Fokus's category: inferring cognitive/emotional states of pupils. The medical exception is both shield and trap (device-classification risk).
7	US regulators and scholars still argue over what counts as “neural data” — the definitional Goldilocks problem (derived vs recorded signals).	fpf.org — defining neural data	Aug 2025	Emerging issue	L	Fokus's attention score is derived — the exact category some definitions exclude. The legal map is unfinished; assume it tightens.
8	BrainCo pitches the “Focus 1” EEG headband to gather student brainwave data and measure classroom engagement.	edsurge.com — “This Company Wants to Gather Student Brainwave Data…”	26 Oct 2017	Established trend	T	Fokus is not first. This product category exists — and has a public history schools remember.
9	A Hangzhou school trials BrainCo's Focus 1 headbands with a teacher portal (Apr 2019). In Oct 2019 video of headbands at a Jinhua primary school, with readings sent to teachers and parents, sparks international backlash, and the local education bureau suspends their use.	thechinaproject.com — “Chinese parents want students to wear dystopian… headbands” · qz.com — Jinhua backlash, Nov 2019	Apr–Nov 2019	Established trend	P	The category's cautionary tale: parents revolted at attention surveillance of children. A playbook for how a Dutch rollout could die.
10	A Dutch nationwide agreement bans mobile phones in secondary classrooms (Jan 2024); a government-commissioned study finds improved pupil concentration, with ~three-quarters of 317 surveyed schools reporting positive effects.	reuters.com — coverage of the July 2025 study, 4 Jul 2025 · bbc.com — “Has banning phones improved performance at Dutch schools?” (Mar 2026)	2024–26	Established trend	S	Buyers (schools, ministry) already treat attention as something you fix by removing technology. Adds more of it — on children's heads — is the opposite bet.
11	Gloria Mark's logging research: average time on one screen fell from ~150s (2004) to ~75s (2012) to ~47s today (APA Speaking of Psychology, Feb 2023). In interviews for her book Attention Span (2023) she puts the return to an interrupted task at roughly 25 minutes.	apa.org — Speaking of Psychology, ep. 225 · gloriamark.com — book page	2023	Established trend	S	The cultural anxiety Fokus monetises is real and measured. But it also explains why “more dashboards” may worsen the 25-minute problem.
12	WHO/HBSC data: ~11% of adolescents show problematic (addictive-like) social media use, in the 2021/22 survey, up from 7% in 2018; girls 13%, boys 9%.	who.int/europe — “Teens, screens and mental health”, 25 Sep 2024	Sep 2024	Established trend	S	Public willingness to “do something” about kids' attention is high — the political energy Fokus sells into. Also: the cohort most sensitised to surveillance.
13	A scoping review of 16 classroom EEG studies finds alpha-band activity is the most-used marker of attention in real lessons, which is known to track attention inversely; the authors call the evidence preliminary, with contradictory findings and small samples.	pmc.ncbi.nlm.nih.gov — Zeng et al. 2025, scoping review	2025	Expected development	T	The signal Fokus claims is being studied, but the classroom evidence is thin. It supports the physics, not the product.
14	Peer-reviewed comparisons find significant spectral differences between consumer-grade and research-grade EEG — consumer devices “need further validation”.	pmc.ncbi.nlm.nih.gov — Mikhaylov et al. 2024	2024	Risk / opportunity	T	The published validity gap between what Fokus wears and what the literature trusts. A claims-discipline problem, not a footnote.
15	Dry-electrode consumer headbands vs gel-based research systems: usable for resting-state and coarse attention metrics, questionable for clinical or fine-grained use.	frontiersin.org — Kleeva, Ninenko & Lebedev 2024	2024	Risk / opportunity	T	Draws the line Fokus must not cross in marketing: coarse state-estimation ≠ clinical attention measurement.
16	A scoping review of consumer-grade EEG for diagnosing neurodevelopmental disorders finds only seven studies. The authors are optimistic about it as a support tool; two of the three are affiliated with a company in the field.	esmed.org — Marchand, Douet Vannucci & O'Connor 2025	2025	Risk / opportunity	T	Seven studies are not a medical evidence base, so D5's “ADHD-adjacent insights” stand on almost nothing. Highest legal-exposure claim in the roadmap.
17	Neurotech devices market growing at ~14.4% CAGR (2025–2034), North America leading.	precedenceresearch.com — Neurotech Devices Market	Aug 2025	Expected development	Ec	Capital will keep arriving for “measurable brains”. Board pressure to scale is structural, not a mood.
18	EEG devices market ~$1.2B (2024) → $1.31B (2025), with the consumer-wearable segment named as a growth driver.	openpr.com — EEG Devices Market Forecast, 27 May 2025	May 2025	Expected development	Ec	Confirms the sub-segment Fokus sits in is the industry's chosen growth story — and that competitors are funded.
19	Meta ships the Ray-Ban Display with a Neural Band (sEMG wristband) at $799 (Sept 30, 2025) — consumer neural interfaces become a mainstream platform play.	cnet.com — hands-on review · about.fb.com — Meta announcement	Sep 2025	Expected development	T	Normalises “neural interface” vocabulary for parents and pupils. Also a platform rival for the same wrists… and foreheads.
20	EndeavorRx holds FDA De Novo marketing authorisation (DEN200026, June 2020), a game-based prescription digital therapeutic for ADHD in children aged 8–12 that created a new Class II category. Its maker Akili has been a Virtual Therapeutics subsidiary since 2024.	accessdata.fda.gov — De Novo decision summary	2020	Emerging issue	EcL	Proof that attention interventions can be regulated medical devices — and a market where medical-grade evidence, not dashboards, is the currency.

Balance note again: Environmental = 0. Reported as a gap, not papered over. #20 is double-tagged Economic + Legislative.

Natural agenda — what the scans cluster into

Theme	Scans	What it means for Fokus	Impact × Likelihood (team's starting judgement)	Newness to Fokus
A · The rules are arriving before the products mature	1–7, 20	Neural data is becoming its own protected class; EU emotional-state inference in classrooms is largely prohibited except medical. Compliance is a product question now, not a legal afterthought.	5 × 5 — high-risk quadrant	Underweighted — treated as “lawyer review before launch”
B · Attention is already a policy problem	10–12	Governments answer the attention crisis by subtracting tech (phone bans). Fokus's growth model assumes institutions will add tech to children instead.	4 × 4 — high-risk quadrant	Named, but not priced into the growth plan
C · The validity gap is published science	13–16	Alpha-attention is real in research-grade rigs; consumer dry-electrode scores are coarse estimates. Any clinical-adjacent claim (D5) sits on the wrong side of the gap.	5 × 4 — high-risk quadrant	Unknown to most of the team
D · This category has a public history	8, 9	2017 and 2019 show the exact narrative that kills school neurotech: “dystopian headbands”, parents in revolt. Legitimacy is a design constraint.	4 × 3	Known to two founders, not to the roadmap
E · Money normalises measurable brains	17–19	14% CAGR capital and a $799 consumer neural wristband make the vocabulary mainstream. Legitimacy risk falls; platform-competition risk rises.	3 × 5	Well understood (it's the thesis)

Drivers of change to take forward

Neural data becomes a distinct legal class (A). Recommendation: design the data model to neural-data standard now — consent age, deletion, derived-data parity.
EU classroom emotion/cognition inference is prohibited ground unless medical (A). Recommendation: make the medical-device question a 2027 board decision, before scaling.
School buyers are trained to prefer subtraction (B, D). Recommendation: reposition from “more measurement” to “less tech per lesson minute” (e.g., sensor on the teacher's desk, not 30 heads).
Published validity gap (C). Recommendation: claims protocol — never diagnose, always report confidence intervals, publish the score's limitations in the pupil UI, not just the procurement PDF.
Backlash playbook exists (D). Research need: parent-panel legitimacy study in two pilot municipalities before any scale-up announcement.
Capital expects measurable-brain growth (E). Recommendation: decide deliberately which growth curve to serve — institutional trust or consumer platform — because they demand different ethics.
Medical pathway is real (20, 16). Research need: regulatory mapping of the EndeavorRx precedent for an EU-MDR equivalent.
Environmental blind spot (gap). Research need: lifecycle estimate for 1:1 wearable EEG in schools (e-waste, charging, replacement rate).

What to watch next

AI-Act enforcement in education (does any edtech get fined for cognitive-state inference?); Dutch/EU guidance translating the UNESCO Recommendation into school procurement; FPF / Council of Europe work on derived neural data of minors; any BrainCo-style device re-entering a European school.

Leads to verify — not signals, not counted

ASSUMPTION Check whether any EU member state has set GDPR Art. 8 guidance on parental-consent age for wearable sensing in schools.
ASSUMPTION Check whether BrainCo-class devices are in any Dutch or Belgian school today.
ASSUMPTION Check the EU MDR classification a device like Fokus Band 2 would attract if it claims attention training.
ASSUMPTION Check the e-waste profile of 1:1 wearables (replacement cycles, battery disposal) in existing school-tablet programmes.

Claim used as input	Type	Basis
UNESCO adopted a global neurotech ethics recommendation in Nov 2025	SOURCED	unesco.org adoption announcement
EU AI Act Art. 5 bans emotion inference in education/workplace from 2 Feb 2025; €35M/7% fines under Art. 99 from 2 Aug 2025	SOURCED	artificialintelligenceact.eu/article/5/ · /article/99/
Colorado, California, Montana and Connecticut classify neural data as sensitive	SOURCED	fpf.org · leg.colorado.gov · kffhealthnews.org
Consumer EEG differs significantly from research-grade EEG spectrally	SOURCED	Mikhaylov et al. 2024
Dry-electrode consumer headbands suit rhythm-based metrics; finer cognitive states need more evaluation	SOURCED	Kleeva et al. 2024
Only seven studies test consumer EEG for diagnosing neurodevelopmental disorders	SOURCED	Marchand 2025
NL phone ban (Jan 2024) improved concentration per government-commissioned study	SOURCED	reuters.com · bbc.com
Screen attention averages ~47s, down from ~150s in 2004	SOURCED	gloriamark.com · apa.org
11% of adolescents show problematic social media use (up from 7%)	SOURCED	who.int/europe (HBSC)
Classroom EEG alpha is the most-used attention marker; evidence preliminary	SOURCED	Zeng et al. 2025
Jinhua school headbands suspended after the 2019 backlash	SOURCED	thechinaproject.com
Neurotech market ≈14.4% CAGR 2025–2034; EEG wearables a growth segment	SOURCED	precedenceresearch.com · openpr.com
Meta Neural Band shipped Sept 30 2025 at $799	SOURCED	meta.com · cnet.com
EndeavorRx holds FDA De Novo authorisation (DEN200026) for ADHD in children 8–12	SOURCED	accessdata.fda.gov
All Fokus figures (pupils, pricing, ARR, schools) and the six design decisions	ASSUMPTION	Fictional case construction, labelled in §01

Handoff

Assumptions surfaced: the brief assumes schools will add sensing tech; it assumes attention is a pupil property, not a lesson property; it assumes derived scores ≠ neural data.

Open decisions: wellness vs medical classification; scale-up vs legitimacy sequencing; data model under neural-data standard.

Suggested next method: fdbx-causal-layered-analysis — the brief is written at the litany layer (a solution and a metric embedded); before scenarios, read it at four depths.

03 · FOUR DEPTHS FDBX · CAUSAL LAYERED ANALYSIS

The brief is a headline.
What's the deep story under it?

Causal Layered Analysis (Inayatullah) reads the issue at four depths (litany, systems, worldview, myth), then climbs back up with an alternative metaphor to produce a reframed brief. Before any scenario work: it makes scenarios differ in kind, not just degree.

fdbx-causal-layered-analysisInayatullah (1998) Futures 30(8) · (2008) Six Pillars pp. 12–14

The brief quoted verbatim in §01 is Fokus leadership's framing — a growth target with a success metric embedded. Context: Dutch and Flemish secondary schools, an industry shadowed by the 2019 Zhejiang headband backlash, and a policy climate that treats attention as something you fix by removing phones.

Depth 1Litany

Problem: “Pupils can't concentrate; teachers can't see who's struggling; parents are blind to it.” Solution: measure attention live, at scale, and act on the numbers. Problem-solver: the product company, with procurement contracts. Where it lives: trade press, pitch decks, this brief. Emotional register: technocratic urgency — someone should build something.

Depth 2Social causes

Problem: attention is shaped by the attention economy's design choices (47-second screen switching; 11% problematic social-media use), by classroom pedagogy, sleep, poverty and lesson design — but the metric the industry can sell is the pupil, not the feed. Solution: systemic fixes — pedagogy, scheduling, curriculum design, platform regulation, phone policy. Problem-solver: ministries, school boards, teacher-training colleges, platform regulators. Where it lives: policy journals, education science, the HBSC/UNESCO literature.

Depth 3Discourse / worldview

Problem: what counts as “attention” depends on the worldview that measures it. Solution: change what classrooms reward, or change how minds are valued in law. Problem-solver: movements, professional communities, standard-setters. Where it lives: academic and peripheral journals, teacher discourse, the neurorights movement.

Depth 4Myth / metaphor

Problem: the story underneath — the child's mind is an engine that idles when unwatched. Solution: a new story to tell. Problem-solver: leaders, artists, teachers themselves. Where it lives: fiction, film, the staffroom, the family dinner table.

Worldviews in play — each constitutes a different problem

Discourse	The problem it constitutes	The future it points toward
Optimisation (Fokus's home discourse; also the discourse of the measurement industry)	Attention is a scarce, quantifiable resource that is being wasted; the task is to make the waste visible so it can be managed. A pupil who scores low is a problem to be corrected.	Continuous measurement, personalised pathways, “attention debt” reported home.
Pedagogical-relational (how Sanne and most teachers actually speak)	Attention is a relational achievement of a well-designed lesson; a drifting pupil tells you about the lesson, the night's sleep, or the hunger — not about the pupil. Measuring the child instead of the lesson reverses the diagnostic arrow.	Lesson design as the unit of improvement; teachers keep interpretive authority; sensors face the room, not the child.
Cognitive liberty (the discourse arriving via Chile, UNESCO and the US states — scans #1–#6)	The inside of a child's head is the last private room. Whoever instruments it owns a piece of the person; derived “attention scores” are a mental-integrity question, not a UX question.	Neurorights in procurement law; children's neurodata treated like medical records; measurement only by consent, only where medical.

Probe (Inayatullah's toolbox): Who is silenced by the brief? The pupil — the only stakeholder whose head is instrumented has no seat in the brief, only a streak counter. Whose knowledge is ordered first? The dashboard's, over the teacher's judgement. What would make the present look strange? A school that measures nothing about attention and still produces focused children.

Root metaphor and alternative

Root metaphor: “The tuned engine” — the child's attention as an engine that idles when unwatched, running best under a gauge on the dashboard.
Alternative metaphor: “The tended garden” — attention as something a classroom grows: soil (sleep, safety, hunger), weather (the feed), gardener (the teacher). You don't gauge a garden; you tend it.

Moving back up — the alternative metaphor rebuilt

Layer	Rebuilt from “the tended garden”
Worldview	Pedagogical-relational with cognitive-liberty guardrails: the lesson is the patient; the child's interior stays sovereign unless they consent, medically, with their parents.
Social causes	Attention improves by changing lesson design, schedules, sleep and the feed — not by instrumenting children. Fokus's assets (signal processing, school relationships) re-point at room-level sensing and lesson analytics.
Litany (reframed brief)	“Design the road to 2040: grow Fokus from 12 pilot schools to 2,500 schools across Northwest Europe — by making every lesson measurable and every child's interior private. By 2040 teachers see rooms, not pupils; parents get support, not surveillance reports; and attention data powers lesson design, never learning paths of named children.” Success = room-level insight coverage ≥90% of lesson minutes · zero named-pupil raw-signal retention · ARR €120M by 2036 (lower ceiling, honest)

Same company. Different patient. The litany brief treats the child; the reframed brief treats the room.

Scenarios by layer — four kinds, not four degrees

Litany scenario (instrumental): 2,500 schools by 2033. Coverage hits 92%; a study finds scores don't predict learning gains; churn arrives as procurement renewals lapse. The obvious fix — more dashboards — ran out of road.
Social-cause scenario (systems): The Dutch ministry extends the phone-ban logic: a national “focus hour”, later school start times, feed-design rules under the AI Act's education clause. Fokus pivots to selling compliance analytics on lesson structure instead of children.
Worldview scenario (different way of seeing): Neurorights language enters EU procurement: no biometric sensing of named minors outside medical care. Cognitive liberty wins as the frame; Fokus survives only as an anonymised, room-level instrument.
Myth scenario (story): “Milan, 14, remembers the headband the way we remember school desk bells — as a thing that watched him. He tells his daughter the story of the year his class went on strike for their heads, and the teacher, Sanne, led it. ‘It wasn't the wire we minded,' he says. ‘It was being a gauge.'”

Design implications

Holds: schools have a real, measured attention problem; Fokus's sensing IP and school relationships are real assets; 2040 is the right horizon.
Question with the client: the unit of measurement (child → room); the customer of the report (teacher → parent); whether “personalised learning paths from attention data” survives contact with the pedagogical-relational discourse at all.
Research next: with three pilot schools, test whether room-level sensing changes lesson design behaviour (the pedagogical claim) before scaling pupil-level instrumentation.

Claim used as input	Type	Basis
The brief's framing, success metrics and six design decisions	ASSUMPTION	Fictional case construction (§01)
Attention shaped by screen design; 47s screen-switching; 25-min refocus cost	SOURCED	gloriamark.com · apa.org
11% adolescent problematic social-media use, rising since 2018	SOURCED	WHO Europe / HBSC, Sep 2024
NL policy treats attention via subtraction (phone ban; positive concentration effects)	SOURCED	reuters.com · bbc.com
Neurorights discourse is entering law (Chile constitutional; UNESCO recommendation; US states)	SOURCED	courier.unesco.org · unesco.org · leg.colorado.gov

Handoff

Assumptions surfaced: “attention is a pupil property” is the brief's deepest move; the parent report assumes parents want surveillance, not support; the metric assumes measurement causes learning.

Open decisions: unit of measurement (child vs room); whose worldview Fokus sells to.

Suggested next method: fdbx-futures-triangle — map the forces pulling, pushing and weighing on both briefs before writing any 2040.

04 · THE TRIANGLE FDBX · FUTURES TRIANGLE

What pulls, what pushes,
and what weighs

Inayatullah's Futures Triangle maps the images of the future that pull, the present drivers that push, and the barriers that weigh on each image, then asks the design question: which future is the current strategy already serving, and was it ever chosen?

fdbx-futures-triangleInayatullah (2008) Six Pillars pp. 5–8, 15 · horizon 2040 · views of: board, teachers, parents, pupils, regulators, capital

Pull — images of the future people actually hold

Image	Archetype	Held by	What it promises
“Every child measured, every lesson optimised.” The classroom as a quantified system whose efficiency keeps rising.	Evolution & progress	Fokus board, growth-stage investors, procurement-minded school networks	No pupil slips through; teaching becomes evidence-based like medicine.
“The sovereign head.” Children's mental privacy wins; no one instruments a minor's mind outside medicine.	New image — name it “the last room” (cognitive-liberty)	Neurorights movement, privacy regulators, UNSECO-aligned educators, many parents	A childhood that isn't a dataset; attention protected like bodily integrity.
“Back to chalk.” Classrooms unplug; attention is rebuilt the old way — books, rote, silence, phones out of the room entirely.	Back to the future	Phone-ban constituencies, traditionalist educators, a visible parent current	Calm rooms; measured improvement (the NL study is its evidence base).
“The room that learns.” Sensing faces the lesson, not the child: room-level acoustics, pacing, engagement feedback to the teacher within seconds.	Gaia (repair & partnership)	Dr. Bakkali's learning-science team, teacher professional bodies, Sanne	Better lessons without a dataset of children; the teacher keeps judgement.
“Measured minds, ranked.” Attention scores become currency: class rankings, school league tables, insurer discounts for focused kids.	Collapse-adjacent — name it “the sorting room”	Nobody openly — an unchosen future pushed by D2/D3/D6 and platform logic	Efficiency for the system; quiet damage for the sorted.

Push — drivers already moving (all sourced or typed)

Driver / trend	Direction & size	Pushes toward	Ledger
Neurotech market growth ≈14.4% CAGR 2025–2034	Positive, sustained, multi-billion	“Every child measured”; “the sorting room”	#17
Consumer EEG validity published as coarse (not clinical)	Constraint on claims, not adoption	Pushes “the room that learns” (coarse metrics fit rooms); pushes against child-level claims	#14, #15
Legislative tide: UNESCO recommendation, US neural-data laws, EU AI Act education clause	Accelerating 2024–2026	“The sovereign head”	#1–7
Screen-attention erosion (47s) + 11% problematic use	Worsening decade over decade	All interventionist images — including “back to chalk”	#11, #12
Government phone bans with measured positive effects	Spreading (60+ education systems by end-2023 per session search)	“Back to chalk”; institutional caution toward sensing	#10
Platform normalisation of neural interfaces (Meta Neural Band at $799)	New, mainstream vocabulary	“Every child measured” — legitimacy transfers	#19

Weight — barriers, specific to each image

Image	Weights holding it back
Every child measured	EU AI Act Art. 5 education clause (medical exception forces device classification); GDPR Art. 8 child-consent patchwork; procurement veto risk; the 2019 Zhejiang headband backlash; teacher-union opposition; published validity gap.
The sovereign head	Weak enforcement instruments today (recommendations, not treaties); capital chasing the 14% CAGR; anxious-parent demand for “something”; Fokus's own sunk Series B narrative.
Back to chalk	Digital-skills curriculum demands; homework platforms already embedded; parents' own screen habits; measurement lobby's evidence marketing; a decade of 1:1-device procurement inertia.
The room that learns	No procurement category for room-level analytics yet; sensor-per-room cost vs per-pupil pricing; Fokus's per-pupil revenue model (€18/head) has no room SKU; teacher-workload scepticism about “another dashboard”.
The sorting room	Streaks and leaderboards already normalized by consumer apps; insurer pilot money (D6) asking for individual-level outcomes; no current legal barrier to derived scores in most jurisdictions (scan #7's Goldilocks gap).

Interaction

Plausible future (not a prediction): a contested drift between “every child measured” and “the room that learns”. The pushes behind measurement are strong and funded, but child-level instrumentation carries the heaviest weights (law + backlash history). Room-level sensing keeps the push (money, data, pedagogy) while shedding most of the weight — which is why it is the image the strategy should choose deliberately rather than arrive at by accident.

Contested images: “Every child measured” (strong pull, heaviest weights) vs “The sovereign head” (rising push from law, cultural momentum, but thin enforcement).

Unchosen pushes: derived-score sorting (D2 + D3 + D6) is drifting toward “the sorting room” — no stakeholder is pulling for it, and nothing currently weighs on it. That asymmetry is the strategy's sharpest risk.

Scenarios from the triangle

“Coverage” — built on “every child measured”: growth wins procurement; validity gap surfaces publicly in 2031; the sorting room arrives as the business model's quiet rescue.
“Procurement wall” — built on driver “legislative tide”: neurorights language enters Dutch/EU school procurement; child sensing dies in legal review; room-level Fokus is the pivot that survives.
“Chalk dividend” — built on “back to chalk”: bans extend from phones to wearables; Fokus's sensing IP is licensed to clinics (the EndeavorRx-style path, #20) instead of classrooms.
“The gardener's console” — built on “the room that learns”: Fokus re-prices per room, becomes lesson-design infrastructure, and parents receive support offers instead of surveillance reports.

Design implications

Which image does the current strategy serve? “Every child measured” — with the sorting room as its uncosted shadow. Chosen or used? Largely a used future: borrowed from the growth playbook of the measurement industry (14.4% CAGR, per-seat SaaS pricing) without a deliberate decision that this is the future Fokus wants for children.

The disowned future: “The sorting room.” It is the image leadership pushes away (“we would never rank children”) while every current decision — streaks, leaderboards, parent percentiles, insurer pilot — builds it. The triangle says: look here.

Claim used as input	Type	Basis
14.4% CAGR neurotech market	SOURCED	precedenceresearch.com (scan #17)
Consumer EEG coarse vs clinical validity gap	SOURCED	Mikhaylov 2024 · Kleeva 2024 (#14–15)
Legislative tide toward neural-data protection	SOURCED	unesco.org · EU AI Act Art. 5 · HB24-1058
Phone-ban momentum & positive concentration findings	SOURCED	reuters.com · bbc.com (#10)
Meta Neural Band normalises consumer neural interfaces	SOURCED	meta.com (#19)
Insurer-pilot demand, per-pupil pricing, streak/leaderboard mechanics	ASSUMPTION	Case construction D2, D3, D6 (§01)

Handoff

Assumptions surfaced: the strategy is serving a borrowed image; the disowned future is the one being built.

Open decisions: room-level vs pupil-level sensing; what to do about the unchosen push toward sorting.

Suggested next method: fdbx-three-horizons — the weights are H1 lock-in; “the room that learns” is a candidate H3; pockets exist now.

05 · THREE HORIZONS FDBX · THREE HORIZONS PRACTICE

What's losing fit, what's emerging,
and which way each experiment points

Sharpe & Hodgson's practice maps three patterns that exist now: first-horizon business-as-usual losing its fit, third-horizon aspirations already alive as pockets in the present, and the turbulent second horizon where innovations either open the way (H2+) or prop up the old system (H2−). This is Sharpe's futures practice, distinct from the growth-portfolio model of the same name.

fdbx-three-horizonsSharpe, Hodgson, Leicester, Lyon & Fazey (2016) Ecology & Society 21(2):47 · Curry & Hodgson (2008) · horizon length: 14 years (2026→2040) — institutional education + regulatory cycles move slower than consumer software

Fig. 1 · The three horizons of attention technology in schools. All three patterns are present in 2026; only their prevalence shifts. ● the H2 fork: heatmap − · room pilot + · ● H1 losing fit: law, validity, backlash.

The map

Horizon	Pattern	Evidence	Who holds it	Direction
H1	Attention managed by adding measurement to children: per-pupil EEG dashboards, parent reports, engagement KPIs	Loss of fit: EU AI Act education clause; consumer-EEG validity gap; 2017/2019 backlash history; phone-ban policy logic runs the other way (#1–10, 14–16)	Fokus today; engagement-analytics vendors; procurement-led school networks	Declining
H2	Transition experiments: room-level analytics, consent-first “focus hours”, streak gamification, research-data sharing	Innovations in play at Fokus and competitors A	Product teams, learning-science labs, insurers	Turbulent
H3	Either (a) “the room that learns” — sensing faces lessons, children stay uninstrumented; or (b) “the sovereign head” — schools exit child instrumentation entirely	Pockets below (#1, #6, #10, #13, #20)	Learning scientists, neurorights policymakers, teacher bodies	Emerging (contested: two H3s)

H1 — losing fit (and the fixes that fail)

The per-pupil dashboard model is losing fit three ways at once: legally (emotional/cognitive-state inference in education is prohibited ground in the EU unless medical — scan #6; neural data hardening into a protected class — #4–5), scientifically (published gap between consumer headbands and research-grade EEG — #14–15; no diagnostic validity for consumer EEG — #16), and culturally (the 2019 Zhejiang headband story is the category's public memory — #9). Fixes that fail: deeper opt-out menus, stronger anonymisation claims, and “teacher can hide a pupil's column” — each patches the dashboard while the law, the science and the story move against the dashboard itself.

H3 — aspirations and pockets of the future in the present

H3 vision	Concrete pocket already existing (ledger-linked)
H3-a · The room that learns. Room-level sensing and lesson analytics; the teacher, not the child, is the instrumented unit; attention improvements are designed into lessons, schedules and sleep.	Classroom EEG research showing attention-linked signals in real lessons (research-grade rigs) — Zeng et al. 2025 · NL's phone-ban evidence that subtracting tech measurably helps — reuters.com
H3-b · The sovereign head. No biometric sensing of named minors in classrooms outside medical care; Fokus's sensing IP exits to clinics.	UNESCO neurotech ethics recommendation adopted Nov 2025 — unesco.org · EU AI Act Art. 5 education clause in force — artificialintelligenceact.eu · medical-digital-therapeutic pathway proven — FDA De Novo DEN200026 (EndeavorRx)

The two H3s conflict: H3-a keeps Fokus in classrooms with consent-first sensing; H3-b ends classroom sensing altogether. Both are legitimate; the roadmap pretends only one future exists.

H2 — innovations in play, and which way each points

Innovation	H2+/H2−	Why
Teacher live heatmap of named pupils (D1)	H2−	Refines H1 on its own terms; deepens child-level instrumentation that law and science are moving against.
Focus streaks & class leaderboard (D2)	H2−	Imports consumer engagement mechanics; builds the “sorting room” nobody is pulling for.
Parent weekly Focus Report with percentiles (D3)	H2−	Surveillance dressed as care; entrenches the child-as-metric frame with families.
Anonymised research & insurer data sharing (D6)	H2−	Data-exhaust economics props up the H1 business model; the Goldilocks derived-data gap it relies on (#7) is closing.
Room-level acoustics & pacing pilot (Bakkali's team, unsourced in roadmap)	H2+	Opens H3-a: shifts the instrumented unit from child to lesson; makes the validity gap (#14–15) manageable instead of fatal.
Consent-first “focus hour” — lesson-structure analytics with no pupil signal at all	H2+	Creates the missing procurement category for H3-a; survives every legislative scenario in §02.
Regulatory-grade clinical spinout (sensing IP to clinics, EndeavorRx-style)	H2+	Bridges toward H3-b if child sensing must die: medical consent, medical-grade evidence, real indication (#20).

Features to maintain from H1

School relationships and procurement craft — twelve pilot schools that let Fokus iterate is the scarcest asset in edtech; carry it into any H3.
On-device signal processing — privacy-preserving compute is H1 engineering that H3-a inherits directly.
The teacher trust Sanne represents — if the pivot to room-level is made with pilots, H1's trust funds H3's legitimacy.

Dilemma and power

Dilemma: “honour the growth promise made at Series B” (H1 value: 2,500 schools, €220M ARR) and “keep the licence to operate in children's spaces” (H3 value: legitimacy under arriving law and science). Both must be respected; the dilemma is re-solved in experiments, not once at a board meeting.

Power: the board and lead investors can enable or veto the pivot (and their return expectations are the heaviest weight). The DPO can legally block D1–D6 features in the EU. Teacher unions and ministries can kill the category. Pupils hold latent power through parents and press — the 2019 Zhejiang backlash shows what happens when they exercise it.

Horizon perspectives

Looking from	At	Fixed mindset	Positive perspective
H1 manager (CPO Jonas)	H3-a / H3-b	“Consent-first analytics is unprocureable philanthropy; clinics are a different company.”	“Room-level and clinical are the only two roads that survive the legislative tide — H1 trust funds the crossing.”
H3 visionary (Dr. Bakkali)	H1	“A surveillance dinosaur collecting its own lawsuit.”	“Twelve schools of earned trust and real signal-processing IP — the launchpad H3 has no other way to buy.”
H2 experimenter (insurer pilot lead)	Both	“Purists on one side, cowboys on the other.”	“My pilot data decides the fork: if room-level beats child-level on outcomes, the argument ends.”

Actions

H2+ moves to start: fund the room-level pilot in all 12 schools this school year; write the consent-first focus-hour spec; scope the clinical spinout with an MDR consultant.
H1 features to protect: on-device processing; school relationships; teacher trust. To wind down: child-level public heatmaps and percentile parent reports (they are the lawsuit, not the product).
H3 pockets to connect with: UNESCO-recommendation-aligned procurement pilots; the HBSC/UNESCO school-health networks; a clinical partnership where sensing IP can live under medical consent.

Portfolio classification — Fokus's current initiatives

Initiative	Class	Reason
Live pupil heatmap (D1)	H1 maintenance	The flagship of the losing-fit pattern.
Streaks & leaderboard (D2)	H2−	Engagement mechanics deepen child-as-metric frame.
Parent Focus Report (D3)	H2−	Extends dashboard logic into the home.
24-month retention + model reuse (D4)	H1 maintenance	Data-exhaust economics; soon legally untenable in the EU.
Mentor “ADHD-adjacent insights” (D5)	H2− (drifting to H1-lock)	Claims pull Fokus toward medical classification without medical evidence (#16).
University/insurer data sharing (D6)	H2−	Funds H1; the derived-data gap it leans on is closing (#7).
Room-level pilot (unfunded)	H2+	The fork that opens H3-a; currently the portfolio's smallest line item.

Balance read: six of seven initiatives are H1 maintenance or H2−. Most of Fokus's resources are currently spent propping up a system the law, the science and the public story are all moving against.

Claim used as input	Type	Basis
EU AI Act education clause in force; neural-data class hardening; UNESCO recommendation	SOURCED	Art. 5 · HB24-1058 · unesco.org
Consumer-EEG validity gap; no diagnostic validity	SOURCED	Mikhaylov 2024 · Marchand 2025
Phone-ban subtraction works; backlash history of category	SOURCED	reuters.com · edsurge.com · thechinaproject.com
Initiative list, staffing, funding, pilot sizes	ASSUMPTION	Case construction (§01)

Handoff

Assumptions surfaced: the portfolio is ~85% invested in the pattern losing fit; two legitimate H3s are being run as zero.

Open decisions: fund H2+ room pilot or not; choose which H3 the company is building toward (or deliberately for neither).

Suggested next method: fdbx-four-futures — the H3 fork is still inside one worldview; experience four fundamentally different 2040s before choosing a preferred future to backcast from.

06 · FOUR TUESDAYS IN 2040 FDBX · FOUR FUTURES

Experience the alternatives
before choosing one

Dator's four generic futures (Continued Growth, Collapse, Discipline, Transformation), each taken seriously, each written from inside and positively, each with an artifact and the product living in it. No future is labelled best, worst or most likely; the four differ in what they believe about how the world works.

fdbx-four-futuresDator (2009) J. Futures Studies 14(2) · horizon 2040 · mode: analysis

Future	Generic type	Underlying logic (what people believe about how the world works)	What changed
The Measured Decade	Continued Growth	Markets and measurement keep solving problems; anything countable improves; childhood is no exception.	Sensors normalised, procurement industrialised, attention a standard school KPI.
The Rip-out	Collapse	Systems built on extracted trust fail suddenly and messily; legitimacy, once lost, is the only non-renewable resource.	A breach-and-scandal cascade ends institutional neurotech in schools; measurement retreats to clinics.
The Neuro-Rights Classroom	Discipline	Some goods are not for sale or measurement; mental integrity is a constitutional value that outranks efficiency.	Neurorights in procurement law; child instrumentation restricted to medicine; attention rebuilt by design of rooms and days.
The Symbiotic School	Transformation	Technology remakes the learner, not just the tools: learning, memory and attention become negotiable, adaptive, personal.	Neuroadaptive learning is everyday and learner-owned; the measurement/teaching distinction dissolved.

FUTURE 1

The Measured Decade

Continued Growth

2040 is a good year to be a school with data. The 2027 “Cognitive Support Framework” turned attention tools into ordinary infrastructure: calibrated, audited, cheap. Teachers start the day with a room view, parents with a digest, pupils with a plan. The old anxieties about “headbands” feel as dated as fears about calculators. Schools in funded regions post their best attainment numbers in a generation, and boards credit the dashboards.

Artifact from this future

Excerpt, Regionaal College Utrecht-Oost annual report, 2040

“Attention-support coverage reached 94% of lesson minutes this year, within the national framework's fairness envelope. Graduate destinations rose for the fourth year; the college attributes this to early-warning analytics and personalised pathways.”

A · Life in this future: most people work with, not against, algorithmic management; school inequality narrowed in rich regions and widened between them; today's worries about screens feel minor — new worries are about attention-boredom and theatre of compliance. B · Probability: the growth signal chain (#17, #18, #19) makes this the default trajectory — if the validity gap (#14–16) stays unexamined and no scandal lands (#9's pattern). C · Preferability: close to Fokus's stated dream; far from Sanne's. D · Toward its desirable aspects: 1) fund the room-level pilot now; 2) build procurement relationships beyond the 12 pilots; 3) publish the score's confidence intervals; 4) make consent architecture best-in-class; 5) train teachers in data literacy. E · Against its undesirable aspects: 1) kill percentile parent reports before they become national; 2) refuse individual-ranking features permanently; 3) cap raw-signal retention at session length; 4) separate “insight for the teacher” from “score for the record”; 5) keep an offline, unscored school mode.

Fokus in this futureThrives as infrastructure — if it rebuilt trust primitives early. It stops selling surveillance and sells calibration, fairness audits and pathway design.

FUTURE 2

The Rip-out

Collapse

In 2033 the “Focus Files” leak — years of named-pupil attention traces from several vendors, including from the “anonymised” research channel. Parents across Europe discover their children were the product. Schools rip the headbands out in a term; some teachers quietly grieve the early-warning system that did catch the quiet kids. Trust in edtech collapses for a decade. People live well here anyway: the dust settles into a smaller, choosier edtech sector, and children's school days are stubbornly analogue — which, parents tell pollsters, suits them fine.

Artifact from this future

NRC headline, 14 March 2033

“Province cancels all neurotech contracts: ‘We measured everything about our children except whether we should.'”

A · Life: local-first schooling, analogue-by-default classrooms, a booming “unmeasured education” premium market; digital-waste remediation employs thousands; problems gone: attention-KPI theatre, dashboard fatigue; new problems: no early-warning for the pupils who needed it, and a black market in bootleg focus tools. B · Probability: D4 + D6 make the leak path real; the category already has its 2019 dress rehearsal (#8, #9). C · Preferability: hard on the industry, kind to most children; Fokus as currently designed does not survive it. D · Toward its desirable aspects: 1) adopt data-minimisation now, before law forces it; 2) build the brand on “we delete”; 3) publish a wipe-on-demand API; 4) fund analogue pedagogy partnerships; 5) make local processing a marketing headline. E · Against its undesirable aspects: 1) no third-party research sharing of child-derived signals ever (D6 dies); 2) insurers out of school data entirely; 3) breach insurance + 72-hour family notification; 4) pseudonymise pupils from day one; 5) board-level incident drills run like fire drills.

Fokus in this futureWho does well: teachers (regained authority), pupils (unmeasured), open-source privacy tooling, and whichever vendor can prove minimal data. Fokus survives only as the company that couldn't leak what it never kept.

FUTURE 3

The Neuro-Rights Classroom

Discipline

The UNESCO recommendation of 2025 turned out to be a seed, not a brochure. Chile's Article 19 became a template; the EU's 2031 “Mental Integrity Act” wrote neurorights into procurement: no biometric sensing of named minors outside medical care. Growth didn't stop — it was re-ordered around values. Schools redirected the measurement budget into smaller classes, later starts, phone-free days and food. Neurotech didn't die; it grew up and moved to hospitals, where consent is informed and evidence is clinical.

Artifact from this future

EU model procurement clause, 2035

“§4.2 The supplier shall not create, infer, retain or transfer physiological or neural measures identifiable to a pupil, except under medical authorisation per §4.3. Room-level analytics that cannot be re-attributed to any pupil are permitted.”

A · Life: people treat mental privacy like water quality — a public good with regulators; teachers hold interpretive authority; problems gone: percentile childhood; new problems: how to evidence learning without extraction, and richer schools buying “private clinical” workarounds. B · Probability: every legislative signal (#1–7) points this way; the open question is speed, not direction. C · Preferability: highest for pupils and teachers; uncomfortable for current Fokus. D · Toward its desirable aspects: 1) volunteer for the strictest procurement regime now; 2) build the room-level, non-re-identifiable product line as a first-class citizen; 3) help draft the model clause so it's implementable; 4) move sensing IP toward clinical indication (#20); 5) certify “cannot re-attribute” with external auditors. E · Against its undesirable aspects: 1) fight the private-clinical workaround loophole; 2) keep tools affordable for poor schools under the new regime; 3) support teacher workload as measurement falls away; 4) guard against a shadow market of parent-purchased bands; 5) document the learning evidence room-level analytics can provide.

Fokus in this futureThrives as the clause-compliant incumbent: room analytics plus a clinical arm. It stops being a child-surveillance company and becomes what Bakkali always said it was — a lesson-science company.

FUTURE 4

The Symbiotic School

Transformation

In 2040 the question “is the child paying attention?” sounds odd — like asking if the river is wet. Neuroadaptive tools became personal property: a learner's own signal pair-bonds with their own learning environment, sharing nothing. School is where you bring your instrument, like a violin. Attention isn't measured by institutions; it's modulated by learners, with teachers orchestrating ensembles. The Meta Neural Band of 2025 (#19) was this future's pediment — barely noticed under the scandal headlines of the late 2020s.

Artifact from this future

Pupil-facing card, “mySynapse”, school year 2040–41

“Today your history class runs at 60% discussion, 40% focus-blocks — your profile, your choice. Your signal never leaves your band. Everything your teachers see is your own report, your own words.”

A · Life: learning is lifelong and adaptive; cognitive augmentation is normalised and stratified by access; problems gone: the attention crisis as public panic; new problems: cognitive inequality between instrumented and uninstrumented learners, “signal identity” theft, what school is when everyone learns at their own edge. B · Probability: technology-pushed and normalising (#19), but gated by the trust collapse (#2's path) and the rights regime (#3's path) — a transformation that must pass through one of the other futures first. C · Preferability: intoxicating for learners; requires an autonomy architecture today's Fokus has never drafted. D · Toward its desirable aspects: 1) prototype learner-owned signal wallets (data stays on-device, learner-controlled); 2) open the scoring model to pupil inspection; 3) design for asymmetry-reversal: pupil sees everything the school sees; 4) partner with teacher colleges on the new role; 5) publish the interoperability spec for bands. E · Against its undesirable aspects: 1) refuse cohort-level optimisation that uses individual signals; 2) build the uninstrumented learner a first-class experience; 3) hard-limit persuasive nudges toward “focus”; 4) no adaptive content that infers mental-health state; 5) keep a human teacher above every algorithmic judgement.

Fokus in this futureIt becomes the neutrality layer — the company that makes learner-owned signals interoperable — or it is commoditised by whoever does. Its streak mechanics (D2) are the exact opposite of what this world rewards.

Across the four futures

Reading	Finding
Robust design decisions (hold in all four)	1) On-device processing with deletion as default; 2) no individual-ranking mechanics, ever; 3) coarse, honestly-bounded claims (never clinical without clinical evidence); 4) teacher-facing value that survives even when pupil signals are refused; 5) consent architecture designed for the strictest plausible regime (#3's clause). At least three, as the method requires — here: five.
Single-future bets in the current plan	Streaks & leaderboard (D2) and parent percentile reports (D3) pay off only in the Measured Decade; the 2,500-school / €220M plan is a Continued-Growth artifact through and through.
Signals to watch	Growth: procurement frameworks normalising attention KPIs. Collapse: any named-pupil neurodata leak anywhere in edtech. Discipline: first national procurement clause banning child instrumentation; AI-Act enforcement in education (#6). Transformation: consumer neural-wearables adoption passing ~10% of households (#19's line).

Toward a preferred future (a sketch, not a conclusion)

Take the Neuro-Rights Classroom's legal spine, the Symbiotic School's learner-owned signals as its 2040 destination, and the Measured Decade's funding engine — minus its percentile childhood. That is §04's “gardener's console”: rooms measured, minds sovereign, teachers richer not poorer. §10 backcasts from exactly this.

Claim used as input	Type	Basis
Legislative direction toward mental-integrity regimes	SOURCED	unesco.org · courier.unesco.org · Art. 5
Market & platform normalisation trajectory	SOURCED	precedenceresearch.com · meta.com
Backlash precedent of category	SOURCED	edsurge.com · thechinaproject.com
2033 leak, 2031 Act, 2040 products, all named artifacts	ASSUMPTION	Fictional future-construction (the method's job), grounded on the sourced signals

Handoff

Assumptions surfaced: the current roadmap contains exactly one future (Measured Decade) and is fully exposed to the Rip-out.

Open decisions: which two robust decisions to adopt this quarter; whether D2/D3 survive the stress test at all (→ §07–08).

Suggested next method: hand the four futures to the ethics wall — edbx-worrystorming and edbx-black-mirror-brainstorming on what the product could do inside these worlds; robust decisions then go to edbx-pledge-works at §10.

07 · THE ETHICS WALL — BAY 1 EDBX · WHOSE BRAIN IS IT?

Before we ask what it does,
ask who it reads

The edbx router's first question for a product involving minors and sensitive data is always the same: whose view is missing, who is excluded, who is exposed? Four methods answer. Then Bay 2 asks what could go wrong.

BAY 1

Whose brain is it?

Exclusion (CIDER) · exploitation of sensitivities (Humane Design Guide) · the designer's own lenses (Another Lens) · how to hear the missing voices (Critical Interviewing). All four run on Fokus Band 2 in its 12 pilot schools.

edbx-ciderOleson, Solomon, Horowitz et al. (2022) — assumption elicitation · #newperspectives #identifyvalues

Stage C — Assumptions surfaced

“This design assumes the user can and will wear a rigid band for 50-minute lessons, five times a day.” — Ability
“This design assumes the user's hair, headwear and head shape leave forehead + mastoids accessible for dry electrodes.” — Ability / Environment
“This design assumes stillness and eye-forward gaze equal attention — that a ‘neutral' attentive body exists.” — Ability / Capacity
“This design assumes pupils read the pupil app at grade level and understand streaks, opt-outs and scores.” — Capacity
“This design assumes parents have a smartphone, Dutch/English literacy and quiet evenings to read a weekly report.” — Resources
“This design assumes classrooms have charging carts, dense WiFi and storage for 30 bands per room.” — Environment
“This design assumes a pupil can opt out without social cost — that being the only red dot is socially free.” — Environment

Stage I — Exclusion scenario

Assumption in focus: the “neutral attentive body.” Amina, 13, second-form vmbo at De Brug, wears her hijab pinned close; the band's electrodes will not seat over fabric, so hers lives clipped to the desk edge, scoring the room instead of her. On Thursday she is actually locked in — she always thinks best while squeezing a stress ball low in her lap — and the screen still paints her dot amber, then red, through the lesson she loved. On Friday the streak she was proud of resets. The Mentor suggests she “try to stay stiller.” Sanne knows Amina was fine; the dashboard doesn't, and the dashboard is what the mentor meeting reads from. Amina starts calling herself “a red kid” in the third person.

Stage D — Redesign proposals

Interaction A pupil-controlled visibility layer: before any teacher view, each pupil self-anchors the score with a two-tap “locked in / drifting” check-in; disagreement between self-report and band opens a calibration note, not a public colour.
High-tech An inclusive fit kit — three band sizes, cap-compatible electrode arrays, and a “hands-free fidget” mode that re-baselines motion artefacts per pupil instead of against a stillness norm.
Systemic Instrument the room, not the child: ceiling acoustic + pacing sensing gives Sanne the same lesson insight with no pupil wearables at all (the §05 H2+ pilot).
Low-tech A paper opt-out card the pupil hands to any teacher, any time, no app, no menu, no questions — honoured class-wide in under a minute.

Stage E & R — shared register and the second cycle team mode

#	Assumption	Category	Surprising?	Source
1	Attentive bodies sit still	Ability	Yes — most contested in room	Sanne
2	Electrodes contact skin of every head	Ability/Environment	Yes	Donders
3	Grade-level reading of the app	Capacity	No	De Wit
4	Parents read reports	Resources	Yes — 9% of pilot parents had no smartphone or data plan A	Bakkali
5	Opt-out is socially free	Environment	Yes	Verhoeven
6	Classrooms have charging infrastructure	Environment	No	De Wit
7	Every pupil can self-advocate to a teacher	Capacity	Yes	Sanne

Stage R — second cycle, assumption: “opt-out is socially free.” Milan, 14, is the only red dot in a leaderboard-winning class. The card system exists now — but opting out means his class loses three streak points in the weekly tally, and his classmates know it. He keeps the band on, rocks slightly, goes red, keeps the streak. The exclusion moved from the interface into the peer group. Redesigns: 1) opt-out is class-anonymous (aggregate shrinks, never a named gap); 2) streaks never depend on pupils who are opted out; 3) an “off week” class ritual — the whole room goes dark together so being unmeasured is normal, not deviant.

Inclusive design commitment: “We commit to shipping the paper opt-out card and the inclusive fit kit to all 12 pilot schools before the January 2027 term, owned by the CPO, because we learned that ‘the neutral attentive body' excludes Amina — and that exclusion re-appears at group level unless opt-out is made socially free.”

edbx-humane-design-guideCenter for Humane Technology — six sensitivities · #breakmydesign #evaluateoutcomes

Value proposition: “See attention. Support every learner.” Measures of success (as briefed): attention-score coverage of lesson minutes ≥90% · ARR €220M · churn <2% · flagged: coverage is an extraction metric — it rewards instrumented minutes, not learning.

Sensitivity	Current state	What's at risk	Level	Named mechanism	Most affected (≥2, with mechanism of harm)	Enabled to…
Attention	Live dashboard pulls teacher attention to the screen; pupil streaks create a second, private attention economy	Pupils self-monitor being monitored; genuine absorption displaced by score-management	🔴 High	Streak-loss aversion + attentional residue: every glance at a red dot costs a return-to-task; ~25-min refocus cost is documented	Adolescents (developmental impulse-control gap); ADHD pupils (scored on a neurotypical stillness baseline they cannot meet)	Find focus and peace
Emotional	Public red/amber/green per named pupil in front of peers	Chronic low-grade shame during the lesson; learned fear of one's own states	🔴 High	Status threat + public-shame mechanics; score becomes identity (“I'm a red kid”)	Adolescents (peak peer-comparison sensitivity); pupils with anxiety disorders (interoceptive vigilance amplified by the band)	Process and adapt
Sensemaking	Score implies “attention → learning” with no uncertainty display; percentiles imply normality	Teachers and parents misread a coarse EEG proxy as cause of learning	🟡 Medium	Metric reification — a derived estimate presented as a measured fact (validity gap #14–16)	Parents reading reports without training (median pupil-report session 90 seconds A); substitute teachers reading dashboards cold	Find truth and clarity
Decision-making	Opt-out exists but is three menus deep; no pupil-facing explanation of what is stored	Agency removed from the people wired	🔴 High	Friction asymmetry + default bias (defaults set for procurement, not the child)	Pupils without adult advocates; low-literacy families (consent forms at B1 reading level or below)	Act with agency and control
Social Reasoning	Class leaderboard; streak visible to classmates	Relationships reframed as ranking; helping a struggling classmate costs the helper points	🔴 High	Elo-style peer ranking carrying historical participation bias; peer-comparison sensitivity	Adolescents; pupils from homes with less homework support (rank correlates with household resources)	Connect meaningfully
Group Dynamics	Class-versus-class league table in staffroom and newsletter	“Green classes” vs “red classes” becomes track and class stigma (vmbo over-represented)	🟡 Medium	Intergroup ranking — aggregate scores convert into group identity	vmbo (pre-vocational) classes vs vwo (pre-university) — Dutch tracking correlates with migration background; part-time and special-ed groups	Build strong communities

Design alternatives for the three 🔴 sensitivities — minimal / structural / radical, each with its metric change opens the full menu

Attention

A (minimal): replace live dots with an end-of-lesson summary the teacher reads after the bell. Metric: coverage ↓ ~10%; teacher-reported interruption incidents ↓. Heuristic: “Nothing scores a pupil while the teacher is teaching.”
B (structural): pupil-owned visibility — teacher sees aggregates, pupils see their own score first, always. Metric: pupil trust index (to be invented) replaces coverage as the north star. Heuristic: “If a pupil can't see it before the teacher does, it's surveillance, not support.”
C (radical): stop instrumenting children; ship the room-level pilot. Metric: lesson-design changes per class per term. Heuristic: “Measure the weather, not the flowers.”

Emotional

A: no colour on any shared surface; words, ranges, and a “not sure” state. Metric: stigma language in pupil interviews ↓.
B: scores never shown to anyone but the pupil unless the pupil chooses to share. Metric: sharing rate becomes the engagement signal. Heuristic: “Consent is a share button, not a settings menu.”
C: retire the emotional register of the product — rename scores after lessons (“this lesson ran hot”), never after children. Heuristic: “Lessons can be bad. Children can't be red.”

Decision-making

A: one-tap opt-out on the pupil home screen; paper card equivalent. Metric: opt-out completion ≤2 taps, ≤60 seconds.
B: opt-out costs nothing social — implemented as aggregate-shrink (CIDER Stage R). Metric: opt-out rate tracks the rate in the no-leaderboard control condition.
C: consent is opt-in, annually, child-and-parent, with the raw-signal report readable by the child. Metric: informed-consent rate (post-comprehension quiz), not mere signature rate.

Algorithm audit (required — the product scores and recommends)

Proxy variables: electrode contact quality correlates with hair texture and headwear (CIDER #2) → fit artefacts become class/culture proxies inside the score; motion features penalise pupils whose bodies regulate through movement.
Training-data provenance: models trained on the 12-pilot population — mostly NL/FL, one region, one product generation A; generalisation to other neurotypes, tracks and cultures is unmeasured.
Amplification × inequality: attention ranking intersects with tracking (vmbo/vwo) and household income; the leaderboard converts it into a visible hierarchy.

Engagement-value paradox

Yes — the paradox is structural. Coverage rewards every instrumented minute; the humane goal (focused, self-regulating pupils) wants fewer watched minutes. “A product that succeeds when children are watched cannot be evaluated by how much they are watched.” The KPI that must change: coverage → lesson-design improvements per class per term.

Exploitation stacks

Stack	Features	Why worse than the sum	Most affected
The shame circuit	Streak (D2) + live heatmap (D1) + parent percentile report (D3)	Streak creates loss aversion; public heatmap adds shame; parent report moves the shame home and makes it weekly. Each feature removes a different exit.	Adolescents; pupils with anxiety or ADHD
The extraction stack	3-menu opt-out (D4) + 24-month retention + research/insurer sharing (D6)	Consent is hard to find, data is kept long, and the reuse channel is a contract the pupil never saw. Individually defensible; together they are the Focus-Files scenario from §06.	All pupils; future-affected populations

Break-the-stack redesign (one move): make the pupil the data's owner and broker — signals stay on-device, only the pupil's chosen summaries leave it. That single architectural decision breaks both stacks: no archive to leak, no channel to sell, no shame the pupil didn't authorise.

Memorable heuristics (written fresh from this audit)

“Coverage is not care — every added instrumented minute must beat the 25-minute refocus cost.”
“If the opt-out takes three menus, it's a setting we're hiding, not a choice we're offering.”
“Percentiles are for growth charts, not children.”
“Lessons can be bad; children can't be red.”

Humane design statement: “Fokus will stop exploiting attention, emotion, social reasoning and decision-making in minors: no live named-pupil colours, no streaks or leaderboards, no percentiles sent home. It will enable instead: pupils see their own signal first, own their data on-device, and share only what they choose; teachers get lesson-level insight that respects the 25-minute return. We measure success in better lessons, not watched minutes.”

edbx-anotherlensUniversal Methods of Ethical Design #07 — Balance Your Bias · Consider the Opposite · Growth Mindset · #breakmydesign

Solo session, Jonas de Wit (CPO), facilitated in the journal. Design examined: the Fokus Band 2 rollout in its pilot form.

Lens 1 — Balance your bias

Named bias 1: “I assumed every pupil has a personal device and stable charging, because I've never known otherwise. The band syncs over my home WiFi.”
Named bias 2: “I designed the heatmap for the pupil I was — the still, bookish one for whom stillness was attention. Milan, who thinks while moving, was never in the picture I optimised.”
Named bias 3: “I built the parent report for the parent I am: university-literate, Dutch, with a Tuesday evening to read it. The metric I chose (percentile) is the one I'd want. That's me, not our users.”

Lens 2 — Consider the opposite

Affected groups: hijab-wearing pupils (fit), ADHD pupils (baseline), low-literacy families (report), classmates of streak leaders (social cost), and the teachers who must defend the dashboard in parents' evenings.
Assumption to test: “attention scores are valid and fair across neurotypes” — currently untested; the consumer-EEG literature (#14–16) suggests caution.
The person I'm nervous to talk to: the chair of the De Brug parent committee, who asked at the spring meeting what happens to the data “after our kids stop being pupils” — and I answered with the retention policy instead of the truth, which is that nobody had decided.
If the assumption is wrong: the product isn't just unfair, it's confidently unfair — a number that misreads neurodivergent children and ships that misreading to parents weekly.

Lens 3 — Embrace a growth mindset

Learning intention: watch one vmbo lesson through the heatmap with Sanne and the special-needs coordinator, then interview two parents who didn't read the report — before the January roadmap review.
Reframe: “I built a dashboard that misreads the pupils who most need support” becomes “I now know exactly which two conversations to have before the roadmap locks, and what to ask.”

Design decision spec (top 3)

Insight	Specific design change	Measurable spec	Owner	Sprint	Pass/fail
Percentile framing harms without informing	Replace percentiles with lesson-level observations and two support offers	0 percentile comparisons in parent surfaces; ≥2 actionable offers per report; reading level ≤ B1; report comprehension ≥70% in parent testing (n=20)	Head of Learning Science	Sprint 27.4	Comprehension test passes; 0 percentiles in QA sweep
Opt-out is three menus deep	One-tap opt-out + paper card; aggregate-shrink so nobody is “the gap”	Opt-out ≤2 taps, ≤60s; paper card honoured ≤1 lesson; WCAG 2.2 SC 2.5.5 touch targets ≥44×44pt; SC 1.4.3 contrast ≥4.5:1 on opt-out copy	Mobile platform lead	Sprint 27.5	QA: 100% of paths ≤2 taps; accessibility audit green
Score validity across neurotypes untested	Neurotype fairness study before any school-year-wide rollout	Validity study with n≥30 per subgroup (ADHD, non-ADHD, fit-variant); disparate-impact ratio ≥0.8 on score distributions; report published to schools	Head of Learning Science + DPO	Q1 2027	Study filed; release gate blocks rollout if ratio <0.8

“After applying Another Lens, I'm now questioning whether we ever designed for pupils who are not me — and I plan to sit one vmbo lesson with Sanne and the special-needs coordinator before the January roadmap locks.”

edbx-critical-interviewingCarspecken (1996) critical ethnography, adapted — protocol design mode · domain: Product Feature

Research context: the team privately doubts whether the attention score means what the roadmap says. This protocol interviews pilot teachers and the learning-science team to surface what they actually believe — before procurement scales the claim.

Component	Content
Lead-off	“Tell me about a specific lesson this term where the dashboard and your own reading of the room disagreed — what happened next?”
Back-up	“Walk me through last Tuesday: when did you last look at the heatmap, and what did you do because of it?”
Emergency	(if narrating as bystander) “Have you ever been the one who decided to ignore the score — or to act on it — for a named pupil? Take me there.”
Descriptive probes	1) What was on the screen? 2) Who else saw it? 3) What did the pupil do?
Normative probes	1) What should the dashboard have shown instead? 2) When the number felt wrong, what made it wrong?
Evaluative probes	1) Was acting on it appropriate, in hindsight? 2) Would you want your own child scored this way?
Construct-building	Watch for: “it's guessing”, “unfair”, “surveillance”, “theatre” → mirror exactly: “When you say the score is guessing — what is it guessing at, and how do you know?”

Covert categories: “I want to understand whether teachers privately treat the score as theatre without asking ‘do you believe the score?'” · whether heatmaps change teaching or just change looking · what teachers think pupils understand about the band · whether anyone has altered a judgement of a child because of a colour.

Domain-specific (Product Feature): What metric is the heatmap optimised for? What's the perverse scenario if teachers game it (keep the room still → green)? Which pupils were excluded from beta? What are the rollback criteria?

Sequencing: pilot teachers before the learning-science team (teachers name their experience before it is reframed technically); learning-science team before leadership (surface the technical doubts executives haven't heard); Sanne last-to-first across three schools so school politics don't ripple.

Non-obvious harms inventory, guardrails & halt checklist required components

Population	Non-obvious harm the research could surface or cause	Question that surfaces it	Question that prevents surfacing it (avoid)
Pupils (minors)	Interview artefacts describing known score unfairness = discoverable evidence (“we knew”) — subpoena risk in a future claim	Ask about decisions and their reasons, not post-hoc judgements	“So you always knew the scores were wrong?”
Pupils (minors)	Aggregation: teacher quotes + rooster data re-identify a child	Strip class/pupil identifiers at transcription	Any request for “the specific case”
Teachers (employees of pilot schools)	Retaliation risk if quoted to school boards	Explicit confidentiality architecture before content questions	Attributed quotes in procurement decks
Parents (low-literacy)	Consent theatre — signing what they can't read	Consent at ≤B1 reading level + verbal walkthrough	“Just have them sign the form”

Guardrails: interviews run by an external researcher (not the CPO — power asymmetry); recordings deleted after transcription (30 days); no legal-privileged topics; participant redaction window of 14 days; compensation as a school-book voucher for the teacher's class.

Halt checklist: distress → pause and offer break · disclosure of a child-safety issue → safeguarding protocol immediately · “if the board hears this…” → re-confirm confidentiality · drift into protected-attribute territory → return to frame.

08 · THE ETHICS WALL — BAY 2 EDBX · WHAT COULD GO WRONG?

Eight ways to worry well

Free-form worries, card-pair provocation, a twelve-cell stress test, three cautionary episodes, a behavioural forecast, a motivation map, a dark-pattern audit and a shared vocabulary for naming manipulative moves. Each method ends in something the team can do, not just fear.

BAY 2

What could go wrong?

Worrystorming → DAH Cards → Bad Design Canvas → Black Mirror → Inverted Behavior Model → Motivation Matrix → Fair Patterns → Anti-Heroes. Ran on D1–D6 (§01) as the objects under audit.

edbx-worrystormingUniversal Methods of Ethical Design #100 — 8 clusters · pre-mortem for the 2027 rollout · #breakmydesign #identifyvalues

Worry list (16, all eight clusters covered)

Cluster	Worries
🗃️ Data & Privacy	1) “Anonymised” lesson data (D6) is re-identifiable: 30 pupils × lesson times × a school rooster is a population of 30. 2) The 24-month archive (D4) is a breach with children's names on it. 3) Derived scores enable inferences nobody consented to (fatigue, puberty-phase, home stress).
🌍 Environmental	4) 4,200 bands × charging nightly × 2-year replacement = an e-waste and energy line no one has costed. 5) Server-side scoring keeps a data centre running for colours a teacher could ask aloud about.
👥 Community & Social	6) The class leaderboard makes helping a struggling classmate expensive. 7) Parents compare percentiles at the school gate; the percentile becomes the child.
💸 Economic	8) €18/pupil/month starves exactly the schools with the least (per-pupil pricing is regressive). 9) The insurer wellness pilot (D6) is the first step of pricing on children's measured attention.
🔒 Safety & Harm	10) Hardware worn by minors for hours has no independent safety certification in the current claim set. 11) A teacher having a bad day reads a red dot as defiance; the score becomes evidence in a behaviour file.
🧠 Psychological	12) Streak-loss aversion in minors (developmental vulnerability) — the shame circuit of §07. 13) “Red kid” identity formation: the score becomes self-concept by 8 weeks of use.
⚖️ Power & Fairness	14) The opt-out (3 menus, D4) is consent theatre; the pupil is the only stakeholder who can't leave. 15) Fokus's model migrates pedagogy: schools start teaching to the measurable.
🔮 Future Consequences	16) Every signed school normalises the measured childhood for the cohort after; and the archive sits discoverable until 2028.

Clusters → reframed values

Theme (affirmative)	Worries	Severity	Ethical objective	Design constraint	Test question
The Ephemeral Archive	1, 2, 3, 16	🔴 Critical	We commit to data minimisation so the worst leak is an empty folder.	Raw signals never leave the device; derived scores auto-delete in 30 days; re-identification tested by an external lab quarterly.	If we were breached tomorrow, what could actually be leaked? (Target answer: nothing named.)
A Childhood Without Percentiles	7, 12, 13, 6	🔴 Critical	We commit to feedback that describes lessons, not ranks children.	No percentiles, leaderboards, or public colours anywhere in the product; streaks removed from pupils under 16.	Could a parent describe the report without comparing their child to another?
Consent That Costs Nothing	14, 15	🔴 Critical	We commit to exit as one tap and silence as one card.	Opt-out ≤2 taps + paper card; aggregate-shrink; no social dependence on measured pupils.	Would Milan opt out if it were free? (Ask him.)
Hardware Worth Wearing	4, 5, 10	🟡 Monitor	We commit to a lifecycle we can defend to a 14-year-old.	Independent safety cert before 2027 term; take-back recycling; per-room mode with zero bands as a SKU.	What does a band cost the planet per lesson-minute?
Price Nobody Regrets	8, 9	🟡 Monitor	We commit to pricing that narrows gaps.	Room SKU priced per class not per child; insurers excluded from any data contract.	Does the poorest school in the pilot pay less than the richest?

Mitigations for 🔴 clusters: on-device processing and 30-day deletion (data architecture decision, Q1 2027); percentile-free parent reports (Sprint 27.4); one-tap opt-out (Sprint 27.5); consult: Dutch DPA guidance on minors' consent, the De Brug parent committee, and the school's paediatrician advisor. Who to tell first when a 🔴 worry lands: the pupil council — they don't know the streaks exist yet.

Worrystorm summary: the core challenge is not any single feature — it is that Fokus is building an archive of childhood while telling itself it is building a coach. Minimise the archive (30 days, on-device, nothing named), remove the ranking (no percentiles, no leaderboards, no public colour), make exit free (two taps or one paper card) — then the product's real value, lesson insight, survives every remaining scenario in this journal.

edbx-dah-cardsDesign Against Humanity — audit (Modes 1+2) · #breakmydesign

DAH card pairs

#	Category	Black side (as it ships)	White side (in the voice of someone harmed)
1	AddictionSystemic	“Focus streak” with loss-state copy when broken	“I stayed in at break to keep my streak. My friends stopped asking why.”
2	Deception	“Anonymised research programme” (D6)	“It was never anonymous — my daughter's rhythms sit in a dataset under a code the insurer's pilot can read.”
3	Coercion	Three-menu opt-out, default-on everything (D4)	“You can refuse — if you find the door. I needed my son to help me find it.”
4	Surveillance	Live named-pupil heatmap (D1)	“Forty minutes of red, in front of everyone, while I was doing my best thinking of the term.”
5	Exclusion	Stillness-baseline scoring + one band size	“My score is red because my body works differently. The app decided I'm the problem.”
6	Systemic	Weekly parent percentile report (D3)	“By Christmas, Milan wasn't a boy who struggled some weeks. He was 63rd.”

Consequence map (selected rows; likelihood · severity · who bears it · design awareness)

Consequence	Likelihood	Severity	Who bears it	Design awareness
Streak-driven lesson choices (pupil suppresses breaks)	Common	Meaningful	Pupil	Known, optimized for
Public colour shame in front of peers	Common	Serious	Pupil	Known, not mitigated
Parent–child conflict from percentile reports	Possible	Meaningful	Pupil + family	Not considered
Re-identification of “anonymised” archive	Possible	Irreversible	All pupils, future	Known, dismissed
Fair-mechanism misread of neurodivergent pupils	Common	Serious	Named subgroup	Not considered (pre-CIDER)
School budget crowding-out (€18 × 30 × 8 subjects)	Common	Meaningful	Low-budget schools	Not considered

Sorting verdict

Item	Verdict	Rationale
On-device processing; lesson-summary analytics	Ethical	Genuine value, minor harms, mitigated.
Room-level sensing pilot; teacher dashboards (aggregate)	Ethical / Unsure	Real promise; watch teacher workload.
Streaks + leaderboard (D2)	Unethical	Exploits minors' loss aversion for a metric; not defensible publicly.
Live named heatmap (D1)	Unethical	Public shame of a named child in real time; no mitigation exists.
Percentile parent report (D3)	Unethical	Ranks children inside families; harms without informing.
24-month retention + research/insurer reuse (D4, D6)	Unsure → Unethical	Benefits are speculative; the archive risk is irreversible.

Microcopy rewrites

Before: “Share anonymised data to help science improve learning for everyone.” → After: “We keep your lesson data for 30 days on your band. After that it is deleted. We will not sell, share or study it.”

Before: “Opt out of data collection” (Settings → Privacy → Preferences). → After: “Stop measuring me — one tap, works today.” (Home screen.)

What changed? “Card 4 was the surprise: we argued about whether the heatmap is even ‘surveillance' since it is ‘care data'. Writing the white side in Sanne's voice ended the argument. And we disagreed about card 1 — one founder defended streaks as ‘motivation'. Sorting it Unethical was the first time the roadmap lost an argument to a card.”

Ship-ready recommendation: replace D2 entirely — delete streaks and leaderboards from the pupil app in Sprint 27.5 and replace with a weekly self-reflection (“what made this lesson stick?”), authored by the pupil, shared only by the pupil. Ticket exists; three taps of design work.

edbx-bad-design-canvasManos, Bad Design Canvas (CC BY-ND) — 12 cells · pre-launch audit · #breakmydesign #evaluateoutcomes

Cell	Finding	Severity
1 · 🌍 Cultural appropriation	The “neutral attentive body” encodes one culture's classroom etiquette as biology; headwear practices are treated as fit problems, not design briefs.	🟡 Possible
2 · 🩹 Band-aid	Measures the pupil instead of the feed, the schedule and the lesson — the actual causes in §03's systems layer. A gauge on the symptom.	🔴 Significant · systemic
3 · 🔒 Unfair control	The instrumented party (child) controls nothing: no opt-in, 3-menu opt-out, no data bargaining power; contract is school↔company.	🔴 Significant · systemic
4 · ⚠️ Exploitation	Insurer/insurer-adjacent reuse (D6) converts minors' physiological signals into commercial asset without their understanding.	🔴 Significant
5 · 🔄 Inefficiency	Teacher attention now splits between lesson and dashboard; charging logistics eat class time; three admin systems per school.	🟡 Possible
6 · 🌱 Environmental & social	Unmodelled: battery churn, e-waste, night charging at scale; 4,200 devices growing to 500k+ on the roadmap.	🟡 Possible
7 · 👥 Stakeholder abandonment	Pupils never consulted; special-needs staff absent from design reviews; substitute teachers untrained on what colours mean.	🔴 Significant
8 · 🛡️ Decreased safety	No independent hardware safety cert; breach exposure (24-month archive); behavioural-file misuse by schools.	🔴 Significant
9 · 🚫 Inappropriate	Percentile comparison of minors to peers in family-facing weekly reports sits badly against Dutch youth-care norms (no diagnosis, just comparison).	🟡 Possible
10 · 😐 Boring	The dashboard aesthetic is 2016 HR analytics. The radical version — pupils authoring their own attention stories — would be more interesting to build and to use.	🟡 creatively overdue
11 · 🏠 Displacement	Displaces teacher judgement with score authority; displaces paediatric assessment pathways (D5's “ADHD-adjacent insights” shadow clinical referral).	🔴 Significant
12 · ⚖️ Inequity	Per-pupil pricing is regressive; benefits accrue to data-rich, well-resourced schools; misreads concentrate on neurodivergent and minority-practice pupils.	🔴 Significant

Count: 6 🔴 · Top 3: (1) Unfair control — the child has no lever (systemic); (2) Band-aid — wrong unit of measurement (systemic); (3) Displacement — shadows clinical referral and teacher judgement (addressable with claims discipline). Clusters: 3+4+12 reinforce each other (control → exploitation → inequity).

Redesigns: 3 → opt-in consent + pupil data ownership (involve: pupil council + DPO) · 2 → fund the room-level unit test (involve: Bakkali + Sanne) · 11 → hard claims wall: “Fokus never suggests diagnoses; Fokus suggests referrals” (involve: paediatric advisor). Hard question: does cell 2 suggest this shouldn't be built at all? Not at all — but in its current child-instrumented form.

Bad design statement: “The most significant unintended consequence of Fokus Band 2 is the removal of control from the one person it instruments — the pupil — which converts a coaching tool into a childhood archive others own. This requires the data-ownership redesign (pupil-brokered signals) or the room-level pivot; shipping both child-instrumentation and child-powerlessness is the one combination this product cannot defend.” Verdict: Rethink fundamentally for the child-level line; Ship with changes for room-level + clinical.

edbx-black-mirror-brainstormingKlassen & Fiesler (2022); Mauldin (2018) — speculative red-team · anti-goals as testable constraints

What could go wrong (10, across three lenses)

Economic How could an insurer use “anonymised” attention archives to price family policies in 2032?
Economic How could a reseller of pupil attention data undercut Fokus by selling better archives than it keeps?
Economic How could streak multipliers turn classroom patience into a purchasable asset (pay to keep your streak)?
Political How could a ministry fuse attendance, attention and behaviour files into a single “school readiness” record?
Political How could a well-meaning minister launch “national attention standards” and rank schools by pupils' measured focus?
Political How could attention data be subpoenaed in a custody case?
Social How could a teacher under pressure use red dots to justify seating, exclusion or report comments?
Social How could “green” become a class identity and “red” a slur by sixth grade?
Social How could parents coach children to perform stillness — teaching kids that attention is a performance for sensors?
Social How could the one opted-out pupil become the class's responsibility, then its suspect?

Episode 1 — “GREEN MEANS GO” lens: Social — tap to read

Protagonist: Milan, 14, vmbo De Brug — wants to be an electrician; loves the practical course where he has never once been “red”. Antagonist: the system itself — a head-of-year who needs measurable improvement by June. Supporting: Sanne, who keeps the heatmap on for the team's sake.

Context: De Brug, school year 2027–28, second year of the rollout.

Beginning: the leaderboard works. Milan's class is third; his streak is a small pride. The staffroom board shows each class's weekly focus score.

Middle: Milan's streak breaks during his grandmother's funeral week. The class slips to fifth. “Motivational” interventions begin: seating changes for red pupils, streak-repair Wednesdays. Milan starts performing stillness — and loses the practical course hour where he was never red.

End: in June the class “recovers” to second. The head-of-year's review deck calls it an intervention success. Milan's report card carries a new line: attention: improving. Nobody remembers what he used to be good at.

Pitch quote: “Milan's class is two leaderboard places from the top, and the school has a plan to get there. It involves his grandmother's funeral, a seating chart, and a number that says he wasn't trying. He was trying. That's the problem — the number can't see it.”

Episode 2 — “THE FOCUS FILES” lens: Economic

Protagonist: Annemieke Oosterhuis, retired teacher turned municipal data auditor, asks a boring question about an invoice. Antagonist: a wellness-data broker and the growth team that signed it. Supporting: Lotte Donders, Fokus DPO, three years gone.

Context: a Dutch province, 2032; procurement freeze pending audit.

Beginning: Annemieke traces an “analytics services” line item to a broker. The contract calls the data derived, anonymised, wellness-grade.

Middle: four re-identification attacks succeed with school roosters and public timetables. The “anonymised” archive of 2019–2026 — legally hers to request, nobody's to read — maps pupils' bad years to home addresses. The broker has already licensed it twice.

End: contracts freeze; the archive is destroyed; two families sue. The audit's last line notes the archive was, on paper, compliant the whole time. The harm never needed to be illegal.

Pitch quote: “Annemieke audits school invoices for a living, and one line item won't add up. It leads to an archive of every focused and unfocused minute a decade of children spent in classrooms — anonymised on paper, identified in an afternoon. The data was collected legally, sold legally, and will hurt legally. Nobody broke the rules. That's what scares her.”

Episode 3 — “THE QUIET ROOM” lens: Political

Protagonist: Sanne, eleven years teaching, who keeps a paper register out of habit. Antagonist: a national “Schoolreadiness” programme that wants attention streams in the safety pipeline. Supporting: her pupils, who organise before the adults do.

Context: the Netherlands, 2034, a coalition government “modernising” school safety.

Beginning: the programme asks vendors for “early-warning indicators”. Fokus's successor company complies, reading only aggregates — but the pipeline re-attributes.

Middle: a pupil list circulates: “elevated distractibility profiles”. Sanne recognises the names: all from one street, all from one community organiser's family. She refuses to hand over her register; the class notices.

End: the school goes on “head strike” — bands off, walkouts at the bell. The programme is paused, not killed. Sanne's final scene: she reads the register aloud, names, not numbers, and the room is silent in the old way.

Pitch quote: “The ministry promised the data would only ever be seen in aggregate. Lists have a way of un-aggregating. When her pupils' names start circulating, a teacher with a paper register becomes the last firewall between a classroom and a watchlist.”

Anti-goals (testable, traced)

Anti-goal	Trace	Test
No named pupil's state is ever visible to anyone but that pupil (and, with their tap, their family)	E1, middle	Dashboard QA: zero named-pupil live colours in any shipped view; code-level test in CI
No ranking of minors by measured attention, in any surface, ever (incl. “friendly” leagues)	E1, beginning	Product grammar audit: no comparative aggregate renders with a named minor
“Anonymised” means re-identification-tested, not declared	E2, middle	Quarterly external re-ID lab on every shared dataset; ≥0 successes to publish
No derived child data leaves the EU school context for commercial reuse	E2, end	Contract registry: zero wellness/insurer/commercial clauses; annual audit
No archive longer than the pedagogical need (30 days derived, 0 raw)	E2, premise	Retention lint on data schema; storage grep in release review
Attention data never joins any safety, attendance or behaviour pipeline	E3, beginning	Data-flow diagram review per release; no egress to government safety systems
Opt-out is free, social and instant	E1/E3, opt-out pupil	Opt-out ≤2 taps ≤60s; mystery-shop twice a term; opt-out pupils invisible to aggregates
A pupil can always be unmeasured without being conspicuous	E3, register scene	“Off week” class ritual shipped; usability test with opted-out pupils

Load-bearing anti-goals (violate any and the concept collapses): 1 no named-pupil live visibility; 2 no ranking of minors; 6 no pipeline fusion. These go to leadership before the Series C deck does.

📄 The featured poster (Episode 1, “Green Means Go”) is rendered as its own title card: fokus-poster.html — open it alongside this journal.

08 · BAY 2, CONTINUED EDBX · WHAT COULD GO WRONG?

From worries to mechanics

The first half of Bay 2 generated the worries. The second half finds the mechanisms: what each prompt actually incentivises, which motivations it pulls, which dark patterns are already installed, and what the team calls these moves when it sees them in a review.

BAY 2 · CONT.

What could go wrong? — the mechanics

edbx-inverted-behavior-modelBJ Fogg B=MAP, inverted for audit — feature under review: the pupil app + teacher dashboard · #breakmydesign #evaluateoutcomes

Stakeholder map (Step 0)

Stakeholder	Relationship	Power they have	Power they lack
Primary users (pupils 12–18)	Wear the sensor; see streaks	Exit (theoretically), complaint	Cannot influence prompt design, the metric, or the contract
Vulnerable subgroups: ADHD/autistic pupils; anxious pupils; low-literacy pupils	Scored against a stillness norm	Advocate adults, sometimes	Cannot re-baseline the score; misreads hit them first
Non-consenting bystanders: classmates of streak leaders; opted-out pupils	Ranked beside, compared to, responsible for	Peer pressure, collective refusal	No data rights despite being scored into aggregates
Workers in supply chain: annotators labelling classroom footage for the model A	Expose pupil behaviour for wages	Exit	Cannot see end-use; no say in retention
Absent regulators: Dutch DPA, EU AI Act enforcement, inspection of education	Authority not in the room	Fines (€35M/7%), procurement bans	Post-hoc only, if notified
Future-affected: pupils at 20 with a decade of traces; the 2040 cohort	Inherit the archive and the norm	None yet	Did not consent to the precedent

Prompt inventory

Prompt	Trigger mechanism	Intended behaviour
P1 · Focus streak counter	Daily lesson minutes → streak state; loss copy on break	Consistent lesson engagement
P2 · Class leaderboard	Weekly aggregate → visible rank in app + staffroom	Collective motivation
P3 · Live teacher heatmap	Per-minute score → colour on class screen	Timely teacher intervention
P4 · Mentor nudge	Score pattern → “try a micro-break” push	Self-regulation skill-building
P5 · Parent weekly report	Sunday evening push with percentile	Informed home support

Worst possible design (the inversion artifact): “If this product were designed to maximise engagement at children's expense, it would: put the streak on the lockscreen with a grief mascot when it breaks; multiply streaks for wearing the band during breaks; publish the class ranking with names, last-place highlighted, and a demotion mechanic; give teachers a live wall view with pupils' photos pulsing red; send parents a Sunday push ‘Milan scored below 4 classmates this week — here's what other parents do'; make the opt-out require a signed form and a meeting; and let the Mentor message pupils directly at 21:00 on school nights, ‘don't lose tomorrow's streak.'”

Convergence check

Prompt	Worst-version element	Actual design	Convergence
P1 streak	Grief copy + lockscreen + multipliers	Streak with loss-state copy; no multipliers yet	🟡 Adjacent — one copy change away
P2 leaderboard	Named rank with demotion	Named class rank, staffroom + app	🔴 Converged
P3 heatmap	Live named colours, photos pulsing	Live named colours, no photos	🔴 Converged
P4 Mentor	Direct evening nudges to pupils	In-app nudges during lessons only	🟡 Adjacent — push channel exists
P5 parent report	Peer comparison + comparative coaching	Percentile vs school year	🔴 Converged (comparison is the feature)

Behavior forecast & per-stakeholder consequence map ✅⚠️🔴🟣 taxonomy

Prompt	Motivations (≥2)	✅ Intended	⚠️ Unintended neutral	🔴 Unintended harmful	🟣 Social amplified
P1	Loss aversion; achievement; habitual	Regular attendance to lessons	Pupils check band state between lessons	Suppressing genuine breaks to protect the streak; anxiety on break days	Streak culture: break-taking becomes visible deviance
P2	Social acceptance; power	Collective effort	Friendly inter-class banter	Helping a struggling classmate costs the helper rank; last place = class shame	Class identity by colour; “red class” stigma (vmbo over-represented)
P3	(teacher) fear of looking inattentive-class; duty	Early support	Teachers glance more than they act	Public shaming justified by the dot; score enters behaviour notes	Whole class performs stillness when observed — “green theatre”
P4	Autonomy; mastery	Micro-breaks taken	Nudges ignored	Nudge timing trains pupil availability to the app	Mentor-speak enters pupil slang (“the band says I need a break”)
P5	(parent) fear, responsibility; love	Informed support	Reports unread (low-literacy homes)	Percentile conflict at home; child's bad week reframed as deficit	Gate comparison culture; tutoring arms-race on “focus”

Behavioral cascade (for P1's 🔴 finding): Week 1: Milan keeps the streak. Weeks 2–6: he times toilet breaks to streak windows; suppresses fidgeting that keeps him regulated. Months 1–3: the app is “a tool he owes”; missed streak triggers guilt he narrates to friends. Months 3–12: “I'm a 40-streak kid” — identity; the practical course (never red, never counted) fades in importance because it doesn't feed the number. Years: streaks normalised as legitimate school motivation; “loss aversion as pedagogy” becomes an industry pattern nobody remembers proposing.

Rationalizations check:

🔴 Finding	The story	Why it fails
Streaks exploit minors' loss aversion	“Streaks motivate — pupils tell us they love them.”	“Love” is downstream of variable reward; the preference was manufactured by the design being defended. Also: the pupils who hate them can't say so in a classroom where the streak is public.
Named live heatmap enables public shame	“Teachers are professionals; they won't shame.”	The design doesn't require bad teachers to produce the harm — visibility does the work. One tired Tuesday is a design test the product fails by existing.
Percentile parent reports rank children	“Parents want to know where their child stands.”	“Where they stand” is a school-choice answer, not a weekly push. Frequency and comparison are our choices, not the information's requirements.

Behavioral risk summary (top 3)

Green theatre → attention-as-performance. P3 + P2 jointly teach that attention is a performance for sensors; intrinsic motivation for learning is displaced (overjustification effect) — the exact competence the product claims to build.
The shame circuit. P1 + P3 + P5 (§07's stack) — adolescent loss aversion, public visibility, home amplification.
Helping becomes expensive. P2 penalises prosocial behaviour; classrooms normalise leaving the struggling behind to protect rank.

Redesign directions

Direction A (minimal): kill P2's named rank (aggregate trend, no names); move P3 to end-of-lesson; strip loss copy from P1. Metric that changes: weekly active streak-keepers ↓ (the metric was the harm); pupil-reported classroom safety ↑. Resistance: “engagement proof” for procurement decks.

Direction B (structural): invert ownership — pupil app becomes the primary surface (their signal, their dashboard, their sharing choices); teacher gets room-level trends only; parent gets pupil-authored summaries. Metric: share-rate chosen by pupils (consent as engagement). Resistance: the €220M plan assumes institutional dashboards are the product.

Behavior forecast statement: “This feature is likely to produce regular lesson engagement and genuine early-warning value for a subset of teachers, but also incentivises streak-protecting self-suppression through loss aversion, public stillness-performance through named live ranking, and percentile anxiety at home — with consequences concentrated on ADHD, anxious and low-literacy pupils, and amplified at classroom scale into colour-based class identity.”

edbx-motivation-matrixZhu (2020) — MOTIVATION TYPE × USER @ CONTEXT · #identifyvalues

Contexts chosen: (1) Milan @ supportive home, loves practical course · (2) Milan @ stressed home, low-literacy parents · (3) Sanne (teacher) @ exam-pressure term. Focus question: whose motivation does the product actually pull — the learner's, or everyone around the learner's?

Motivation	Milan @ supportive home	Milan @ stressed home	Sanne @ exam term
Achievement	🟡 Achievement + the streak replaces the practical-course pride — progress now counts only if the band sees it	🔴 Achievement + broken streaks arrive as one more failure in a life already short of wins	🟡 Achievement + class focus-score becomes her KPI proxy; teaching to the green
Social acceptance	🟡 Social acceptance + leaderboard gives belonging-by-colour; green is inclusion	🔴 Social acceptance + red in a green class marks him; peers enforce what the app starts	🟡 Social acceptance + staffroom league table makes her class's colours her reputation
Fear	🟡 Fear + mild loss aversion on streak break; recoverable	🔴 Fear + the Sunday push lands on a household where the phone is also the bailiff's channel	🔴 Fear + “your class is 14th” arrives as inspection-adjacent threat
Power	🟡 Power + pupils can't control the sensor; no lever at all (the finding)	🔴 Power + zero lever; opt-out needs a literate adult	🟢 Power + the dashboard gives her data authority she never had — used well, real
Incentive	🟡 Incentive + streak multipliers promised in roadmap — extrinsic crowd-out of intrinsic interest	🔴 Incentive + nothing: the paid version's perks don't reach his household	🟡 Incentive + “insight” promise vs another admin surface

Ethical risk summary: the same design element reads as gentle for a resourced pupil and punitive for a stretched one — five of fifteen cells are 🔴 and every 🔴 lands in context 2. The product's motivational load is regressive. Redesign directions: 1) remove comparison from all Fear/Achievement prompts for under-16s (cells: C2-row-Fear, C1-row-Achievement); 2) give pupils one real lever (self-anchoring + off-week ritual — the CIDER fixes) so the Power row stops being the product's blind spot; 3) teacher-facing: make the room view diagnostic, never ranking, and keep it off the staffroom wall.

Missing contexts: pupils who cannot be banded at all (fit/comfort) — the unmeasured experience of a measured classroom; parents of opted-out pupils (are they punished?); substitute teachers (no baseline, all colours). Each is a real stakeholder the matrix exposed only by trying to fill the column.

edbx-fair-patternsGray et al. (2024) ontology + fair countermeasures — audit of pupil app, opt-out flow and consent surfaces

Dark pattern audit

#	Pattern	Manifestation	Agency stolen
1	Roach motel	Data reuse is default-on; opt-out lives three menus deep (D4)	Leaving on the pupil's own terms
2	Privacy zuckering	Consent screen says “help improve learning”; page 4 authorises research & insurer reuse (D6)	Informed consent over data flows
3	Misdirection	Parent report leads with percentile (big numeral); the “what this can't tell you” note is a footer link	Honest weighting of the number
4	Confirmshaming	Opt-out copy: “I'd rather not help my class today”	Refusal without guilt
5	Trick question	Checkbox: “I prefer not to receive fewer insights”	Plain reading of a refusal

Root-cause reframes: Roach motel exists because retention was owned by the growth team, whose KPI is measured minutes — every exit was someone else's bug. Privacy zuckering exists because the research consent was drafted to survive procurement review, not to be understood by a 14-year-old. Confirmshaming exists because the opt-out rate landed on a dashboard next to churn, and copy is cheaper than a better product.

Legal compliance mapping + vulnerable-population harm matrix named statutes, jurisdictions

Pattern	Statute / article	Jurisdiction	Penalty exposure	Risk
Roach motel / default-on reuse	GDPR Art. 7(3) (withdraw consent as easily as give) + Art. 25 (data protection by default)	EU / NL	Up to €20M or 4% global turnover	🔴
Cognitive/emotional inference of pupils	EU AI Act Art. 5(1)(f) (prohibited practice, education); fines under Art. 99	EU	Up to €35M or 7% turnover	🔴
Children's data consent architecture	GDPR Art. 8 + Recital 38 (children merit specific protection)	EU / NL (age 16 threshold as set in Dutch implementation)	Enforcement orders; processing suspension	🔴
Percentile misdirection in family reports	EU Omnibus Directive 2019/2161 (fair information practices) — adjacent	EU	Member-state consumer-authority fines	🟡
Confirmshaming / trick question	FTC Act §5 (deceptive practices) — US pilot school (Massachusetts)	US-federal	Consent decrees; per-violation penalties	🟡

⚠️ Flagged 🔴 in two or more jurisdictions: patterns 1–2 → must-fix before next release. The skill's rule is mechanical; so is the fix.

Pattern	Disproportionately harmed populations	Mechanism of differential harm
Roach motel	Low-literacy families; pupils without advocates; non-native speakers	Exit requires exactly the resources (reading, navigation, adult time) these households have least
Privacy zuckering	Minors generally (developmental consent capacity); first-generation school families	Consent capacity is still forming; page-4 clauses trade on trust in school branding
Misdirection	Parents under time poverty; anxious parents	The numeral is designed to be seen; the caveat is designed to be scrolled
Confirmshaming	Adolescents (peer-norm sensitivity); conscientious pupils	Guilt copy weaponises the exact trait schools want to encourage

Countermeasures, defaults & copy rewrites

Pattern	Fair countermeasure	Default audit
Roach motel	Summary path: one Privacy page, everything on it, one tap each	🔴 Harmful → Protective: reuse default-off
Privacy zuckering	Adequate information: consent in 120 words + 1-tap per flow, no page-4	🔴 → Protective: research reuse opt-in, annually
Misdirection	Nonobstructive info: percentile removed (per §07); leading with lesson observations	🟡 → Protective: observation-first layout
Confirmshaming	Plain language: “Stop measuring me — one tap, works today”	🟡 → Protective: neutral refusal copy
Trick question	Plain language: one checkbox per decision, double-negative ban in content linter	🟡 → Protective: linter rule ships with the design system

Implementation sequence (ranked: Legal Risk × User Impact ÷ Effort)

Pri	Fix	Legal risk	Impact	Effort	Target metric	Owner	Sprint
1	Default-off data reuse + single Privacy page	🔴	High	Low	Reuse-consent comprehension ≥85% (test n=40); opt-out ≤2 taps	DPO + Mobile lead	27.5
2	AI-Act claim scrub of Mentor (no state inference in education surfaces)	🔴	High	Med	0 cognitive-state claims in EU surfaces; legal sign-off pre-release	General Counsel	27.6
3	Neutral copy pass + double-negative linter	🟡	Med	Low	100% surfaces pass linter; support tickets re: consent ↓50% in 90 days	Design lead	27.6

Fair design statement: “This product currently ships five dark patterns that take consent, refusal and honest reading from the exact people with least power to reclaim them — minors and low-literacy families. The countermeasures are small: default-off reuse, one-page consent, neutral copy, observation-first reports. Together they restore the agency the dashboards quietly took.”

edbx-anti-heroesMehta, Chivukula, Gray et al. (2024), arXiv:2405.03674 — evaluation mode · #breakmydesign

1 · Frame

Design under review: the consent/opt-out journey + the “caring” surfaces (parent report, Mentor copy). User task & emotional state: a parent deciding whether their child should be measured — anxious, time-poor, trusting the school. Business pressure: measured-minutes KPI and Series C data story. Mode: evaluation.

2 · Anti-Hero hits

Card	What is happening	Where	Why anti-user
Trap-Setter	Entry is a signature at enrolment; exit is three menus, one confirmshame, no confirmation to the school	Settings → Privacy → Preferences	The off-ramp is structurally harder than the on-ramp
Empathy Manipulator	“We want Milan to thrive — this report helps you help him” atop a percentile	Parent report header	Care language pre-loading a comparison the data can't support
Camouflager	“Anonymised” describing re-identifiable lesson traces	Research programme page (D6)	The word does the work of the audit that never happened
Two-Faced	Privacy page says “your data stays yours”; architecture keeps 24 months and shares “anonymised” channels	Website vs D4/D6 reality	The promise and the pipe disagree

3 · Hero counter-moves

Paired card	Concrete change	Why it restores value
Empowerer	“Stop measuring me” lives on the pupil home screen; paper card equivalent; school notified in-product	Exit symmetric with entry
Life Coach	Report leads with one strength, one friction, two offers; comparison removed	Information, not pressure
Unveiler	Replace “anonymised” with the re-ID test result and date	Claims audited, not adjectived
Consistent Communicator	Privacy page and data schema generated from the same source of truth	Promise = pipe

4 · First three changes to ship

Ship the pupil-side opt-out (two taps) — Sprint 27.5.
Rewrite the parent report header without care-language-over-comparison — Sprint 27.4.
Publish the re-ID test result on the research page or stop using the word “anonymised” — this sprint.

5 · Necessary friction kept

Age-appropriate dual consent (child + parent) stays — legally required and genuinely protective; confirm dialog before a parent removes measurement stays, phrased neutrally. Neither is a Trap-Setter; both are load-bearing doors.

09 · THE ETHICS WALL — BAY 3 EDBX · HOW DO WE DECIDE?

Seven ways to decide
with the worries on the table

The audit produced findings; decisions still have to be made. Bay 3 runs the Stanford Ethics Toolkit's full five-tool chain, then weighs the same decisions through normative ethical theory, a four-category compass, a five-axis prism, stakeholder value conflicts, team culture levers — and finally reads the design team's own meeting history for its ethical trajectory.

BAY 3

How do we decide?

edbx-stf-etStanford McCoy Family Center Ethics Toolkit V1.1 (2025), CC BY 4.0 — full chain: Explore → Evaluate → Decide

Chain status: Future Story ✓ · Impacts Explorer ✓ · Ethics Frame ✓ · Ethics Gauge ✓ · Weighing Options ✓ — Entry phase: Explore (no prior frame existed).

Tool 1 — Future Story Kenn Adams' story spine, told from 2031

1 · “Once upon a time…” Schools in Northwest Europe faced a real, measured attention problem — 47-second screen switching, a phone-ban decade, teachers asking for eyes in the back of their heads.

2 · “Until one day…” Fokus put a soft band on 4,200 pupils and sold teachers a live attention map; the Series B deck said 2,500 schools by 2040.

3 · “And because of that…” Thousands of lessons were redesigned around real engagement data; quiet struggling pupils were noticed weeks earlier; two universities published with the data; teachers reported fewer surprises at parents' evenings.

4 · “But also…” Pupils learned to perform attention for a sensor; ADHD and hijab-wearing pupils were systematically misread; parents' Sundays filled with percentile dread; one breach-ready archive of childhood sat on two continents.

5 · “In turn…” The misreads compounded into track advice and behaviour files (value: dignity ⬇️). The archive normalised measured childhood, and competitors raced to the bottom on consent (autonomy ⬇️, trust ⬇️). And in turn, the pupils who had been “green theatre” champions arrived at university unable to notice their own drift — the skill was scaffolded outside them (well-being ⬇️, autonomy ⬇️).

6 · “One thing we could have done differently is…” Instrument the room instead of the child from day one, and let pupils own their signal — the same data for the teacher, none of the archive for the industry.

→ Carries forward to Tool 2

Action/Creation: pupil-worn EEG attention band with teacher heatmap, streaks, parent percentile reports.
Direct effects seeding Effects ring: engagement gains; early-warning catches; misreads of neurodivergent pupils; family percentile anxiety.
Cascading impacts seeding Secondary Effects ring: performance-of-attention culture; measured-childhood normalisation; archive & resale risk; teacher authority shift.
Values touched: well-being, autonomy, trust, dignity, privacy, justice.

Tool 2 — Impacts Explorer futures-wheel, values-annotated

Direct effect	Secondary effects	Values hit
Teachers see attention live and intervene earlier	a) interventions arrive before consent or context · b) pupils manage the teacher's screen, not their attention · c) teacher skill migrates to dashboard reading	Autonomy ⬇️ (a, b) · Relationships ↔ · Trust ⬇️ if hidden (c)
Pupils wear a physiological sensor all lesson	a) stillness-performance displaces self-regulation · b) misreads concentrate on neurodivergent pupils · c) body-sensor normalisation for life	Well-being ⬇️ · Justice ⬇️ (b) · Dignity ⬇️
Parents receive weekly percentiles	a) home conflict reframed as data · b) comparison culture at the gate · c) tutoring arms-race on “focus”	Relationships ⬇️ · Justice ⬇️ (c) · Trust ↔
Attention data retained & reused (D4/D6)	a) re-identification of “anonymised” archives · b) insurer/commercial gravity · c) precedent for the next cohort	Privacy ⬇️⬇️ · Responsibility ⬇️ · Justice ⬇️
Streaks & leaderboards gamify attention	a) helping classmates becomes costly · b) streak anxiety · c) extrinsic crowd-out of intrinsic interest	Relationships ⬇️ · Well-being ⬇️ · Autonomy ⬇️

Group-specific: ADHD pupils — misread as inattentive while hyperfocused (justice ⬇️); vmbo classes — leaderboard aggregates into track stigma (justice ⬇️); low-literacy families — report is decoration or provocation (justice ⬇️); teachers — new accountability without new authority (responsibility ↔).

→ Carries forward to Tool 3

Benefits: early warning; lesson redesign evidence; fewer surprises for families.
Harms: misreads by design; shame circuits; archive risk; crowd-out of intrinsic attention.
Values: autonomy, justice, privacy, dignity, well-being, trust, relationships.
Stakeholder groups & differential impacts: ADHD/autistic pupils (misread); low-literacy families (excluded from consent); vmbo classes (stigma); teachers (accountability shift).
Fairness patterns: harms track existing marginalisation; benefits track existing resources.

Tool 3 — Ethics Frame (9 sections) full worksheet

Action: scale pupil-instrumented attention measurement to 2,500 schools by 2040 (current design D1–D6).
Values driving: educational fairness (“no pupil invisible”), efficiency, evidence-based teaching. Values impacted: privacy ⬇️, autonomy ⬇️, dignity ⬇️, justice ⬇️ (misreads concentrate on the already-marginalised), trust ⬇️ (hidden reuse), responsibility ⬇️ (duty shipped to teachers). Values those affected care about most (from CIDER/IBM): being seen accurately; exit without cost; not being compared.
Benefits (extent): early-warning catch (magnitude med; scope all pupils in measured classes; likelihood med-high; duration years); lesson-design evidence (mag med; scope teacher-level; likelihood high; duration ongoing).
Harms (extent): systematic misread of neurodivergent pupils (mag high for the individual; scope ~10–15% of pupils; likelihood high; duration until corrected); shame circuit (mag high; scope wide; likelihood common; duration formative years); archive exposure (mag catastrophic in worst case; likelihood low-med; duration irreversible). Unintended uses: behaviour files, custody subpoenas, insurer gravity. Failure mode: dashboard outage → school blind; worse, school confident.
Who is affected: spectrum from great benefit (data-comfortable vwo classes, resourced schools) to great harm (ADHD/autistic pupils, hijab-wearing pupils, low-literacy families, opted-out minorities). Fairness: already-marginalised groups face outsized harms; privileged groups capture most benefit; some cannot access the benefit at all (misread ≠ insight).
Values prioritised: justice and autonomy over coverage and ARR; transparency over the data story.
Expand benefits: room-level mode so insight survives without child data; open the scoring model to schools; publish validity bounds per pupil subgroup.
Reduce harms: prevent — no named live colours, no ranking, 30-day on-device retention; protect — one-tap exit, aggregate-shrink, independent re-ID audits; support — fund analog alternatives in poor schools; referral-not-diagnosis wall in Mentor.
Bottom line: pause the child-level scale-up; ship the room-level pilot and the consent redesign; decide medical-vs-wellness at board level in Q1 2027. Specific enough to gate the roadmap.

→ Carries forward to Tool 4

Action: as above. Benefits with extent: early-warning (med/high), lesson evidence (med/high). Harms with extent: misreads (high/high), shame (high/common), archive (catastrophic/low-med). Fairness: harms concentrate on marginalised; benefits on resourced. Autonomy/dignity: pupils are instrumented without levers; shame is public and comparative.

Tool 4 — Ethics Gauge 12 spectra, marked

Dimension	Spectra (mark − / ○ / +)	Reflection note
Beneficial?	Individual benefit: ○ moderate over existing tools	Measure effects against Sanne's judgement, not vendor claims; investigate whether green theatre negates the benefit entirely.
	Breadth: ○ many pupils, unevenly
	Likelihood of benefit: −/○ unproven at consumer-grade validity
Harmful?	Individual harm: − shame + misread beat the status quo of being unnoticed-but-uncharted	Indicators to track: opt-out requests, wellbeing-survey deltas, referral-rate shifts. Endurance: identity effects; remediability: archive = none; preventability: most harms preventable by design.
	Breadth of harm: − whole measured cohorts
	Certainty/preventability: −/+ harms likely but largely preventable
Fair?	Distribution: − unequal by design (misreads concentrate)	The burden falls on pupils least able to contest a score; unjustifiable as currently distributed.
	Less-advantaged harmed more: − yes (fit, literacy, neurotype, track)
	Burden justifiable? − no
Empowering?	Informed choice: − consent theatre	Autonomy could be justified only by preventing greater harm — the data shows the reverse. The pupil-brokered architecture is the route to +.
	Control over life: − the instrumented party controls least
	Coercion/manipulation: − streaks and confirmshaming are textbook

Synthesis: the gauge's story is a split verdict: a genuinely promising benefit (early warning, lesson evidence) sits on top of a fairness and empowerment failure that is architectural, not incidental. No spectrum in the fair/empowering dimensions scores above the dark pole in the current design — the benefit dimension cannot redeem that, and the synthesis says so plainly.

→ Carries forward to Tool 5

Current situation: Series B raised on a growth plan; 12 pilots; legislative tide arriving; validity gap published.
Hot spots: fair (all three spectra −), empowering (all three −), harm breadth.
Knowledge gaps: subgroup validity; green-theatre magnitude; parent comprehension.
Values priority: justice & autonomy > coverage & ARR; transparency > data story.

Tool 5 — Weighing Options two real options, then the Future Direction

Current situation: Fokus must choose its 2027 architecture now; the legislative tide (#1–7) makes the child-level line legally fragile in the EU; the room-level alternative exists internally (Bakkali's pilot) but has no revenue model yet.

Option A — “Scale as briefed.” Child-level line to 2,500 schools; keep D1–D6; lawyer-up. Societal: benefits scale with the misreads; archive risk multiplies by 200×. Organizational: ARR path intact; enforcement exposure (Art. 5, GDPR); the Rip-out scenario is priced at zero. Obstacles: DPO veto, school-board legal reviews, parent revolt; contingency: EU-only “lite” consent mode — which is Option B wearing a costume. Prioritising: growth, speed, coverage — over justice, autonomy, privacy.

Option B — “Room-first, consent-first.” Ship room-level as the flagship; pupil band only under medical/consent architecture; kill D2/D3; 30-day retention. Societal: keeps the real benefits, removes the misread/shame architecture; archive collapses to nothing. Organizational: ARR ceiling lower (€120M by 2036); new procurement category to create; clinical spinout option preserved (#20). Obstacles: investor narrative rewrite; contingency: bridge revenue from consent-first analytics + fairness-audit services. Prioritising: justice, autonomy, trust — over speed and the €220M line.

Future Direction: Option B — with the Ethics Gauge's hot spots (fair, empowering) as the acceptance criteria, and the knowledge gaps (subgroup validity, green theatre) as its research agenda. What we still need before the final board call: the parent-committee conversations Another Lens promised, and the room-level pilot's first-term data.

Closing synthesis: Values prioritised: justice, autonomy, dignity, trust (over coverage, ARR, speed). Trade-offs accepted: a lower revenue ceiling; a harder procurement story; the loss of the streak metric. Actions: pause child-level scale-up; ship room-level + consent redesign; board decides medical-vs-wellness Q1 2027; red lines (§10) fixed before Series C.

edbx-normative-design-schemeThree co-equal lenses — Virtue (intention) · Consequentialism (effect) · Deontology (duty) — assess mode

Step 0 · Stakeholder & power map (abridged — full map feeds all three lenses)

Stakeholder	Relationship	Power they have	Power they lack	What they'd lose
Pupils (primary users)	Instrumented subjects	Collective refusal; parent channel	Contract standing; data leverage; opt-out reality	Interior privacy; self-authored attention
Vulnerable: ADHD/autistic pupils; hijab-wearing pupils; low-literacy families	Misread first	Advocates, sometimes	Re-baselining; comprehension of consent	Fair interpretation; equal benefit
Non-users: classmates, siblings in other schools	Ranked beside, normalised into	Cultural refusal	Any data right	An unmeasured cohort norm
Workers: annotators, support	Label pupil data	Exit	End-use visibility	Clean conscience by design
The designers themselves	Build the gauge	Speech, exit, design choices	Cover over the metric	The professionals they become
Future-affected: 2040 cohort; subpoenaed archives	Inherit	None	Consent retroactively	The precedent of childhood-as-data

Power asymmetry: the operator holds the sensor, the schema, the retention clock and the contract; the pupil holds a streak. All three asymmetries are effectively irreversible once data exists — exit, undo and contest are all closed after 30 days. Non-obvious harms (5): inferred-state harms (mood/fatigue inferences nobody asked for); aggregation re-ID of “anonymised” archives; competence foreclosure (self-regulation scaffolded outside the child); intrinsic-motivation crowd-out (overjustification via streaks); future-you harm (track advice and behaviour files written from 12-year-old data).

Lens 1 · Virtue — intention

User-facing: the design trains performance of attention, not attention; patience, honesty and self-knowledge are what the streak undercuts. Creator impact (the skipped move): “What kind of engineers does the heatmap make us? Ones who read a child's worst week as a churn-risk. What habits of moral attention does the metric demand we suppress? The habit of asking what the number is. What would we have to believe to ship D3 in good conscience? That a percentile, weekly, from a company to a parent, is love. What are we embarrassed to explain to our own children? That we sold their attention to their school, and called it insight.” Ideas: 1) a “moral attention” review step in the design system — every metric change answers “what does this make us”; 2) ship the pupil-authored “attention story” feature the team would be proud to sign.

Lens 2 · Consequentialism — effect

Assessment: benefits are real but concentrated where resources already are; harms are real and concentrated where resources are not; at 2,500 schools the archive becomes systemically attractive (breach economics) and the normalisation effect irreversible. Ideas: 1) ship the room-level SKU — best consequence-per-euro once shame architecture is removed; 2) publish an annual “harms ledger” — consequence transparency as product feature.

Lens 3 · Deontology — duty

Assessment: duties owed: to pupils — truth about the number, exit without cost, no treatment as means to a data business; to parents — information without comparison; to teachers — support without surveillance of their craft; to the future — no archive it must forgive. Ideas: 1) a duty clause in every contract: “the pupil is never the product”; 2) data minimisation as categorical imperative — build only what deletion would destroy.

Universal law test

Design decision	If every product did this…	Livable?	Verdict
D1: live named-pupil state visible to authority figures in real time	Every classroom tool shows children's inner states live; childhood becomes a performance for every dashboard	No — attention becomes universal performance	Fails
D2/D3: rank minors weekly, in class and at home	Every youth product ranks children to motivate them	No — ranking becomes the pedagogy	Fails
D4: “anonymised” reuse of derived child data with buried opt-out	Every product buries exits under aspirational labels	No — consent stops meaning anything	Fails
Room-level analytics + pupil-owned signals + one-tap exit	Every lesson tool measures rooms, minors own their signals, exit is free	Yes — insight without extraction	Passes

Triad conflict matrix

Decision	Virtue	Consequence	Deontology	Alignment
Room-level pivot	⬆️ promotes honesty of craft	⬆️ net positive	⬆️ honors duty to pupils	🟢 All-aligned
Kill streaks/leaderboard (D2)	⬆️	⬆️ (extrinsic crowd-out removed)	⬆️	🟢 All-aligned
Consent redesign (D4)	⬆️	↔ mixed (short-term growth cost)	⬆️	🟡 Two-against-one — dilemma named honestly: the consequence lens genuinely loses near-term growth and school reach; the team accepts slower scaling as the price of consent meaning something, and this is a commitment, not a resolution.
“ADHD-adjacent insights” (D5)	⬇️	⬇️ (referral shadowing, misread amplification)	⬇️ (treats pupils as diagnostic means without evidence)	🔴 All-conflict — no ethical configuration of D5 exists on current evidence (#16); the team must decide to stop, not to phrase it better.

Strongest 🟢: the room-level pivot — all three lenses endorse it. Most serious 🔴: D5 — resolve by removal, not reframing.

Normative design statement: “This design is grounded in honesty of craft — promoting self-knowledge in pupils and moral attention in our team. It produces earlier support and better lessons for measured classrooms while imposing misread shame on the neurodivergent and tracked pupils it most misreads. It fulfills the duty of truthful instrumentation, but cannot fully reconcile its growth promise with free exit. We commit to the room-level pivot and the consent redesign to honor what the lenses agreed on, and to removing D5 entirely and naming the €220M ambition as deferred — not rephrased — to acknowledge what they did not.”

Priority recommendation: adopt the room-level pivot first — it benefits the named stakeholder who currently has no power (the pupil), and it eliminates the non-obvious harm with the longest half-life (competence foreclosure).

edbx-digital-ethics-compass“Put the Human in the Cosmos” — four categories · audit of Fokus Band 2 v2.3

Stakeholder & power map: §09's Step-0 map is imported by reference (six named populations, three asymmetries, all irreversible post-collection). Historical-justice prior: schools have a record of adopting measurement on the least powerful — from IQ-sorting's origins to attendance surveillance of migrant pupils; the burden of proof sits on Fokus to show this won't recur, not on children to prove it might.

Objective function risk table

Stated objective	Optimized objective (what metrics actually maximize)	Stakeholder value in conflict	Drift mechanism	Mitigation owner
“Help every learner be seen”	Attention-score coverage of lesson minutes; ARR; churn ↓	Pupil autonomy; honest interpretation; family trust	A/B loop rewards stickier dashboards and more instrumented minutes; coverage buys “insight” that is mostly presence-theatre	Product + Education Lead (joint)

Non-obvious harms inventory (5)

Harm	Population	Mechanism	Considered before?
Inferred-state leakage (fatigue, puberty-phase, home stress read from traces)	All pupils	Derived-data inferences beyond the score's purpose	No
Attentional-architecture formation: pupils habituate to being watched into focus	Adolescents	External scaffold replaces self-regulation	Partially
Fit-artifact proxy harm (hair/headwear → degraded score)	Hijab-wearing pupils; thick-hair pupils	Hardware physics becomes cultural proxy	No (pre-CIDER)
Pedagogy migration: schools teach to the measurable	All pupils; curriculum	Goodhart drift via procurement KPIs	Partially
Annotation-labour exposure: staff watch pupil footage for wages	Workers; pupils	Supply chain invisibility	No

The four categories (findings only; heuristics answered in full worksheet)

Category	Status	Key findings
1 · Data	🔴 High risk	24-month retention + reuse channel (D4/D6) disproportionate to a lesson tool; derived data escapes “neural data” definitions today (#7) — plan for parity anyway; deletion exists for accounts, not for archives; shutdown fate of the archive undefined.
2 · Avoid manipulation	🔴 High risk	Streaks (variable reward), confirmshaming, percentile framing — §07/§08's audit stands; nudges currently designed to exploit habit and loss aversion, not support it.
3 · Make intentions transparent	🟡 Medium	Users cannot see what the score can't do (no uncertainty display); business incentives (insurer pilot) invisible; personalisation not disclosed in Mentor copy; “anonymised” unverified.
4 · Automation	🔴 High risk	Mentor recommendations land on minors without challenge path; automated seating/pace suggestions unequally distributed; no explainability at pupil level; human oversight exists for teachers but not for pupils (the affected party).

Business pressure attribution & priorities

🔴 Finding	Business pressure behind it	Type
Retention/archive architecture	ARR narrative requires a data moat	Requires Leadership
Streak/leaderboard mechanics	Engagement proof for procurement decks	Requires Leadership
Three-menu opt-out	Retention team owns the exit	Quick Win
Mentor's unchallengeable advice	AI-demo velocity for Series C	Strategic Fix

Priority improvements: 1) Quick win — one-tap opt-out + default-off reuse (sprint 27.5); 2) Strategic fix — Mentor gains a pupil-side challenge path and an uncertainty display; 3) Requires leadership — replace coverage/ARR as the optimised objective with lesson-improvement and comprehension metrics; the board decides.

Compass reading: “This product puts children's autonomy and interpretive dignity at risk primarily through Data and Automation, driven by a coverage-and-ARR objective that rewards watched minutes over better lessons. The most urgent reorientation is to move the optimised objective itself — a leadership decision, not a feature fix.”

edbx-responsible-design-prismFive-axis spectrum: dark patterns → responsible design · audit + reflection

Prism audit

Axis	Score	Evidence
User relationship	Red	Pupil as data-source for a business-to-school market; the measured party holds no lever (means-to-an-end structure)
Information	Red	False expectations via “anonymised” label, uncertainty-free scores, care-framed percentiles
Error handling	Yellow	Misreads are quietly corrected for schools, never surfaced to pupils; no “the band was wrong” UX exists
Business alignment	Red	Revenue model monetises instrumented minutes and data reuse — exploitative incentive at the core, fair practices in the margins
Usability	Green	Genuinely good hardware and dashboard craft — which is what makes the other axes worth fixing

Mechanism audit (Red/Yellow axes): user relationship — default bias + friction asymmetry (easy entry, buried exit) · information — metric reification + false consensus (a coarse estimate dressed as fact, and “everyone's doing it” procurement framing) · error handling — silent repair (corrections happen above pupils' heads, denying them the learning that their state was misread) · business alignment — data-exhaust economics (the archive is the moat; the moat is the harm).

Reflection (synthesized)

Where we stand when we build: under Series B return math; the heatmap shipped because a competitor demo had one.
Org ecology: coverage as the north-star KPI makes every humane fix look like revenue arson; leadership rewards the deck that demos, not the one that deletes.
Transition: smallest meaningful shift per axis — exit symmetry (one tap); uncertainty display; “the band was wrong” state; room-level SKU.
Recovery: the shipped base can be recovered by deleting more than we add — retention 30 days, comparison removed, reuse default-off — and saying so publicly.

Redesign pathways

Axis	Redesign	Business case	Risk of inaction
User relationship	Pupil-brokered data architecture	The only posture that survives procurement under the 2031-style clause world of §06	Category ban; The Rip-out
Information	Uncertainty-first score UI; audited “anonymised” claims	Trust as differentiator as parents get literate	AI-Act / GDPR enforcement; reputational cascade
Error handling	“The band was wrong” surfaced to pupils & teachers	Calibration data improves the model — error honesty is free training data	Misread harm compounds silently
Business alignment	Room-level SKU priced per class; insurers excluded	New procurement category; defensible under every §06 future	The sorting room arrives by drift

Spectrum placement: transitional-leaning-dark — excellent craft carrying a dark business alignment. Leadership framing: “We sell to schools a product whose buyers can't use it legally in five years if a competitor's lawyer reads the news first. The redesign is cheaper than the enforcement action, and it wins the procurement cycles that matter: the ones that ask what happens to the data.”

edbx-value-dams-and-flowsStakeholder value conflicts — what's blocked, what flows · #alignmyteam

Dam-flow map

Feature / policy	D or F	Opposed by	Supported by	Underlying value	Strength	Intensity	Power to unblock
Live named-pupil heatmap (D1)	DAM	Pupils, most parents, DPO, unions	Sales; two pilot head-teachers	Mental integrity; dignity	🔴 Hard	5	Board — but it would cost them the Series C data story
Percentile parent report (D3)	DAM	Parents' committee, Bakkali	Growth team	Childhood without ranking	🔴 Hard	4	CEO — costs a launch feature
Insurer research reuse (D6)	DAM	DPO (legal), pupils, teachers	CFO, insurer partner	Non-commercial childhood data	🔴 Hard	5	Board + DPO jointly — legal veto is real
Streaks (D2)	DAM	Teachers, humane-design read	Part of growth; some pupils	Intrinsic motivation	🟡 Soft	3	CPO — dissolves if replaced by self-authored reflection
24-month retention (D4)	DAM	DPO, future cohorts	ML team	Data minimisation	🟡 Soft	4	CPO — dissolves at 30 days + on-device (model quality loss is the honest price)
On-device processing	FLOW	—	Everyone incl. engineers	Privacy by architecture	🟢 Strong	—	—
Teacher training & lesson analytics	FLOW	—	Teachers, schools, Bakkali	Teacher authority	🟢 Strong	—	—
Room-level pilot	FLOW	Growth team (margin scepticism)	Everyone else	Insight without extraction	🟡 Conditional	—	— (condition: dedicated budget line, not spare cycles)
University research partnership (strict consent)	FLOW	Some parents (case-by-case)	Universities, science team	Public knowledge	🔴 Contested	—	—

Hard-dam navigation (D1 as example): redesign alternative: room-level live view — the teacher keeps “eyes in the back of her head”, no child is named. Negotiation framing: “What does the heatmap give you that a room view and an end-of-lesson summary don't?” Minimum viable ethics: pupil-first visibility + no comparative aggregates. What would have to be true for D1 to dissolve as a dam: live named-pupil state would need to be the only clinically effective form of the intervention — which the evidence (#13–16) actively contradicts. Contested flow safeguard: research partnership proceeds only with per-family consent, external re-ID audits, and a school-level veto.

Values balance statement: “This design proceeds with on-device processing, teacher-first analytics and the room-level pilot, while blocking named live heatmaps, percentile reports and insurer reuse — protecting children's mental integrity and teachers' authority without asking the business to abandon insight.”

edbx-values-leversTeam-culture layer — run from Dr. Bakkali's seat (team lead, partial authority) · #identifyvalues #designresponsibility

Lever activation plan

Lever	Applicable?	How to activate	Owner
Interdisciplinary teams	Yes	Put the DPO, a paediatric advisor and the special-needs coordinator into roadmap reviews permanently (not as reviewers — as deciders on their domains)	CPO
Designing around constraints	Yes	Run a 60-minute spike: “redesign the parent report with zero pupil data in it”	Bakkali
Leader advocacy	Yes	Arm Donders with the Art. 5 analysis as a business risk brief, not an ethics memo	Donders
Self-testing	Yes	Every PM wears the band for one full school week, including the staffroom leaderboard	People lead
Gaining/funding	Partial	Price one ethics incident (The Rip-out) against the room pilot's cost in the board pack	CEO

Whose values are visible — and whose aren't

A · Served now: investors (growth, optionality) — stated: “evidence-based education”, unstated: “a defensible data moat”; schools (control, inspection-proofing) — stated: “support pupils”, unstated: “visible teachers”. B · Absent from the room (3+): pupils themselves — would want exit without cost and truth about the number; their absence is why the streak shipped. Future teachers of 2040 — would want a profession not organised around dashboards; absent, the pedagogy migrates. Substitute & special-needs staff — would want to know what colours mean before using them; absent, misreads happen on their watch. The 2040 cohort — would want the precedent debated; absent, it is set. C · Systemic map: aggregation — 30 pupils × a term re-identifies (proven in E2's mechanism); power — the operator holds schema, retention, and contract; reversal ~0 after 30 days; future discriminatory use — attention archives enable disability profiling of adults who were once pupils; non-user externalities — classmates ranked beside, siblings normalised into.

Communication audit & facilitation

Channel	Values reinforced	Space for ethics?	Recommendation
Sprint review	Velocity, demo-ability	No	Add a standing 10-min “who does this touch?” slot
Metrics standup	Coverage, ARR	No	Pair each KPI with its humane counter-metric
Pilot school calls	Pragmatism	Partial	Pupils join one call per term (they are the users)
New ritual: “Red-tea meeting” (monthly)	Candour	Yes	Anti-Heroes cards as shared vocabulary; no roadmap decisions may be cited without one “who loses” answer

30-day plan: Week 1 — Donders' Art. 5 business-risk brief lands with the CEO (deliverable: one-page risk memo). Week 2 — constraint spike “parent report with zero pupil data” (deliverable: two concepts). Week 3 — PM band-week completes; retro uses Anti-Heroes cards (deliverable: three named cards with cites). Week 4 — checkpoint: room pilot funded or not; success = a funded line, not a nod.

Values-based design decision statement: “In this project, we are choosing to prioritise exit without cost and honest interpretation over coverage and the data moat, because the users with least power are the ones most misread — and this will manifest in the pupil-brokered architecture, the one-tap exit, and the uncertainty-first score UI.”

edbx-ethicographyChuvaala, Gray et al. (2019; 2024) — team reflection mode · the 2026 roadmap meetings as data

Ethical moves inventory (from the roadmap-minutes archive)

#	Actor	Decision / speech act	Quadrant	Value impact
1	Verhoeven (CEO)	“We would never rank children” — while approving D2 (leaderboard) three weeks later	Implicit + Coercion	⬇️ honesty, trust
2	De Wit (CPO)	Live heatmap framed as “safety net for the quiet kids”	Implicit + Persuasion	↔ care ⬆️ / autonomy ⬇️
3	Bakkali (Learning science)	Proposed validity study for subgroup fairness; deferred to “post-Series C”	Explicit + Persuasion	⬆️ justice (deferred)
4	Donders (DPO)	Flagged D6 insurer reuse as “unlikely to survive Art. 5 review”; overruled on commercial grounds	Explicit + Persuasion (overruled)	⬇️ privacy, responsibility
5	Growth team	Opt-out moved from one menu to three to “reduce confusion”	Implicit + Coercion	⬇️ autonomy
6	Verhoeven (CEO)	Approved on-device processing for v2 despite cost	Explicit + Persuasion	⬆️ privacy (real, structural)

Performative ethics moves: Move #1 is the archive's cleanest specimen — ethical language (“we would never rank children”) performing cover for a ranking mechanic shipping in the same quarter; the privacy page (Move: “your data stays yours”) vs D4 architecture is its twin. Values alignment: privacy: 1 supporting (6), 2 undermining (4-overruled, 5) · honesty: 0 supporting, 2 undermining (1, privacy-page twin) · care: 1 supporting, 1 ambivalent (2) · justice: 1 deferred (3).

Affected populations (every move, named): #1 — pupils (ranked while told they aren't; voice: no) · teachers (professional honesty implicated; voice: partly) · #2 — quiet pupils (genuinely caught early; voice: no) · ADHD pupils (misread by same surface; voice: no) · #3 — neurodivergent pupils (fairness study deferred; voice: no) · #4 — all pupils in reuse channel (voice: no) · insurer customers-to-be (voice: no) · #5 — low-literacy families (exit priced in literacy; voice: no) · #6 — all future pupils (on-device default; voice: no, benefit anyway).

Ethical trajectory narrative: the archive shows a team that talks value-centred and ships implicitly coercive: every explicit ethical proposal (#3, #4) was deferred or overruled on commercial grounds, while every implicit coercion move (#1, #5) shipped without a meeting. The trajectory is not malicious — it is structural: the metric did the deciding.

Trajectory forward (12-month projection): “If current decision patterns continue, Fokus will drift toward implicit coercion as default: the Series C deck will inherit the coverage metric, D5's claims will creep from ‘insight' toward ‘screening', and the first explicit ethics proposal to survive will be one that's also good for growth. The inflection point is the January 2027 roadmap lock. The team can prevent this by giving the DPO a real veto (not an advisory flag) and by adopting the red lines in §10 before the deck is written.”

Recommendations: record ethical objections with the same rigour as legal ones; make one explicit-persuasion move ship per quarter (the trajectory needs wins); re-run this coding at the roadmap lock.

10 · THE ETHICS WALL — BAY 4 EDBX · WHAT DO WE PROMISE?

Findings are cheap.
These two methods make them expensive to break.

Everything upstream lands here as commitments with thresholds, owners, breach consequences and signatures: the worry clusters, the anti-goals, the red lines implied by the Universal Law Test.

BAY 4

What do we promise?

edbx-pledge-worksPledge Works — Simple Pledges · Responsible Stories · To Whom · red-line register · #alignmyteam #designresponsibility

Anchors imported: the 🔴 worry clusters (§08), the anti-goals (§08 Black Mirror), the 🟢/🔴 triad findings (§09), the prism pathways (§09).

Simple pledges

“We will keep raw EEG signals on-device and delete all derived scores after 30 days, to ensure no archive of named childhood exists — for pupils.” DEFAULT-CHALLENGING
“We will design every parent-facing surface without comparison to other children, to protect family trust and child dignity.”
“We will give every pupil one-tap exit and a paper equivalent, to ensure refusal costs nothing social.”

Responsible stories

“As the CPO, I pledge to block any release in which a named pupil's state is visible live to anyone but that pupil, because visibility without consent is the mechanism every other harm in this journal rides on.”

“As the learning-science team, we pledge to publish the score's validity bounds per subgroup with every school-year release, because an unqualified number is a misread waiting for a procurement deck.”

To-whom pledge

“As Fokus Labs, we pledge to the pupils of 2040: no decision we ship this year will make your interior measurable by anyone you didn't choose.”

Pledge register (with the five required fields per pledge)

Pledge	✅ Success indicator	⚠️ Failure signal	Owner	Consequence for breach	Cadence	Early warning
On-device, 30-day, nothing named	0 raw-signal egress events; 100% derived rows <30 days at quarterly audit	Any named raw row in storage; any egress to third parties	CTO	Release freezes; incident logged publicly in the quarterly responsibility report	Quarterly audit + pre-release gate	Any ticket requesting “just a longer window for research”
No comparison in family surfaces	0 percentile/rank elements in QA sweep; parent comprehension ≥70% (n=20/term)	Any comparative aggregate rendered for a parent	Head of Learning Science	Feature rolls back within one sprint	Per release + termly parent test	Copy requests for “where does my child stand”
One-tap exit + paper card	Opt-out ≤2 taps ≤60s in mystery-shop; card honoured ≤1 lesson in 100% of tested classes	Any new confirmation step added to exit	Mobile platform lead	Ship-blocked until fixed	Twice-termly mystery-shop	Any “we should show them what they'll miss” proposal
CPO release veto (visibility)	0 named-pupil live views in shipped code (CI test)	CI test disabled or skipped once	CPO + CI owner jointly	Release halts; board notified within 24h	Every release	Any request to “temporarily” re-enable for a demo
Validity bounds published	Bounds doc in every termly school pack; DI ratio ≥0.8 per subgroup (n≥30)	Any school-year release without bounds; any DI ratio <0.8	Head of Learning Science + DPO	Rollout gates shut	Pre-launch + 30/90/180 days	“Marketing needs the number without the caveat”

Stress test (for the default-challenging pledge): “Q3 growth is at risk; a big network asks for a live heatmap as a condition of signing.” The pledge leadership would be tempted to waive is the CPO veto (demo exemption). Protective mechanism: the CI test cannot be disabled by product — only by a signed board resolution logged in the responsibility report. Making the waiver harder than the sale is the point.

Red-line register — what we refuse to build

Refusal (feature-level)	Why it crosses the line	Owner with veto	Consequence if attempted
No live named-pupil attention display for any audience	The mechanism of the shame circuit; fails the Universal Law Test; AI-Act-adjacent	CPO — CI-enforced	Release halts; board notified in 24h
No ranking of minors by measured attention, in any surface, including “friendly” leagues	Dator-Growth-only bet; E1's engine; fails universalisation	Head of Learning Science — product-council veto	Feature removed within one sprint; incident logged
No sale, licence or share of pupil-derived data with insurers or any commercial third party	E2's entire mechanism; GDPR Art. 8/7(3) exposure; the sorting room's door	CEO + DPO joint veto	Contract voided; public disclosure in responsibility report
No diagnostic or screening claims from consumer-grade signals (D5 dies)	Triad 🔴 with no ethical configuration; #16 evidence; MDR exposure	General Counsel + Head of Learning Science joint veto	Feature shipped = product withdrawn from schools pending review

Pledge-to-product traceability

Pledge / red line	Feature affected	Decision it produces	Decision it prevents
On-device / 30-day	Data architecture	Session-scoped schema; auto-delete job; egress monitor	24-month archive; cross-cohort model training on named data
No comparison in family surfaces	Parent app	Observation-first report with two offers	Percentile numerals; “compared to classmates” modules
One-tap exit	Pupil app + school ops	Home-screen exit; paper card; aggregate-shrink	Multi-step flows; “speak to your teacher” gates
No named live display	Teacher dashboard	Room-level live view; end-of-lesson named summary to the pupil first	Live pupil tiles; photo pulses; behaviour-file exports
No screening claims	Mentor	Referral-suggestion wording with clinician loop	“ADHD-adjacent insights”; risk scores in school reports

edbx-ethical-contractSigned shared agreement — five steps + bias & harm audit · #identifyvalues #alignmyteam

1 · Ethical challenge

Fokus instruments the bodies of minors for a commercial data product while the affected party holds no lever; without a signed cross-disciplinary contract, commercial pressure will keep resolving every dilemma the same way (§09's Ethicography projection). The cost of not addressing it is the Rip-out scenario: a category ban and a company that cannot explain its own archive.

1.5 · Stakeholder tension map

Growth needs a data moat and engagement proof → DPO/Legal needs minimisation and lawful basis → structurally opposed because the moat is the exposure.
Schools need inspection-proof visibility → pupils need exit without social cost → opposed because the dashboard's value to adults is the pupil's loss of interior.
Engineers' job is to build what's specified → ethics requires pushing back on specifications → opposed because the refusal has no slot in the sprint.

2 · Responsibilities & red lines by role

Stakeholder	Role	Ethical responsibility	Red line (never, under any pressure)
M. Verhoeven	CEO	Protect the contract in board and investor rooms	Will not sign any contract monetising pupil-derived data
J. de Wit	CPO	Keep exit symmetric with entry in every release	Will not approve a live named-pupil view
A. Bakkali	Head of Learning Science	Only ship claims the evidence supports, with bounds	Will not ship any screening or diagnosis claim
L. Donders	DPO / Legal	Flag rights implications before build, not before launch	Will not sign off processing that fails Art. 5 / Art. 8 review
Eng & data leads	Development	Implement only what the contract authorises; keep CI guards	Will not disable a CI ethics guard without a board resolution
Sanne (teacher-partner) + pupil council	Stakeholder representatives	Surface classroom reality that dashboards hide	Will not endorse any feature they wouldn't defend in their own staffroom

3 · Shared ethical objectives (each with all four parts)

#	Commitment	Quantitative threshold (breach trigger)	Named owner	Review cadence
1	“We commit to on-device processing and 30-day derived-data life in order to make the worst breach an empty folder.”	0 raw egress events; 100% derived rows <30 days at audit; storage schema lint green	CTO	Quarterly audit + every release gate
2	“We commit to comparison-free family and classroom surfaces in order to protect children from ranking.”	0 comparative aggregates rendering for named minors; comprehension test ≥70%	Head of Learning Science	Every release + termly parent test
3	“We commit to fairness testing across neurotypes, headwear and literacy in order that the score misreads no group systematically.”	Disparate-impact ratio ≥0.8 per subgroup (n≥30); validity bounds published each school year	Head of Learning Science + DPO	Pre-launch + 30/90/180 days post-term

4 · Bias & harm audit (Section 4 of the signed contract)

Population	Specific harm	Severity	Reversibility
ADHD/autistic pupils	Systematic misread; identity as “red kid”	High	Partly (identity effects persist)
Hijab-wearing / thick-hair pupils	Degraded or absent scoring → exclusion from benefit, presence in ranking	High	Yes if fixed early
Low-literacy families	Consent theatre; reports unread or confrontational	Medium	Yes
All pupils	Archive exposure (breach/subpoena/commercial reuse)	Catastrophic (worst case)	No
Teachers	Accountability without authority; staffroom ranking	Medium	Yes
2040 cohort	Normalised measured childhood	High (systemic)	No

Bias audit thresholds: disparate-impact halt at ratio <0.8 for any subgroup with n≥30 · named proxies to monitor: electrode-contact quality (hair/headwear), motion features (body-regulation practices), device/storage assumptions (income) · model-confidence floor: recommendations below 0.7 escalate to a human who talks to the pupil before anything changes.

Gray-zone register: anonymised academic research on pupil data — allowed only with per-family consent + external re-ID audit + school veto (would need all three to expand) · teacher-facing “risk lists” — not built; would require a court order, a published policy, and a pupil-side challenge path that does not exist · insurer wellness product — not built; would require a different company.

5 · Design goal & signatures

“Our ultimate design goal for Fokus is better lessons and earlier support for every pupil, while no child is ranked, no interior is archived, and no exit costs anything.”

Name	Role	Date	Signature
Marit Verhoeven	CEO	30 Sep 2026	____________
Jonas de Wit	CPO	30 Sep 2026	____________
Dr. Amira Bakkali	Head of Learning Science	30 Sep 2026	____________
Lotte Donders	DPO / Legal	30 Sep 2026	____________
Sanne	Teacher partner, De Brug	30 Sep 2026	____________
Pupil council, De Brug	Stakeholder representatives	30 Sep 2026	____________

Renewal triggers: any Series C raise · any new school-network contract · any ethics incident · any change to the AI Act enforcement landscape · each new school year (thresholds re-baselined).

Commitment risk register

Objective	Likely failure moment	Early warning signal	Protection mechanism
On-device / 30-day	Series C data-room due diligence (“show us the moat”)	Request for “longitudinal retention” slides	Board resolution required; investor letter includes the red lines
Comparison-free surfaces	Big-network procurement asks for league tables	“Just for the staffroom version”	Contract §: no comparative features, product-council veto
Fairness testing	Launch deadline vs missing subgroup n	“Ship first, test after the holidays”	Release gate is automated on the DI ratio

11 · WALKING HOME FDBX · BACKCASTING

Stand in the preferred 2040.
Remember how we got there.

The UK Futures Toolkit's backcasting, with Elise Boulding's move (via Inayatullah): work backwards from the end state, score every critical event for control, name who loses out, then take the same walk in avoid mode through the feared future, to know what to block. Overcoming present bias is the method's whole point.

fdbx-backcastingGO-Science Futures Toolkit (2024) pp. 91–96 · Inayatullah (2008) p. 18 · mode: preferred, then avoid · backcasters: the Fokus strategy group + Sanne + pupil council

The preferred future in 2040 — “the gardener's console”

Dimension	In 2040
What people can do	Any teacher can see how a lesson breathed — pacing, talk-ratio, room energy — within minutes of the bell, without any child being measured. Any pupil can keep their own attention signal, on their own device, and share what they choose with whom they choose. Any parent can get honest, comparison-free observations about their child's school weeks — and two concrete offers when support would help.
Conditions that are true	Room-level analytics are a normal procurement line (EU model clause §4.2 exists); pupil-owned signal wallets interoperate across bands; Fokus runs a clinical arm under medical consent for the children who need it; raw pupil signals are retained by nobody, including Fokus.
What is no longer true	No live named-pupil colours anywhere; no percentile reports; no streaks or class leaderboards; no 24-month archives; no “anonymised” commercial reuse of child-derived data; no screening claims from consumer-grade EEG.

Key differences, now and then

PESTLE	Now (2026)	In 2040	Driver of the change
Political	Voluntary UNESCO recommendation; phone-ban policy energy S	Procurement law embeds mental integrity; subtraction first	Regulators + parent movements
Economic	Per-pupil SaaS; data-moat logic; 14% CAGR capital S	Room-level SKUs; clinical reimbursement; audit services	Procurement categories created deliberately
Societal	Attention anxiety high; surveillance fatigue growing S	Measured childhood abnormal; signal wallets personal	A decade of parents who read the Focus Files reporting
Technological	Consumer EEG coarse; on-device compute feasible S	Interoperable learner-owned signals; room sensing mature	On-device trajectory + open standards
Legislative	AI Act Art. 5 in force; US neural-data states; derived-data gap open S	Derived-data parity closed; child sensing medical-only	The 2027–2031 legislative window
Environmental	Unmodelled device footprint A	Per-room mode shrinks device count 30×; take-back norm	The environmental audit nobody did in 2026

Memories from 2040

“It is 2040. Looking back, the turning point was the consent redesign of 2027 — the quarter we deleted more than we shipped. Everything after got easier because the archive never existed.” — Jonas de Wit, retired CPO

“The turning point for me was when the room pilot beat the bands on outcomes. After that, keeping the heatmaps was like arguing for chalk dust.” — Dr. Amira Bakkali

“We went on head strike in '33 — bands off at the bell, whole school. I used to think we were protesting technology. We weren't. We were insisting on being asked.” — Milan, 28, electrical engineer

“The clause — EU model procurement §4.2 — did more than any product. It made ‘cannot be re-attributed to any pupil' a line item, like fire exits.” — Lotte Donders, former DPO, now at the Dutch DPA

Backward timeline (read from the top: each row had to happen before the row above)

2040

Preferred future in place: rooms measured, minds sovereign, clinical arm consented.

end state

2036–39

Pupil-owned signal wallets interoperate across ≥3 band makers (open spec, learner-controlled egress).

critical · control 2

2034–36

EU model procurement clause adopted by ≥500 school boards; child-level sensing restricted to medical consent.

critical · control 3

2033–34

The Rip-out scare (near-miss, not the full leak) makes clause adoption politically irresistible.

control 3

2031–33

Fokus's clinical arm achieves reimbursable indication (EndeavorRx-style pathway) — sensing IP finds its lawful home.

critical · control 2

2029–31

Room-level pilot evidence published: lesson-design outcomes beat child-scoring in matched schools.

critical · control 1

2028–29

Fairness study (n≥30/subgroup) either validates bounded scoring or kills child-level claims — either way, the claims wall holds.

critical · control 1

2027–28

Consent redesign ships: one-tap exit, default-off reuse, percentile-free reports, D2/D5 removed; red lines signed by board.

critical · control 1

2026–27

Board chooses Option B (§09): room-level flagship; coverage metric retired; ethics contract signed.

critical · control 1

Now

This journal: 27 methods, one decision due at the January 2027 roadmap lock.

Critical events in our control (scored 1–2)

Event	Control	Impact	Who benefits / loses out	Certainty	Enablers	Barriers	Steps now
Board chooses Option B (2026–27)	1	Everything downstream depends on it	Pupils, teachers benefit · Series-C story loses its moat narrative (investors may feel loss without losing value)	Med-high	Ethicography projection; Art. 5 exposure; Donders' brief	€220M anchor; competitor demo pressure	Option B memo to board before roadmap lock; red lines pre-signed
Consent redesign ships (2027–28)	1	Converts the decision into product reality	Pupils & low-literacy families benefit · growth team loses a lever	High	CIDER/IBM redesigns already specced	Sprint capacity; “engagement proof” habit	Specs 27.4–27.6 into the January plan
Fairness study runs (2028–29)	1	Decides whether any child-level claim survives	Neurodivergent pupils benefit either way · marketing loses a claim	High	Pilot population; university partners	Recruitment; uncomfortable result risk	Fund in Q1 2027; preregister the analysis
Room pilot beats bands (2029–31)	2	Makes the pivot undeniable to buyers	Schools benefit · no one materially loses (bands redeploy to clinical)	Med	12-school trust; Bakkali's design	No procurement category yet; margin scepticism	Matched-school design with an external evaluator
Clinical arm reimbursed (2031–33)	2	Gives sensing IP a lawful, consented home	Children with clinical need benefit · consumer roadmap loses a feature	Med	#20 precedent; MDR consultants	Evidence burden; capital patience	Scoping study with a university hospital in 2027

Critical events outside our control (scored 3 — kept on the timeline)

Event	Who has control	If it doesn't happen	What we can influence	Who loses out, or may feel they will
EU model clause adopted at scale (2034–36)	Legislators; school boards	Pocket solutions persist; grey market of parent-bought bands grows	Help draft the clause; volunteer pilots under the strictest reading; publish compliance economics	Vendors relying on the derived-data gap; schools that bought child-level systems — they may feel stranded, and they will lobby
Rip-out scare, not catastrophe (2033–34)	The whole industry; luck	Either no scare (complacency slows clause adoption) or full leak (Fokus's clinical arm catches blame by association)	Industry standards body; incident-response pacts; be the company that couldn't leak what it never kept	All vendors feel exposed; insurers may feel an opportunity — watch them
Signal-wallet interoperability (2036–39)	Standards bodies; platform rivals (Meta-class)	Wallets stay proprietary; transformation future arrives gated by platforms	Publish the spec early; open-source the reference client; ally with teacher colleges	Fokus's services margin; competitors' lock-in — they will feel it as loss

Forward check

A forward timeline written independently (Q3 pipeline, procurement cycles, model roadmap) meets the backcast at the same decision points: January 2027 (Option B), Q3 2027 (fairness study funding), 2029 (room pilot readout). Where the forward plan was more optimistic — it assumed the clause lands without a scare — the backcast's version wins, because present bias loves a quiet legislator.

Action plan (type: action learning — small experiments create the future; not a budgeted certainty)

Action	Owner	Start	Decision point it serves
Option B board memo + pre-signed red lines	Verhoeven + Donders	Nov 2026	Jan 2027 roadmap lock
Ship consent redesign specs 27.4–27.6	De Wit	Jan 2027	2027–28 consent redesign
Preregistered fairness study	Bakkali	Q1 2027	2028–29 claims wall
Room-level matched-school pilot w/ external evaluator	Bakkali + Sanne	Sep 2027	2029–31 readout
Clinical scoping with university hospital	Bakkali + GC	Q3 2027	2031–33 indication
Open signal-wallet spec v0 + reference client	CTO office	2028	2036–39 interoperability
Model procurement clause drafting input	Donders	2027–33, ongoing	2034–36 adoption

Unknowns to fill (not invented)

What share of pilot parents read the Focus Report at all today? (asks the analytics team)
What does one band cost per lesson-minute over its life? (asks the hardware team + a lifecycle study)
What is the pilot schools' current opt-out rate, and why? (asks De Brug's office)
Which EU body would host the model clause, and what evidence would they need? (asks Donders)

Avoid mode — walking back from “The Rip-out” to know what to block feared future, same method

Feared end state, 2033: the Focus Files leak; provinces cancel contracts; category credibility gone for a decade. Memories backward: “the archive existed for four years before anyone leaked it”; “the broker licence was signed in 2031 because a quarter needed the revenue”; “we knew re-identification worked — the lab told us in 2029 and we shipped anyway.”

Step that leads to the feared future	Control	Block with
Named archive accumulates past the pedagogical need (2027→)	1	30-day derived life; on-device raw; retention lint — pledge 1, signed
“Anonymised” reuse channel opens (2028–31)	1	Red line: no commercial third-party reuse — CEO+DPO joint veto, signed
External re-ID testing skipped for speed (2029)	1	Quarterly lab audit is a release gate, not a research project
Brokers court the category; a competitor says yes first (2030–31)	3	Can't control rivals — can make “never sold” a marketable public commitment, shifting buyer defaults
Regulator arrives after the fact (2033)	3	Voluntary early compliance reporting — be the case study of the clause, not its corpse

Who loses out if we block these steps: nobody who matters to the product — the revenue the channel promised was the Rip-out priced as ARR. The avoid-mode action plan is already inside the preferred-mode plan; that overlap is the strongest signal this journal can give the board.

Claim used as input	Type	Basis
Legislative window & procurement-clause direction	SOURCED	Art. 5 · unesco.org · HB24-1058
Clinical pathway precedent exists	SOURCED	FDA De Novo DEN200026 (EndeavorRx)
Validity gap constrains claims to coarse, bounded use	SOURCED	Mikhaylov 2024 · Marchand 2025
All 2040 conditions, timeline events, memories	ASSUMPTION	Constructed end state per method (fictional, labelled)

Handoff

Assumptions surfaced: the preferred future assumes the board can over-ride the Series C narrative — untested; the clause world assumes no full-scale leak arrives first — luck is a dependency.

Open decisions: the January 2027 roadmap lock is the one date everything serves.

Suggested next method: edbx-pledge-works and edbx-ethical-contract (already run in §10) — the action plan's owners and thresholds are now commitments with signatures. The chain is closed.

12 · THE RECEIPTS EVIDENCE · PROVENANCE · WHAT CHANGED

What twenty-seven methods
actually did to one roadmap

A methods journal owes you an audit of itself. Here is the consolidated evidence ledger, the provenance of every technique, and the list of decisions each technique changed, which answers “were the skills any good?”

Methods run (6 fdbx + 21 edbx)

Sourced signals · 0 invented

6 / 6

Briefed decisions materially changed

Red lines with veto owners

Roadmap ambition honestly retired

The decisions the methods changed

D1 · Live named heatmap — killed

Four independent methods converged (Humane 🔴 shame circuit, IBM 🔴 convergence, Universal Law Test fail, DAH Unethical). Replaced by room-level live view + pupil-first end-of-lesson summary. CI-enforced red line.

D2 · Streaks & leaderboard — killed

Worrystorming's crowd-out worry became IBM's cascade, became the Triad's 🔴, became a red line. Replaced with pupil-authored reflection. Product-council veto.

D3 · Percentile parent report — redesigned out of existence

Another Lens' named bias (“the parent I am”), CIDER's gate scenario, Motivation Matrix's 🔴 column, humane heuristics. Replaced by observation-first reports with two offers.

D4 · 24-month retention & buried reuse — inverted

Fair Patterns' legal mapping (GDPR 7(3), Art. 25) + Compass data category + E2's mechanism. Now: on-device, 30-day, default-off, CI lint.

D5 · “ADHD-adjacent insights” — removed entirely

Normative Design Scheme's only 🔴 row: no ethical configuration exists on evidence #16. Referral wording + clinician loop instead. Joint GC+science veto.

D6 · Insurer research reuse — dead

Black Mirror's Episode 2 gave it a story; the contract gave it a veto. Nothing named pupil-derived ever leaves for commercial reuse.

Method receipts — the sharpest finding of each, and what it changed

Method	Sharpest finding	Decision it changed
fdbx-horizon-scanning	The legislative tide is ahead of the product's legal design — and Environmental PESTLE is empty (reported as a gap, not filled)	Compliance became a Q1 2027 board decision, not a launch chore
fdbx-causal-layered-analysis	The brief's deepest move is “attention is a pupil property”; the reframed brief measures rooms, not children	The unit of measurement — the pivot everything else serves
fdbx-futures-triangle	The strategy serves a used future (borrowed growth playbook); the disowned future (“sorting room”) is the one being built	Naming the unchosen push made the drift preventable
fdbx-three-horizons	Six of seven current initiatives are H1 maintenance or H2− — most resources prop up the pattern losing fit	Room-level pilot moved from unfunded idea to funded line
fdbx-four-futures	The roadmap contains exactly one future (Measured Decade) and is fully exposed to the Rip-out	Five robust decisions identified; single-future bets flagged for the board
fdbx-backcasting	The avoid-mode and preferred-mode plans overlap almost entirely — the safest path is also the preferred one	Action plan with owners; January 2027 named as the decision point
edbx-cider	“The neutral attentive body” excludes Amina — and exclusion reappears socially even after the interface is fixed	Inclusive fit kit + paper opt-out card + aggregate-shrink
edbx-humane-design-guide	Two exploitation stacks (shame circuit, extraction stack) — one architectural move breaks both	Pupil-brokered data ownership
edbx-anotherlens	The CPO designed for the parent he is; the person he's nervous to face is the parent-committee chair	Three-row design decision spec with numbers and owners
edbx-critical-interviewing	Covert category: do teachers privately treat the score as theatre? — plus discoverability discipline for the research itself	Interview protocol run before procurement scale-up
edbx-worrystorming	The product is an archive of childhood told it is a coach	Three 🔴 clusters became constraints with test questions
edbx-dah-cards	Card 4 in Sanne's voice ended the “is it even surveillance” debate	Ship-ready ticket: replace streaks, Sprint 27.5
edbx-bad-design-canvas	Cell 3 (unfair control): the instrumented party controls nothing — systemic, not cosmetic	“Rethink fundamentally” verdict on the child-level line
edbx-black-mirror-brainstorming	Eight anti-goals, three load-bearing; E2 showed the harm can be entirely legal	Anti-goals became CI tests and contract clauses
edbx-inverted-behavior-model	Two prompts are 🔴 converged with the worst possible design — not adjacent, already there	Killed P2 named rank; moved P3 to end-of-lesson
edbx-motivation-matrix	Every 🔴 cell lands in the same context: the stressed, low-literacy home — motivational load is regressive	Context-conditional design rules
edbx-fair-patterns	Two patterns are 🔴 in ≥2 jurisdictions — must-fix before next release, mechanically	Default-off reuse + one-page consent, Sprint 27.5
edbx-anti-heroes	“Two-Faced”: the privacy page and the data schema disagree	Promise = pipe (shared source of truth)
edbx-stf-et	The Gauge's synthesis: real benefit sitting on architectural fairness/empowerment failure; Options weighed, B chosen with acceptance criteria	Option B + its knowledge-gap research agenda
edbx-normative-design-scheme	D5 is 🔴 in all three lenses — no ethical configuration exists	D5 removed, not rephrased
edbx-digital-ethics-compass	Stated ≠ optimised objective: coverage rewards watched minutes; the fix is a metric change only leadership can make	North-star metric question put to the board
edbx-responsible-design-prism	Transitional-leaning-dark: excellent craft carrying a dark business alignment	Leadership framing that survives a board room
edbx-value-dams-and-flows	D1 is a Hard Dam (intensity 5/5) that no facilitation can dissolve — only a redesign honours it	Dam-flow map used as the stakeholder meeting agenda
edbx-values-levers	Pupils are the absent stakeholder whose absence explains the streak	Standing “who does this touch?” slot; PM band-week
edbx-ethicography	Every explicit ethics proposal was deferred or overruled; every implicit coercion move shipped without a meeting	DPO veto made real; trajectory forward dated to the roadmap lock
edbx-pledge-works	Pledges without failure signals are aspirations — five fields or nothing	Register + stress test + early warnings
edbx-ethical-contract	Tension map first, signatures last — a contract that ignores real conflicts is performative	Signed contract, thresholds, renewal triggers

Why the chain works — the router's argument, demonstrated

The two routers (fdbx §4 “cross-box handoffs” and edbx §3 “chaining patterns”) claim that methods are worth more in sequence than alone. This journal is the test, and the receipts agree:

Scan → everything. Because the scan came first with real links, CLA's systems layer, the triangle's pushes and the backcast's uncontrollable events are all sourced claims instead of confident guesses. The Evidence Ledger never had to invent a trend.
CLA → scenarios. The four futures differ in worldview (measurement vs. integrity vs. transformation vs. trust-collapse) because CLA surfaced the discourses first — Dator's rule “differ fundamentally” was achievable, not aspirational.
Triangle → Three Horizons → Backcasting. The disowned future named in §04 became the avoid-mode backcast in §11; the H2+ pocket funded in §05 became the 2029–31 timeline event. Nothing in the final plan is unowned.
fdbx → edbx handoffs. Each cross-box handoff the fdbx router prescribes was used: four futures → black-mirror and worrystorming; robust decisions → pledge-works; Three Horizons dilemma → value-dams; backcast plan → ethical-contract. The ethics wall audited the futures, and the futures gave the wall its scenarios.
edbx chaining patterns. Worry → Contract (§08→§10), Lens → Spec (§07), Forecast → Redesign (§08→§07), Decision deadlock → Three-lens (§09) — all four recognized patterns occurred naturally in sequence.

Evidence ledger, consolidated

Claim typing per the skills' shared rule: SOURCED = a link from this session's searches that a reader can open · USER-SUPPLIED = stated in the brief · ASSUMPTION = model-supplied, never counted as evidence. Every per-method ledger in §02–§11 follows it; the counts:

Type	Rows across all ledgers	Notes
SOURCED	38 rows · 24 distinct source links	UNESCO ×2, EU AI Act, Colorado bill, KFF, FPF, Stanford Law, EdSurge, China Project, Quartz, Reuters, BBC, APA, Gloria Mark, WHO Europe (HBSC), Zeng, Mikhaylov, Kleeva, Marchand, Precedence, OpenPR, Meta, CNET, FDA. Every link opens the cited page itself, not a site homepage; all 20 signals were re-checked against their sources on 30 Sep 2026.
USER-SUPPLIED	0	The brief itself is case construction (fictional), so its facts are typed ASSUMPTION — the honest choice for a composite case.
ASSUMPTION	All Fokus figures, people, initiatives, and every fictional future artifact	Always labelled at point of use. Leads to verify listed in §02; unknowns listed in §11 rather than invented.

Method provenance

fdbx method	Published technique it runs	Source
horizon-scanning	GO-Science Horizon Scanning (5 steps, PESTLE, natural agenda, 2×2 rating)	Futures Toolkit, 2nd ed. (2024), pp. 34–41
causal-layered-analysis	Inayatullah's CLA (four layers, toolbox, move back up)	Futures 30(8) (1998); Six Pillars (2008)
futures-triangle	Inayatullah's Futures Triangle (pull/push/weight, archetypes, used & disowned futures)	Six Pillars (2008), pp. 5–8, 15
four-futures	Dator's four generic futures (Manoa School), rules, A–E exercise	J. Futures Studies 14(2) (2009)
three-horizons	Sharpe/Hodgson Three Horizons practice (H2±, pockets, dilemma, perspectives)	Ecology & Society 21(2):47 (2016); Curry & Hodgson (2008)
backcasting	GO-Science backcasting (control scoring, winners/losers) + Boulding's move	Futures Toolkit (2024), pp. 91–96; Inayatullah (2008), p. 18

edbx method	Published technique it runs	Source
cider	CIDER elicitation (Critique-Imagine-Design-Expand-Repeat)	Oleson et al. (2022)
humane-design-guide	Center for Humane Technology six-sensitivity worksheet	humanetech.com; UMED #47
anotherlens	Another Lens (balance bias · opposite · growth mindset)	UMED #07
critical-interviewing	Carspecken critical ethnography interview protocol	Carspecken (1996), adapted
worrystorming	Worrystorming (8 clusters → values reframe)	UMED #100
dah-cards	DAH card pairs + ethical sorting	Design Against Humanity
bad-design-canvas	Manos' 12-category Bad Design Canvas (CC BY-ND)	Manos, gum.co/baddesign; UMED #09
black-mirror-brainstorming	Speculative episode red-team → anti-goals	Klassen & Fiesler (2022); Mauldin (2018)
inverted-behavior-model	Fogg B=MAP inverted for audit (worst design, convergence, cascades)	BJ Fogg model, audit adaptation
motivation-matrix	Motivation Matrix (5 motivations × contexts)	Zhu (2020); UMED
fair-patterns	Dark-pattern taxonomy (Gray et al. 2024) + fair countermeasures	Gray et al. (2024); deceptive.design
anti-heroes	Anti-Hero/Hero card deck (evaluation mode)	Mehta, Chivukula, Gray et al. (2024), arXiv:2405.03674
stf-et	Stanford Ethics Toolkit five-tool chain	McCoy Family Center for Ethics in Society (2025), CC BY 4.0
normative-design-scheme	Three normative lenses + Universal Law Test + Triad matrix	Normative Design Scheme
digital-ethics-compass	Four-category compass heuristics (Data · Manipulation · Transparency · Automation)	The Digital Ethics Compass
responsible-design-prism	Five-axis dark↔responsible spectrum + mechanism audit	Responsible Design Prism; dark-pattern taxonomy
value-dams-and-flows	Stakeholder dams/flows + conflict intensity	Value Dams and Flows
values-levers	ENABLE model, five levers, values-surfacing	Values Levers
ethicography	2×2 ethical-move coding + trajectory forward	Chuvaala, Gray et al. (2019; 2024)
pledge-works	Pledges / Responsible Stories / red-line register	Pledge Works
ethical-contract	Five-step signed agreement + bias & harm audit	Ethical Contract

The honest disclaimer, one last time. A clean ledger and a complete chain prove the methods were followed — not that practitioners will find the output useful, and not that the future will cooperate. What this journal can claim is narrower and stronger: every external claim is checkable, every fictional number is labelled, every method ran with its own required artifacts and guardrails, and the roadmap that entered this process in September 2026 is not the roadmap that left it.

“The methods didn't predict 2040. They made the 2027 roadmap harder to lie to.”

Fokus 2040The attention company

A company, a headband,and a brief with one future in it

The company

The people in the room

What is already changingaround this product

Natural agenda — what the scans cluster into

Drivers of change to take forward

What to watch next

Leads to verify — not signals, not counted

The brief is a headline.What's the deep story under it?

Worldviews in play — each constitutes a different problem

Root metaphor and alternative

Moving back up — the alternative metaphor rebuilt

Scenarios by layer — four kinds, not four degrees

Design implications

What pulls, what pushes,and what weighs

Pull — images of the future people actually hold

Push — drivers already moving (all sourced or typed)

Weight — barriers, specific to each image

Interaction

Scenarios from the triangle

Design implications

What's losing fit, what's emerging,and which way each experiment points

The map

H1 — losing fit (and the fixes that fail)

H3 — aspirations and pockets of the future in the present

H2 — innovations in play, and which way each points

Features to maintain from H1

Dilemma and power

Horizon perspectives

Actions

Portfolio classification — Fokus's current initiatives

Experience the alternativesbefore choosing one

The Measured Decade

Excerpt, Regionaal College Utrecht-Oost annual report, 2040

The Rip-out

NRC headline, 14 March 2033

The Neuro-Rights Classroom

EU model procurement clause, 2035

The Symbiotic School

Pupil-facing card, “mySynapse”, school year 2040–41

Across the four futures

Toward a preferred future (a sketch, not a conclusion)

Before we ask what it does,ask who it reads

Stage C — Assumptions surfaced

Stage I — Exclusion scenario

Stage D — Redesign proposals

Attention

Emotional

Decision-making

Algorithm audit (required — the product scores and recommends)

Engagement-value paradox

Exploitation stacks

Memorable heuristics (written fresh from this audit)

Lens 1 — Balance your bias

Lens 2 — Consider the opposite

Lens 3 — Embrace a growth mindset

Design decision spec (top 3)

Eight ways to worry well

Worry list (16, all eight clusters covered)

Clusters → reframed values

DAH card pairs

Consequence map (selected rows; likelihood · severity · who bears it · design awareness)

Sorting verdict

Microcopy rewrites

What could go wrong (10, across three lenses)

Anti-goals (testable, traced)

From worries to mechanics

Stakeholder map (Step 0)

Prompt inventory

Convergence check

Behavioral risk summary (top 3)

Redesign directions

Dark pattern audit

Countermeasures, defaults & copy rewrites

Implementation sequence (ranked: Legal Risk × User Impact ÷ Effort)

1 · Frame

2 · Anti-Hero hits

3 · Hero counter-moves

4 · First three changes to ship

Fokus 2040
The attention company

A company, a headband,
and a brief with one future in it

What is already changing
around this product

The brief is a headline.
What's the deep story under it?

What pulls, what pushes,
and what weighs

What's losing fit, what's emerging,
and which way each experiment points

Experience the alternatives
before choosing one

Before we ask what it does,
ask who it reads

Seven ways to decide
with the worries on the table

Findings are cheap.
These two methods make them expensive to break.

Stand in the preferred 2040.
Remember how we got there.

What twenty-seven methods
actually did to one roadmap