# From sealed research to the AppStore2031 Top 10s **Forecast target:** 31 July 2031 **Evidence cutoff:** 2 August 2026 **Edition:** `2031-2026-08-02` **Status:** research draft; not frozen or deployed This is the shortest honest account of how the visible AppStore2031 charts were made. It separates evidence about the world today from scenarios about 2031 and from the fictional products used to make those scenarios understandable. Every app, developer, icon, rating, review and exact rank is invented. The sources do not predict those products. They support or challenge the present conditions and pressures from which the forecast was inferred. ## What was researched before any product was ranked The replacement forecast did not begin with today's App Store categories or with a list of existing apps. It moved through these stages: 1. A global horizon scan covering AI and compute; work, firms and finance; health and demography; relationships and culture; energy, climate, food and water; physical systems and robotics; and geopolitics and public infrastructure. 2. Six contrasting 2031 worlds. They are conditional scenarios, not six predictions with assigned probabilities. 3. Sixty changed actors and 74 recurring institutional or lived needs. 4. Eight possible marketplace categories. Six passed into product authoring; two remained watch categories rather than being forced into the store. 5. Seventy-two new product concepts: twelve for each active category. Authors were not given the withdrawn inventory or a current-product catalogue. 6. A collision review to check that each concept was materially different from the other eleven concepts in its category. The exact source and claim registers, world documents, need documents, category proposals, criticism and receipts are available in the Research library. The candidate set was sealed before final ranking so that names and chart positions could not quietly rewrite the underlying concepts. ## Why the attempted present-day audit was stopped We attempted a stricter second audit in which independent researchers would compare every sealed concept with current products and published concepts. The audit machinery became too large: one category packet was 360–410 KB and the agent interface repeatedly truncated it after the first concept. Earlier runs also showed that an unusable opened source and context compaction could make an otherwise honest session impossible to validate. No result from those interrupted sessions was accepted. We did not restart the worlds, needs, categories or 72 concepts. We also did not build another audit protocol after the transport limit became clear. Continuing to optimise proof machinery would have created more apparent rigour without more knowledge about 2031. This is a real limitation. The final Top 10s have a bounded exact-name web screen, strong future-distance tests and public counter-cases, but not a complete claim that no present-day product overlaps any part of a concept. ## How the final Top 10 in each category was chosen Six separate category authors compared all twelve sealed concepts in one category. They selected ten and documented why two were excluded. Each concept was scored from 1 to 5 on six questions: | Question | Weight | What it means | | --- | ---: | --- | | Future distance | 5 | Does it depend on a credible 2031 structural change, rather than being a normal 2026 app with better AI? | | Need scale | 4 | Would the need recur often enough, for enough people or organisations, to support a major marketplace listing? | | Global breadth | 3 | Is it useful across several of the nine geographic lenses without pretending those places are the same? | | Marketplace clarity | 3 | Can someone call, buy or switch this service, and can they tell when it has finished? | | Delivery readiness in 2031 | 3 | Could the technical, physical, legal and institutional dependencies plausibly exist by 2031? | | Trust and safety | 2 | Can authority remain with the right people, and can harms, failure and exit be bounded? | The maximum weighted score is 100. The score is an authored judgement, not a measured probability. Ties were broken by reading the product's complete causal case: its recurring need, future dependence, regional variation, failure mode and strongest counter-case. Every row explains why it occupies that position and why it sits above the next concept. Rank 10 is compared with the strongest excluded concept. Regional views then apply small, transparent adjustments from the candidate's own nine-lens fit and world fit. They can reorder the same selected ten; they do not claim that an official local chart has been observed. ## How the fictional listings were written The visible store copy was written for an interested 16-year-old: - the first line promises a result, not a technology; - “What it does” explains the trigger, steps and finished outcome; - “Why it is on the list” connects a present condition to a world shift and a recurring 2031 need; - “Why this is not a 2026 product” names what is structurally missing today; - “How it could be built” names the smallest complete system and its technical, physical, financial, legal and human dependencies; - harms, abuse, stop behaviour and the strongest counter-case remain visible; and - two imagined reviews add texture but are labelled fictional beside the review itself. No technology is mandatory across the inventory. A listing mentions AI, digital currency, cryptography, a distributed ledger, identity infrastructure, robotics, energy or another rail only when that dependency performs a necessary job. Adding “blockchain” or “AGI” as decoration does not make a product future-native. ## Name screening Every selected fictional product name received a fresh, quoted web search with app, software and company terms. Names that materially collided with an app, software product, technology company or closely related digital service were replaced and searched again. The result and most relevant URLs are preserved in the edition's name-audit records. All 60 final fictional names finished with the bounded verdict `no-material-collision-found`. This is only a bounded exact-name screen. It is not trademark, company-name, domain, linguistic, cultural or legal clearance. A real entrepreneur must run proper clearance before using any name. ## What would change the forecast The forecast should move when evidence changes, not simply because three months have passed. Examples include: - dependable autonomous work expands much faster or slower than the scenario range; - governments create or reject enforceable human-review, portability, repair, local-capacity or infrastructure duties; - physical bottlenecks in power, water, cooling, materials or skilled labour materially change; - new institutions make a proposed marketplace category unnecessary; - a current product already performs the complete future-native job at scale; or - harm, exclusion or liability makes a concept illegitimate. The refresh skill creates a new dated edition. It must not silently edit this edition or treat today's six categories, 60 selected listings, nine lenses or their ranks as quotas for the next forecast. ## Remaining limitations - Scenarios are coherent possibilities, not probabilities of 2031. - AI authors share model pretraining and unavoidable project context; workflow separation reduces anchoring but cannot create a blank mind. - The source register is broad, but access, language and institutional evidence remain uneven across regions. - The final score is transparent and contestable, not statistically calibrated against future app-store outcomes. - Exact ranks create a useful object for debate; they are more precise than the evidence warrants and therefore include rank ranges, move conditions and counter-cases. - The stopped present-day overlap audit means readers should treat “new” as a reasoned design claim, not a proof of global novelty. That incompleteness is part of the educational value. AppStore2031 is designed to let a reader disagree with the chain—not merely admire a futuristic list.