{"schema_version":1,"research_id":"eoa_inverse_innovation_exp06_external_evaluation_20260803","source_assessment_id":"catalytic_pathway_enablement__human_computer_interaction:P3:v0","cell_id":"catalytic_pathway_enablement__human_computer_interaction","search_queries":["site:w3.org WAI personalization semantics user preferences accessibility settings specification","GPII auto personalization preferences across devices applications accessibility research","Cloud4all automatic personalization accessibility user preferences applications study","cross application accessibility preferences users repeated settings configuration research","site:raisingthefloor.org GPII auto personalization preferences applications official","site:github.com GPII preferences framework documentation auto personalization","ISO IEC 24751 AccessForAll user preferences accessibility standard official","European Commission Cloud4all auto-personalisation user preferences project results","Cloud4all evaluation users automatic personalization trial results paper","GPII automatic personalization evaluation user study preferences settings across devices paper","\"auto-personalization\" accessibility evaluation Cloud4all study","accessibility personalization users configure settings repeated devices qualitative study","Morphic accessibility preferences apply settings any computer official","MorphicPlus auto personalize computers settings official documentation","site:morphic.org accessibility settings personalization computers","site:raisingthefloor.org Morphic settings computers personalization","\"Creatures of habit\" personalization accessibility settings DOI","\"An Open-source Tool for Simplifying Computer\" Morphic DOI","site:eur-lex.europa.eu GDPR Article 9 disability data consent Article 25 data protection by design official","site:ico.org.uk disability data special category data accessibility preferences GDPR"],"sources":[{"source_id":"S1","title":"“Creatures of habit”: influential factors to the adoption of computer personalization and accessibility settings","publisher":"PubMed / U.S. National Library of Medicine","url":"https://pubmed.ncbi.nlm.nih.gov/37361673/","source_class":"PRIMARY_RESEARCH","publication_date":"2023-03-04","accessed_at":"2026-08-03","claims_supported":["A grounded-theory study used 49 interviews with 15 participants with and without disabilities.","Accessibility and ease-of-use settings can improve experience, but adoption remains poor and personalization is affected by multiple barriers.","The study directly examined built-in settings, assistive technology, and Morphic-based personalization."]},{"source_id":"S2","title":"Visual design and accessibility settings: an exploratory study on their role in technology adoption among older adults","publisher":"Frontiers in Communication","url":"https://www.frontiersin.org/journals/communication/articles/10.3389/fcomm.2026.1808537/full","source_class":"PRIMARY_RESEARCH","publication_date":"2026-03-26","accessed_at":"2026-08-03","claims_supported":["In two focus groups totaling 16 participants, many older adults struggled to independently configure accessibility settings.","Access paths varied across devices, labels were confusing, and participants often depended on other people for configuration and recovery.","System-wide text enlargement often failed to affect third-party applications, requiring per-application configuration.","Accessibility changes sometimes damaged layout or usability, supporting the need for preview, validation, and rollback, while limiting generalization because the study was small and population-specific."]},{"source_id":"S3","title":"Cloud platforms Lead to Open and Universal access for people with Disabilities and for All (CLOUD4all) project fact sheet","publisher":"European Commission CORDIS","url":"https://cordis.europa.eu/project/id/289016","source_class":"OFFICIAL_ORGANIZATION_DATA","publication_date":"2017-04-25","accessed_at":"2026-08-03","claims_supported":["The European Commission funded a multi-sector consortium to create preference profiles, interface descriptions, matching mechanisms, and automatic on-demand modification across operating systems, browsers, web applications, mobile devices, kiosks, televisions, and smart homes.","The project explicitly sought to avoid repeated negotiation, explanation, qualification, and justification of user requirements.","CLOUD4all ran from 2011 to 2015, had total recorded cost of €12,980,843 and EU contribution of €7,582,661, establishing strong historical funder pull and a large-program cost anchor.","The project is very close prior art to cross-application compilation of functional preferences into supported settings."]},{"source_id":"S4","title":"Requirements for WAI-Adapt specification","publisher":"World Wide Web Consortium","url":"https://www.w3.org/TR/adapt-requirements/","source_class":"OFFICIAL_GUIDANCE","publication_date":"2022-06-09","accessed_at":"2026-08-03","claims_supported":["WAI-Adapt defines requirements and use cases for personalization semantics.","It anticipates user agents, extensions, assistive technology, or scripts using preferences to manipulate content for individual users.","It recognizes conflicting user needs, confusion caused by unfamiliar designs, application-level notification control, and the need for reversible personalization, but leaves user-agent implementation out of scope."]},{"source_id":"S5","title":"ISO/IEC 24751-3:2008 — Individualized adaptability and accessibility in e-learning, education and training — Part 3: Access for all digital resource description","publisher":"International Organization for Standardization","url":"https://www.iso.org/standard/43604.html","source_class":"STANDARD","publication_date":"2008-10; confirmed 2026-03-26","accessed_at":"2026-08-03","claims_supported":["The standard supplies an information model and common vocabulary for matching individual needs and preferences to user interfaces and digital resources.","It addresses interoperability, consistent implementation, defaults, bindings, conformance roles, and implementation scenarios.","The standard was confirmed current in March 2026, showing that standardized preference-to-capability matching is established prior art."]},{"source_id":"S6","title":"Morphic makes computers easier to use","publisher":"Morphic / Raising the Floor","url":"https://morphic.org/","source_class":"OFFICIAL_PRODUCT_DOCUMENTATION","publication_date":"undated","accessed_at":"2026-08-03","claims_supported":["Morphic is a currently offered application that exposes built-in accessibility features and supports use of assistive technology across computers.","Morphic states that the product and organizational use are free.","A University of Michigan assistive-technology manager reports institutional benefits, identifying a concrete adopter and practitioner pull for portable accessibility support."]},{"source_id":"S7","title":"An Open-source Tool for Simplifying Computer and Assistive Technology Use: Tool for simplification and auto-personalization of computers and assistive technologies","publisher":"ACM ASSETS; archived by PubMed Central","url":"https://pmc.ncbi.nlm.nih.gov/articles/PMC8620129/","source_class":"PRIMARY_RESEARCH","publication_date":"2021-10","accessed_at":"2026-08-03","claims_supported":["Morphic already saves accessibility and usability settings, applies them on another computer, and supports assistive technology on shared computers.","The implementation demonstrates technical feasibility for authenticated preference storage, settings capture, portable application, and restoration.","Its scope centers substantially on operating-system and assistive-technology settings rather than the proposal's full bundle of versioned per-application contracts, per-change preview, transactional confirmation, postcondition validation, and transient-binding deletion."]},{"source_id":"S8","title":"Regulation (EU) 2016/679 (General Data Protection Regulation), consolidated text","publisher":"EUR-Lex / European Union","url":"https://eur-lex.europa.eu/eli/reg/2016/679","source_class":"GOVERNMENT_OR_REGULATOR","publication_date":"2016-05-04","accessed_at":"2026-08-03","claims_supported":["A functional accommodation profile is personal data when linked to an identifiable user and may be health data when it reveals physical or mental health status.","Article 9 generally prohibits processing health data unless a condition such as explicit consent for specified purposes applies.","Articles 5 and 25 require purpose limitation, data minimization, storage limitation, and data protection by design and default.","Article 35 can require a data-protection impact assessment where new-technology processing is likely to create high risk, especially for large-scale special-category data."]}],"problem_evidence":{"support":"STRONG","rationale":"Two independent primary studies visibly document poor discoverability, complex configuration, variable access paths, reliance on helpers, and third-party applications failing to honor system-wide settings. The consequences include frustration, dependency, and unusable layouts. Evidence is nevertheless qualitative and small-sample, so population prevalence and aggregate burden remain unknown.","source_ids":["S1","S2"]},"stakeholder_evidence":{"support":"STRONG","rationale":"The European Commission committed €7.58 million to a consortium pursuing nearly the same auto-personalization objective, and a named University of Michigan practitioner reports benefits from Morphic. These identify funder, institutional adopter, and application/accessibility stakeholders, although they demonstrate pull for the broader established approach rather than for this proposal's exact contract-and-validation bundle.","source_ids":["S3","S6"]},"prior_art":{"proximity":"SUBSTANTIAL_COLLISION","closest_analogues":[{"name":"CLOUD4all / Global Public Inclusive Infrastructure","similarity":"It used user profiles, interface descriptions, matching, and automatic on-demand modification across operating systems, browsers, web applications, mobile devices, kiosks, and other platforms—the proposal's central intent-to-supported-setting transformation.","remaining_difference":"The reviewed fact sheet does not establish the proposal's exact combination of contemporaneous per-change preview and confirmation, versioned application contracts with protected invariants, transactional rollback, post-application reachability and persistence checks, and mandatory transient-binding deletion.","source_ids":["S3"]},{"name":"Morphic auto-personalization","similarity":"It is an implemented, currently available system that saves and reapplies accessibility/usability and assistive-technology settings across computers.","remaining_difference":"Evidence reviewed emphasizes operating-system and assistive-technology portability, not validated compilation into heterogeneous per-application setting schemas with application-owner contracts and per-session postcondition testing.","source_ids":["S6","S7"]},{"name":"ISO/IEC 24751 AccessForAll and WAI-Adapt","similarity":"They standardize or specify machine-readable needs, preferences, interface/resource descriptions, and user-agent-driven personalization.","remaining_difference":"They provide vocabulary, semantics, and requirements rather than evidence that the proposal's complete transactional execution, validation, reset, and governance workflow improves user outcomes.","source_ids":["S4","S5"]}],"distinctive_claim_remaining":"For a fixed set of reversible settings in two sandbox applications, adding versioned application contracts, a previewable diff with contemporaneous user confirmation, transactional rollback, reachability/persistence validation, and transient-binding deletion to preference portability will reduce median elapsed time and user actions to a user-confirmed state by at least 25% versus the better of manual native configuration and platform/Morphic-like portable settings, while keeping unintended or incomplete configurations at or below 5% and producing zero protected-control changes, cross-user residue events, or rollback failures.","confidence":"HIGH"},"implementation_evidence":{"support":"MODERATE","rationale":"CLOUD4all and Morphic demonstrate that preference capture, settings matching, automatic application, portability, and restoration are technically feasible, while ISO and W3C provide semantic foundations. The exact proposal still depends on application-specific reversible setting APIs, trustworthy versioned mappings, identity and consent integration, validation that is itself accessible, and sustained application-owner maintenance. GDPR creates manageable but material controller, lawful-basis, minimization, security, and possible DPIA obligations.","source_ids":["S3","S4","S5","S7","S8"]},"scores":{"meaningful_impact":{"score":3,"rationale":"The documented burden affects autonomy and access, but no representative prevalence, time-loss, or outcome estimate supports a higher score.","source_ids":["S1","S2"]},"stakeholder_pull":{"score":4,"rationale":"Substantial public funding, a 29-participant consortium, a deployed product, and an institutional testimonial demonstrate strong historical and current pull for this problem class.","source_ids":["S3","S6"]},"incremental_advantage":{"score":2,"rationale":"Preview, transactional rollback, validation, and ephemeral bindings could improve safety and trust, but no comparative evidence shows benefit beyond mature auto-personalization approaches.","source_ids":["S3","S7"]},"distinctiveness_plausibility":{"score":2,"rationale":"CLOUD4all, Morphic, AccessForAll, and WAI-Adapt occupy most of the claimed territory; only a narrow safety-and-validation bundle remains contrastive.","source_ids":["S3","S4","S5","S7"]},"technical_implementability":{"score":4,"rationale":"Core mechanisms have been implemented, although heterogeneous application contracts and validation increase integration and maintenance risk.","source_ids":["S3","S5","S7"]},"adoption_authority_feasibility":{"score":3,"rationale":"A user can authorize a local sandbox transaction and an accessibility-platform owner can run a bounded pilot, but scaled deployment requires separate authorization and maintenance commitments from every application owner plus privacy and security approval.","source_ids":["S3","S6","S8"]},"evidence_readiness":{"score":3,"rationale":"The candidate supplies a bounded comparator-based probe, but needs executable mappings, preregistered thresholds, accessible consent, controller assignments, and independent correctness adjudication.","source_ids":["S1","S2","S8"]},"safety_net_benefit":{"score":4,"rationale":"If effective, the intervention would reduce repeated configuration and helper dependence for users facing access barriers while preserving a manual route; the magnitude is not yet quantified.","source_ids":["S1","S2"]},"scalability":{"score":2,"rationale":"Compiler reuse is scalable in principle, but each application/version needs a trusted reversible interface, testing, refresh, and exception support; CLOUD4all's large budget illustrates ecosystem-level resource demands.","source_ids":["S3","S7"]}},"score_confidence":"MODERATE","costs":{"first_evidence":{"band_2026_usd":"50K_TO_250K","scope":"Preregister, build or harden six to ten reversible mappings for two sandbox applications, conduct privacy/security and accessibility review, recruit and compensate at most 12 participants, run the crossover comparison, and analyze correctness, effort, rollback, and residue.","confidence":"MODERATE","assumptions":["A reusable prototype and identity test harness already exist or can be assembled without production integration.","Two application owners provide sandbox setting interfaces and mapping review.","The trial uses non-sensitive tasks and local or test identities."],"source_ids":["S3","S7","S8"]},"initial_deployment_startup":{"band_2026_usd":"250K_TO_1M","scope":"Productionize an enterprise-limited service for roughly five applications, including contract tooling, encrypted profile storage, consent and rollback UX, automated conformance tests, security assessment, DPIA where required, monitoring, documentation, and support preparation.","confidence":"LOW","assumptions":["Applications expose stable reversible setting APIs.","Existing enterprise identity and hosting controls can be reused.","No medical diagnosis data or security-sensitive settings are collected."],"source_ids":["S3","S7","S8"]},"operational_launch":{"band_2026_usd":"1M_TO_5M","scope":"Launch a governed multi-application service across one large organization, with accessibility, platform, privacy, security, application-maintainer, incident-response, user-support, and evaluation capacity.","confidence":"LOW","assumptions":["Scope is tens rather than hundreds of applications.","Application teams accept mapping ownership and service-level obligations.","The launch does not include rebuilding inaccessible applications or a global standards ecosystem."],"source_ids":["S3","S6","S8"]},"annual_recurring":{"band_2026_usd":"250K_TO_1M","scope":"Maintain mappings and test suites, respond to application changes and incidents, operate encrypted profile and audit infrastructure, perform accessibility/privacy/security reviews, support users and exceptions, and measure rollbacks and off-target effects.","confidence":"LOW","assumptions":["Five to twenty applications with moderate release frequency.","Two to six resource-equivalent staff plus infrastructure and external review.","No large-scale assistive-technology licensing or remediation work is included."],"source_ids":["S3","S7","S8"]}},"verified_pipeline_gates":{"externally_supported_problem":{"status":"YES","reason":"Independent qualitative studies directly observe configuration difficulty, inconsistent access paths, helper dependence, third-party non-propagation, and adverse usability effects.","source_ids":["S1","S2"]},"externally_credible_adopter_or_authorizer":{"status":"YES","reason":"The European Commission previously funded a large consortium for this objective, and the University of Michigan is identified as an organizational beneficiary of a close product analogue.","source_ids":["S3","S6"]},"distinct_testable_incremental_claim":{"status":"YES","reason":"After removing the collided preference-portability claim, the contract/preview/transaction/validation/reset bundle can be compared against manual configuration and existing portable-settings approaches using prespecified benefit and safety thresholds.","source_ids":["S3","S4","S5","S7"]},"bounded_next_evidence_step":{"status":"YES","reason":"A randomized crossover sandbox study with at most 12 participants, two applications, a fixed reversible setting set, two comparators, measurable outcomes, and hard safety stops is bounded and falsifiable.","source_ids":["S1","S2","S7"]},"no_unresolved_safety_or_authority_stop":{"status":"UNCERTAIN","reason":"The proposed local pilot has user confirmation, rollback, manual alternatives, and application-owner review, but the actual controllers, lawful basis, DPIA determination, threat model, accessible consent implementation, and application-owner authorizations have not been evidenced.","source_ids":["S2","S8"]},"credible_cost_scope_and_range":{"status":"YES","reason":"The bands are scoped as resource equivalents and anchored by an implemented open-source analogue and the much broader €12.98 million CLOUD4all program, but production estimates remain low-confidence because application count and integration maturity are unspecified.","source_ids":["S3","S7"]}},"next_evidence_step":"After obtaining written authorization from two application owners and completing a privacy/security review, preregister and run a randomized within-participant crossover with at most 12 consenting participants, two non-sensitive sandbox applications, and six to ten reversible presentation/navigation settings. Compare (A) native manual configuration, (B) platform-level or Morphic-like portable settings, and (C) the proposed versioned-contract compiler. Hold application capabilities, intent, tasks, and support availability constant. Measure elapsed time, user actions, specialist minutes, user edits/declines, mapping conflicts, incomplete and unintended changes, task reachability, persistence, rollback success, cross-user residue after reset, prohibited-data events, and maintenance work over repeated invocations. Falsify the incremental claim if condition C fails to improve median time and actions by at least 25% over the better comparator, requires bespoke specialist work proportional to sessions, exceeds 5% unintended/incomplete configurations, changes any protected control, leaks a profile, leaves any cross-user residue, or has any failed rollback. Generalize only to the tested participants, applications, settings, and compiler version.","blocking_evidence":["No representative estimate establishes population prevalence, aggregate configuration time, specialist burden, or downstream exclusion.","No live comparison tests the narrow remaining claim against both manual setup and an existing portable-settings approach.","No evidence identifies the actual data controller and processor roles, lawful basis, DPIA determination, retention schedule, or completed security threat model.","No signed application-owner commitment demonstrates access to reversible setting interfaces or long-term contract maintenance.","No measured evidence establishes mapping refresh cost, failure rate after application updates, exception workload, or scalable support capacity.","The 2026 cost bands are resource-equivalent estimates, not vendor quotes or a bottom-up estimate from an approved architecture."],"research_disposition":"PARTNERED_RESEARCH_PROGRAM","world_novelty_boundary":"This eight-source open-web evaluation finds substantial collision with CLOUD4all/GPII, Morphic, ISO/IEC 24751 AccessForAll, and WAI-Adapt. It does not establish exhaustive world novelty, patentability, freedom to operate, market size, or realized impact. The only remaining boundary evaluated is the contrastive performance and safety of the explicit versioned-contract, preview, transactional rollback, postcondition-validation, and transient-reset bundle against close existing approaches.","arm":"COMPLETE_PROPOSAL_PORTFOLIO","candidate_version":0,"controller_recommendation":{"action":"STOP_EMPIRICAL_RESEARCH_NEEDED","repairable":false,"material_progress_observed":true,"progress_targets":["Secure written sandbox authorization and mapping ownership from two application owners.","Document controller/processor roles, lawful basis, purpose and retention limits, DPIA determination, threat model, incident handling, and accessible consent/withdrawal procedures.","Implement and independently verify versioned contracts, protected-setting exclusions, transactional rollback, reachability/persistence checks, and cross-user reset tests.","Preregister the three-condition crossover, 25% benefit threshold, 5% unintended/incomplete ceiling, zero-tolerance safety events, exclusion rules, and analysis plan.","Run the at-most-12-participant study and report all configuration attempts, declines, exceptions, repairs, rollbacks, residue tests, and application-update failures.","Produce a bottom-up production and annual-maintenance cost estimate tied to application count, release frequency, staffing, security controls, and support workload."],"reason":"Bounded web research verifies the problem, adopter/funder interest, technical plausibility, and substantial prior-art collision, leaving a narrow incremental claim whose truth requires proprietary application integrations and a live matched user test. Under the required decision rule, that evidence need is an empirical-research stop."},"proposal_index":3}