{"schema_version":1,"research_id":"eoa_inverse_innovation_exp06_external_evaluation_20260803","source_assessment_id":"representation_independent_interface_contract__human_computer_interaction:P1:v0","cell_id":"representation_independent_interface_contract__human_computer_interaction","search_queries":["site:w3.org/TR multimodal architecture interfaces recommendation interaction manager data model","site:w3.org/TR scxml state chart XML recommendation dialog control","site:gov.uk/service-manual joined up experience all channels service standard","cross modality user interface model based task state conformance research","site:dl.acm.org cross-modal interface consistency voice graphical user interface accessibility study","site:pmc.ncbi.nlm.nih.gov voice interface screen reader web forms usability errors study","MARIA model based language interactive applications abstract user interface paper PDF Paterno","Cameleon Reference Framework multi-target user interfaces task abstract user interface primary paper","site:w3.org/TR/WCAG22 consistent help error identification status messages recommendation","site:rfc-editor.org RFC HTTP idempotent methods semantics retry request automatically","site:bls.gov software developers median pay 2025 occupational outlook web developers digital designers","site:iso.org ISO 9241-210 human-centred design interactive systems 2019","\"The Accessibility and Usability of Online Job Applications for Screen Reader Users\" journal","site:pubmed.ncbi.nlm.nih.gov \"online job applications\" screen reader users accessibility usability"],"sources":[{"source_id":"S1","title":"Provide a joined up experience across all channels","publisher":"UK Government Digital Service, GOV.UK Service Manual","url":"https://www.gov.uk/service-manual/service-standard/point-3-join-up-across-channels","source_class":"OFFICIAL_GUIDANCE","publication_date":"2025-12-16","accessed_at":"2026-08-03","claims_supported":["An identifiable public-service authority expressly requires services to meet user needs across needed channels.","The guidance says disconnected channels can exclude users, create inferior experiences, workarounds, and failure demand.","Service teams are expected to test online and offline channels together and involve operations staff in research and prioritization."]},{"source_id":"S2","title":"Web Content Accessibility Guidelines (WCAG) 2.2","publisher":"World Wide Web Consortium","url":"https://www.w3.org/TR/WCAG22/","source_class":"STANDARD","publication_date":"2024-12-12","accessed_at":"2026-08-03","claims_supported":["Interactive submissions require accessible error identification and, depending on conformance level and transaction type, reversibility, checking, or confirmation.","Compatibility with assistive technologies and complete-process conformance are established accessibility requirements.","A shared task contract could encode some WCAG-relevant obligations, but conformance to the proposed contract would not establish WCAG conformance or usability."]},{"source_id":"S3","title":"Multimodal Architecture and Interfaces","publisher":"World Wide Web Consortium","url":"https://www.w3.org/TR/mmi-arch/","source_class":"STANDARD","publication_date":"2012-10-25","accessed_at":"2026-08-03","claims_supported":["W3C already specifies an Interaction Manager separated from black-box modality components.","The architecture places shared application flow and data synchronization in the Interaction Manager while modality components retain private representations.","Public-data modification and backend submission are recommended to occur through the Interaction Manager, closely matching the proposed opaque task-state boundary."]},{"source_id":"S4","title":"State Chart XML (SCXML): State Machine Notation for Control Abstraction","publisher":"World Wide Web Consortium","url":"https://www.w3.org/TR/scxml/","source_class":"STANDARD","publication_date":"2015-09-01","accessed_at":"2026-08-03","claims_supported":["A standardized event-driven state-machine execution model already supports states, transitions, guards, data, cancellation, invocation, and external communication.","SCXML provides a technically credible representation for the proposed abstract report-session state and dialogue-independent transition logic.","SCXML has explicit conformance provisions and was developed partly for voice-browser control."]},{"source_id":"S5","title":"MARIA: A Universal, Declarative, Multiple Abstraction-Level Language for Service-Oriented Applications in Ubiquitous Environments","publisher":"ACM Transactions on Computer-Human Interaction; author copy hosted by ISTI-CNR","url":"https://giove.isti.cnr.it/attachments/publications/2009-A0-055.pdf","source_class":"PRIMARY_RESEARCH","publication_date":"2009-11","accessed_at":"2026-08-03","claims_supported":["MARIA already models multidevice interfaces at an abstract level independent of platform, modality, and implementation language.","Its abstract model includes data, presentations, enabled events, dialogue behavior, effects, and conditional transitions.","The proposal's core representation-independent interaction-state idea therefore has substantial prior-art overlap; the remaining difference is its mutation-tested substitutability oracle and maintenance-report-specific laws."]},{"source_id":"S6","title":"The Accessibility and Usability of Online Job Applications for Screen Reader Users","publisher":"Journal of Visual Impairment & Blindness; indexed by the U.S. National Library of Medicine","url":"https://pubmed.ncbi.nlm.nih.gov/38529353/","source_class":"PRIMARY_RESEARCH","publication_date":"2023-12-07","accessed_at":"2026-08-03","claims_supported":["Three blind expert testers attempted 90 application trials across 30 randomly selected Fortune 500 companies.","Only 50 of 90 attempts succeeded, 23 of 30 sites had a critical issue for at least one tester, and accessibility varied even among applications from the same vendors.","The study directly supports the existence and consequence of modality- and implementation-dependent failures in complex reporting-like forms, but it does not establish prevalence in maintenance reporting."]},{"source_id":"S7","title":"RFC 9110: HTTP Semantics","publisher":"Internet Engineering Task Force, RFC Editor","url":"https://www.rfc-editor.org/rfc/rfc9110.html","source_class":"STANDARD","publication_date":"2022-06","accessed_at":"2026-08-03","claims_supported":["Idempotent request semantics are established and permit retry after uncertain communication failure without changing the intended effect.","Clients should not automatically retry non-idempotent requests without independent knowledge that the semantics are idempotent or that the original request was not applied.","The proposed intent-token invariant is technically plausible but must be enforced by the service, not merely by interface adapters."]},{"source_id":"S8","title":"National employment and wage data by occupation, May 2025","publisher":"U.S. Bureau of Labor Statistics","url":"https://www.bls.gov/news.release/ocwage.t01.htm","source_class":"OFFICIAL_ORGANIZATION_DATA","publication_date":"2026-05","accessed_at":"2026-08-03","claims_supported":["May 2025 national mean hourly wages were $71.20 for software developers, $53.60 for software QA analysts and testers, and $56.49 for web and digital interface designers.","These wages provide an official labor anchor for resource-equivalent cost ranges after adding benefits, overhead, accessibility expertise, and 2026 rounding.","The source does not provide project-specific effort, contractor premiums, organizational overhead, or integration costs."]}],"problem_evidence":{"support":"MODERATE","rationale":"The problem class is visible and consequential: a primary study found critical screen-reader barriers on 23 of 30 complex application sites and only 50 successful attempts out of 90, while GOV.UK explicitly warns that disconnected channels can exclude users and generate workarounds and failure demand. WCAG establishes error, confirmation, and assistive-technology obligations relevant to the proposed state laws. However, no source measures desktop, touch, voice, and screen-reader state divergence in a field-maintenance reporting organization, so prevalence and effect size in the stated setting remain unverified.","source_ids":["S1","S2","S6"]},"stakeholder_evidence":{"support":"MODERATE","rationale":"The UK Government Digital Service is an identifiable institutional authorizer with an expressed requirement for joined-up multichannel services and cross-channel testing. WCAG supplies a standards-based accessibility rationale. No named field-maintenance organization, service owner, funder, accessibility representative, or technician has expressed demand for this specific contract or committed authority and resources, limiting transfer from the public-service analogue.","source_ids":["S1","S2"]},"prior_art":{"proximity":"SUBSTANTIAL_COLLISION","closest_analogues":[{"name":"W3C Multimodal Architecture and Interfaces","similarity":"Separates shared interaction flow and public data from black-box modality components, centralizes synchronization and submission, and defines an interface between modality components and an Interaction Manager.","remaining_difference":"It does not provide the proposed maintenance-report semantic laws, property-based action-sequence generator, mutation controls, or an acceptance rule comparing abstract outcomes across independently implemented adapters.","source_ids":["S3"]},{"name":"MARIA abstract user-interface language","similarity":"Specifies interaction semantics, data, dialogue behavior, effects, enabled events, and transitions independently of platform, modality, and implementation language, then refines them into modality-specific interfaces.","remaining_difference":"The proposal emphasizes a separately authored behavioral contract used as a reusable black-box substitutability oracle, with maintenance-specific invariants and leakage governance rather than primarily model-driven generation or migration.","source_ids":["S5"]},{"name":"SCXML state-machine control abstraction","similarity":"Provides standardized state, transition, event, data, cancellation, and invocation semantics suitable for modality-independent dialogue and task-flow control.","remaining_difference":"SCXML is a general execution notation, not a complete user-task contract, accessibility model, cross-adapter conformance method, or maintenance-report acceptance policy.","source_ids":["S4"]},{"name":"WCAG 2.2 complete-process and error-prevention requirements","similarity":"Already requires accessible complete processes and specifies checking, correction, confirmation, reversibility, and assistive-technology compatibility for relevant web interactions.","remaining_difference":"WCAG assesses accessibility requirements rather than behavioral equivalence among desktop, touch, voice, and assistive-technology adapters or representation independence of their workflow implementations.","source_ids":["S2"]}],"distinctive_claim_remaining":"Compared with separate modality specifications and with established abstract-UI or interaction-manager architectures, a contract-derived, mutation-tested black-box oracle can admit independently designed form and dialogue adapters without shared presentation structure while detecting task-level divergences in dependency invalidation, reachable semantic actions, cancellation, confirmation, and at-most-once submission. This is falsified if seeded violations survive, conforming adapters disagree at the declared abstract level, the oracle must encode presentation details to distinguish them, or an accessibility reviewer identifies a necessary accommodation that cannot be represented without bypassing the contract.","confidence":"HIGH"},"implementation_evidence":{"support":"STRONG","rationale":"Standards and primary research make the core architecture technically credible: W3C MMI separates modality components from shared flow/data, MARIA demonstrates modality-independent abstract interaction and dialogue models, SCXML supplies executable state-machine semantics, RFC 9110 supports the rationale for idempotent retry, and WCAG supplies relevant error and confirmation obligations. Remaining uncertainty is organizational and empirical: abstraction mappings may omit modality-specific agency, the backend must implement idempotency, and contract passing cannot substitute for accessibility, usability, security, or user validation.","source_ids":["S2","S3","S4","S5","S7"]},"scores":{"meaningful_impact":{"score":4,"rationale":"Preventing inaccessible task paths, unintended commits, and duplicate reports could materially improve task correctness and user agency. Empirical accessibility failures show consequential downside, although maintenance-specific frequency and severity are unmeasured.","source_ids":["S1","S2","S6"]},"stakeholder_pull":{"score":3,"rationale":"GOV.UK expresses clear institutional demand for joined-up channels and testing, but no maintenance organization or named budget holder has requested this intervention.","source_ids":["S1"]},"incremental_advantage":{"score":2,"rationale":"The shared abstract state and modality boundary substantially overlap W3C MMI and MARIA. The plausible increment is a mutation-tested substitutability oracle with maintenance-specific laws, whose advantage over separate modality tests is not yet measured.","source_ids":["S3","S5"]},"distinctiveness_plausibility":{"score":2,"rationale":"The remaining claim is contrastive and falsifiable, but combines established state-machine, abstract-UI, conformance-testing, accessibility, and idempotency practices rather than establishing a visibly new technical category.","source_ids":["S2","S3","S4","S5","S7"]},"technical_implementability":{"score":4,"rationale":"Existing standards support the required state model, event boundary, modality separation, data ownership, and retry semantics. Difficulty lies in specifying complete semantics and independent abstraction mappings rather than basic technical availability.","source_ids":["S3","S4","S5","S7"]},"adoption_authority_feasibility":{"score":2,"rationale":"A service owner and interaction-system owner are plausible authorizers, and analogous government service teams are empowered to change cross-channel experiences. The candidate has no named organization, partner, data owner, or accessibility sign-off authority.","source_ids":["S1"]},"evidence_readiness":{"score":4,"rationale":"A synthetic, non-production, two-adapter test with seeded mutants is bounded and requires no proprietary operational data. It can test technical contrast but not realized usability or workplace impact.","source_ids":["S3","S4","S5","S8"]},"safety_net_benefit":{"score":4,"rationale":"Pre-commit validation, explicit confirmation, cancellation guarantees, and idempotent retry directly address erroneous or duplicate submissions. Safety depends on backend enforcement and separate accessibility and usability review.","source_ids":["S2","S7"]},"scalability":{"score":4,"rationale":"A parameterized contract suite and representation-independent state model can be reused across adapters and report types, but contract stewardship and modality-specific accommodation review create continuing costs.","source_ids":["S3","S4","S5"]}},"score_confidence":"MODERATE","costs":{"first_evidence":{"band_2026_usd":"10K_TO_50K","scope":"Four-to-six-week synthetic sandbox study for one report type: specify the abstract model and 12 traces, build a simple form adapter and dialogue adapter, create two seeded mutants, run conformance tests, and obtain a limited accessibility review.","confidence":"MODERATE","assumptions":["Approximately 160-320 blended hours across software, QA, interaction design, and accessibility review.","2025 national mean wages are increased for benefits, overhead, specialist time, and 2026 rounding.","No participant recruitment, production integration, live data, procurement delay, or formal accessibility audit."],"source_ids":["S8"]},"initial_deployment_startup":{"band_2026_usd":"50K_TO_250K","scope":"Production-quality contract and shared oracle for one report type, adapter instrumentation, backend idempotency design, threat/privacy review, accessibility review, CI integration, and migration planning without full rollout.","confidence":"LOW","assumptions":["Approximately 800-1,800 blended labor hours.","Existing interfaces and backend expose testable seams and do not require major replacement.","Organization supplies product, domain, security, and accessibility personnel."],"source_ids":["S8"]},"operational_launch":{"band_2026_usd":"250K_TO_1M","scope":"Roll out the contract boundary and conformance gate across desktop, tablet, voice, and screen-reader-supported production interfaces for an initial reporting workflow, including security testing, telemetry, training, usability/accessibility evaluation, staged migration, and rollback capability.","confidence":"LOW","assumptions":["Approximately 3,000-8,000 blended hours across engineering, QA, design, accessibility, security, operations, and program management.","No hardware replacement, enterprise-wide backend rewrite, regulated clinical deployment, or large-scale workforce study.","The organization already operates the four interface channels."],"source_ids":["S8"]},"annual_recurring":{"band_2026_usd":"50K_TO_250K","scope":"Contract stewardship, regression and mutation-suite maintenance, adapter recertification, accessibility review of changes, incident triage, and periodic leakage audits for the initial workflow.","confidence":"LOW","assumptions":["Approximately 0.5-1.5 full-time-equivalent staff plus specialist reviews.","Scope remains one workflow family and a small number of adapters.","Excludes general operation of the reporting service and unrelated interface development."],"source_ids":["S8"]}},"verified_pipeline_gates":{"externally_supported_problem":{"status":"YES","reason":"Primary research documents substantial screen-reader barriers in complex submission workflows, while official guidance and WCAG establish the importance of coherent, accessible cross-channel processes. Direct maintenance-setting prevalence remains a gap but does not erase the externally visible problem class.","source_ids":["S1","S2","S6"]},"externally_credible_adopter_or_authorizer":{"status":"UNCERTAIN","reason":"The UK Government Digital Service is a credible analogous authorizer expressing cross-channel need, but no field-maintenance organization, service owner, accessibility authority, or funder is identified for this candidate.","source_ids":["S1"]},"distinct_testable_incremental_claim":{"status":"YES","reason":"The remaining claim compares a mutation-tested, task-state conformance boundary against separate modality specifications and tests; seeded violations, cross-adapter state divergence, presentation coupling, and missing accommodations are explicit falsifiers.","source_ids":["S3","S4","S5"]},"bounded_next_evidence_step":{"status":"YES","reason":"One synthetic report type, 12 preregistered traces, two adapters, two seeded mutants, one accessibility review, and fixed pass/fail criteria form a bounded four-to-six-week study.","source_ids":["S3","S4","S5","S8"]},"no_unresolved_safety_or_authority_stop":{"status":"YES","reason":"The first test can remain synthetic and non-production, with no employee or incident data and no live commits. It must stop if a prototype can reach live records, a required accommodation needs a bypass, or a seeded invariant violation passes.","source_ids":["S2","S7"]},"credible_cost_scope_and_range":{"status":"YES","reason":"The four ranges are scoped as resource-equivalent labor estimates and anchored to official occupational wage data, with explicit exclusions and uncertainty. Actual local rates, procurement, integration complexity, and effort are unverified.","source_ids":["S8"]}},"next_evidence_step":"With a named service-owner sponsor and accessibility reviewer, preregister a four-to-six-week synthetic study for one report type. Independently specify the abstract model and 12 traces before implementation; have separate developers build a form adapter and a dialogue adapter; compare them with the baseline of separate modality specifications and incumbent end-to-end tests. Seed at least four defects: failure to invalidate a dependent answer, navigation that mutates facts, cancellation that commits, and duplicate commit after an ambiguous retry. Primary comparators are mutant-kill rate, agreement on abstract state and reachable actions after every trace step, unauthorized transition count, presentation-specific assertions required by the oracle, and implementation effort. Continue only if all seeded defects are rejected, both valid adapters produce identical declared task outcomes without shared presentation structure, the baseline detects fewer seeded divergences or requires more modality-specific assertions, and the accessibility reviewer finds no required semantic action outside the contract. Falsify the intervention if any seeded defect passes, a declared obligation diverges, presentation details are needed to obtain agreement, or a necessary accommodation must bypass the shared task state.","blocking_evidence":["No direct measurement shows how often desktop, tablet, voice, and screen-reader maintenance-report interfaces currently diverge at the task-state level.","No named maintenance organization, service owner, funder, accessibility authority, or field-technician group has expressed demand or authorized a pilot.","No head-to-head evidence shows that the proposed oracle detects more consequential cross-modality defects than separate modality specifications and end-to-end tests.","No independent accessibility review has established that the shared state model preserves modality-specific accommodations and user agency.","Backend support for intent-token idempotency, privacy-safe telemetry, security controls, and rollback has not been verified.","Project-specific effort, local loaded labor rates, integration complexity, and recurring governance burden are unknown."],"research_disposition":"PARTNERED_RESEARCH_PROGRAM","world_novelty_boundary":"This evaluation measures neither world novelty nor patentability, freedom to operate, market size, or realized impact. Bounded searching found substantial conceptual prior art in W3C multimodal architecture, SCXML, MARIA, WCAG, and HTTP idempotency. The only remaining evaluated distinction is the maintenance-specific, mutation-tested black-box substitutability claim; novelty beyond the eight opened sources is unmeasured.","arm":"COMPLETE_PROPOSAL_PORTFOLIO","candidate_version":0,"controller_recommendation":{"action":"STOP_EMPIRICAL_RESEARCH_NEEDED","repairable":false,"material_progress_observed":true,"progress_targets":["Obtain a named field-maintenance service-owner sponsor, accessibility reviewer, and written sandbox authority.","Measure matched task-state traces on existing modalities to establish whether the proposed problem occurs in the target organization.","Run the preregistered two-adapter study against separate-specification and incumbent-test baselines.","Demonstrate 100% rejection of seeded invariant violations and zero divergence on declared task obligations.","Show that at least one materially different modality-specific accommodation fits through declared extension points without weakening shared task semantics.","Replace resource-equivalent estimates with partner-specific engineering, integration, accessibility, security, and recurring-governance estimates."],"reason":"Web research establishes that the problem class matters and that implementation is feasible, but it also reveals substantial prior-art collision. The remaining advantage—whether a mutation-tested task-state oracle improves cross-modality defect detection while preserving necessary accommodations—cannot be resolved by additional bounded web search; it requires target-organization traces, an authorized sandbox, independent adapters, and live human accessibility judgment."},"proposal_index":1}