{"schema_version":1,"research_id":"eoa_inverse_innovation_exp06_external_evaluation_20260803","source_assessment_id":"catalytic_pathway_enablement__human_computer_interaction:P2:v0","cell_id":"catalytic_pathway_enablement__human_computer_interaction","search_queries":["site:rfc-editor.org floor control protocol floor request grant release BFCP RFC","site:w3.org real-time communication accessibility user requirements raised hand turn taking","site:support.zoom.com raised hand meeting participant queue order","site:support.microsoft.com Teams raise hand order meeting","remote meetings turn taking overlap interruptions study videoconferencing primary research CSCW","multimodal floor control collaborative systems research voice chat shared editing floor broker","videoconferencing turn-taking delay overlap experimental study","site:bls.gov software developers occupational employment wage 2025","OpenFloor protocol official floor management multimodal voice text interoperability","Zoom Meeting SDK participant raise hand API official documentation","Microsoft Graph Teams meeting app real time media participant controls API official","EU GDPR data minimisation official regulation personal data telemetry meetings","\"Whose turn is it anyway?\" Seuren Shaw Greenhalgh 2021 journal","\"Zoom disrupts the rhythm of conversation\" publisher 2022 journal"],"sources":[{"source_id":"S1","title":"Whose turn is it anyway? Latency and the organization of turn-taking in video-mediated interaction","publisher":"Elsevier, Journal of Pragmatics; hosted by NCBI PubMed Central","url":"https://pmc.ncbi.nlm.nih.gov/articles/PMC7819463/","source_class":"PRIMARY_RESEARCH","publication_date":"2021-01-01","accessed_at":"2026-08-03","claims_supported":["A secondary analysis of 25 recorded video consultations found that latency can create perceived silence, overlapping talk, and difficulty restoring one-speaker-at-a-time interaction.","The evidence directly supports the existence of recurrent turn-transition problems, but its clinical setting and emphasis on voice latency do not establish prevalence across general multimodal collaboration."]},{"source_id":"S2","title":"Zoom disrupts the rhythm of conversation","publisher":"American Psychological Association, Journal of Experimental Psychology: General; indexed by NCBI PubMed","url":"https://pubmed.ncbi.nlm.nih.gov/34748361/","source_class":"PRIMARY_RESEARCH","publication_date":"2022-06","accessed_at":"2026-08-03","claims_supported":["Controlled and unscripted experiments found longer turn-transition times in remote Zoom interaction than local or face-to-face interaction.","For the same dyads, reported average transition times were 135 milliseconds face-to-face and 487 milliseconds over Zoom, supporting a meaningful interaction-timing problem."]},{"source_id":"S3","title":"RFC 8855: The Binary Floor Control Protocol (BFCP)","publisher":"RFC Editor and Internet Engineering Task Force","url":"https://www.rfc-editor.org/rfc/rfc8855.html","source_class":"STANDARD","publication_date":"2021-01","accessed_at":"2026-08-03","claims_supported":["Floor control for shared conferencing resources is an established standardized practice.","BFCP already specifies authenticated participants, floor requests, acknowledgments and status, grants or denials, releases, queues, moderator or chair actions, authorization, and coherent server-maintained floor state.","The request-grant-release core of the candidate therefore substantially overlaps established prior art."]},{"source_id":"S4","title":"Raise your hand in Microsoft Teams meetings","publisher":"Microsoft","url":"https://support.microsoft.com/en-us/teams/meetings/raise-your-hand-in-microsoft-teams-meetings","source_class":"OFFICIAL_PRODUCT_DOCUMENTATION","publication_date":"undated; current as accessed","accessed_at":"2026-08-03","claims_supported":["Microsoft identifies a user need to indicate an intention to contribute without interrupting conversation.","Teams already makes bids visible, orders multiple raised hands, notifies presenters, automatically lowers a hand after speech in some conditions, permits withdrawal, and gives organizers a lower-all control.","This is a deployed adopter analogue and a close comparator, though it does not establish demand for an independent cross-tool broker."]},{"source_id":"S5","title":"RTC Accessibility User Requirements","publisher":"World Wide Web Consortium Accessible Platform Architectures Working Group","url":"https://www.w3.org/TR/raur/","source_class":"OFFICIAL_GUIDANCE","publication_date":"2021-05-25","accessed_at":"2026-08-03","claims_supported":["W3C documents accessibility needs spanning audio, video, text, sign language, augmentative and alternative communication, assistive technologies, status information, and separately routed alerts.","Participant identity and status must be accessible to assistive technologies, while preserved participant metadata should be handled with consent.","The document is a Working Group Note rather than a conformance standard and does not validate the candidate's allocation policy."]},{"source_id":"S6","title":"Open Floor Inter-Agent Message Specification Version 1.1.0","publisher":"Open Floor Project, Open Voice Interoperability Initiative, LF AI & Data Foundation","url":"https://openfloor.dev/protocol/specifications/inter-agent-message","source_class":"STANDARD","publication_date":"2026; version 1.1.0","accessed_at":"2026-08-03","claims_supported":["An open specification already models multiparty conversations with human and autonomous participants, a platform-independent floor manager, media proxying, extensible media types, assigned floor roles, and grant and revoke events.","It is a close interoperability analogue to the proposed broker, although oriented toward conversational agents rather than synchronized human speaking, editing, pointing, and assistive-input access across existing collaboration products.","Authentication, authorization, accounting, and security remain outside this specification, exposing an implementation gap."]},{"source_id":"S7","title":"Working with the cloud communications API in Microsoft Graph","publisher":"Microsoft Learn","url":"https://learn.microsoft.com/en-us/graph/api/resources/communications-api-overview?view=graph-rest-1.0","source_class":"OFFICIAL_PRODUCT_DOCUMENTATION","publication_date":"2024-06-26","accessed_at":"2026-08-03","claims_supported":["Microsoft supports service applications that act as participants in Teams meetings and exposes call, meeting, participant-control, and presence capabilities.","Relevant permissions require administrator grants, making the platform owner or tenant administrator an identifiable authorizer.","Microsoft prohibits recording or persisting accessed meeting media or data derived from that media through these APIs; whether the candidate's event telemetry and intended controls fit every platform's terms remains unverified."]},{"source_id":"S8","title":"Software Developers, Quality Assurance Analysts, and Testers","publisher":"U.S. Bureau of Labor Statistics","url":"https://www.bls.gov/ooh/Computer-and-Information-Technology/Software-developers.htm","source_class":"GOVERNMENT_OR_REGULATOR","publication_date":"2025; May 2024 wage reference period","accessed_at":"2026-08-03","claims_supported":["The median annual wage was $133,080 for software developers and $102,610 for software quality-assurance analysts and testers in May 2024.","These wages provide an official labor anchor for resource-equivalent cost ranges; estimates still require assumptions about 2026 inflation, benefits, overhead, accessibility expertise, research labor, security review, and platform-integration effort."]}],"problem_evidence":{"support":"STRONG","rationale":"Two primary studies independently show that video-mediated latency disrupts rapid turn transitions through longer gaps, overlap, and repair difficulty. Microsoft’s deployed raise-hand feature explicitly addresses contributing without interruption. Evidence is strongest for spoken turn-taking; the frequency and consequence of simultaneous editing, pointing, chat, and assistive-input contention across ordinary workplaces remain unmeasured.","source_ids":["S1","S2","S4"]},"stakeholder_evidence":{"support":"MODERATE","rationale":"Microsoft is an identifiable platform analogue and its documentation expresses the contribution-without-interruption need. W3C expresses accessibility needs for multimodal communication, assistive-technology-visible status, and participant control. Microsoft also identifies tenant administrators as required API authorizers. No organization has expressed demand for this particular independent cross-tool broker or committed to a pilot, so stakeholder pull is not strong.","source_ids":["S4","S5","S7"]},"prior_art":{"proximity":"ESTABLISHED_PRACTICE","closest_analogues":[{"name":"IETF Binary Floor Control Protocol","similarity":"Very high: a server receives authenticated floor requests, returns status, grants or denies temporary access to shared resources, supports queues and chair actions, processes release or revocation, and maintains coherent floor state.","remaining_difference":"BFCP does not itself provide the candidate's specific human-facing adapters across proprietary voice, chat, canvas, keyboard, and assistive-input products, its accessibility policy, or comparative outcome evidence.","source_ids":["S3"]},{"name":"Microsoft Teams raised-hand workflow","similarity":"High: participants submit explicit content-free bids; bids are visible and ordered; presenters can grant practical speaking access; participants can withdraw; hands may lower after speech; organizers retain control.","remaining_difference":"Teams is product-native and mainly coordinates speaking rather than applying one declared policy across speaking, shared editing, pointing, chat, and assistive modalities or across vendors.","source_ids":["S4"]},{"name":"Open Floor Inter-Agent Message Specification","similarity":"High: it defines a platform-independent multiparty floor manager, human and agent conversants, media gateways, floor roles, and floor-grant state across technology stacks.","remaining_difference":"It primarily targets interoperable conversational agents and leaves authentication, authorization, accounting, and security outside scope; the candidate targets already-authorized human collaborators and proposes accessibility, exception, telemetry, and cross-tool workflow governance.","source_ids":["S6"]}],"distinctive_claim_remaining":"Under held-constant participation rules, a content-blind broker applying one published policy across at least voice and shared-edit bids can reduce collision-adjusted bid-to-completed-contribution time and moderator work relative to native raised-hand ordering, ordinary coordination, and human moderation, without increasing abandonment, wrong grants, modality-specific exclusion, side-channel displacement, or reset failures.","confidence":"HIGH"},"implementation_evidence":{"support":"MODERATE","rationale":"BFCP proves the request/status/grant/release state machine is technically conventional; Open Floor demonstrates platform-independent JSON floor management and media proxy patterns; and Microsoft Graph permits meeting-participant bots and participant controls. W3C provides concrete multimodal accessibility requirements. However, cross-vendor real-time control parity, shared-canvas hooks, assistive-input adapters, ordering under network delay, and exact product terms were not verified. Open Floor leaves security and authorization out of scope, and Microsoft requires administrator permission and restricts persistence of media-derived data. A standalone sandbox is implementable; a production cross-tool service is not yet demonstrated.","source_ids":["S3","S5","S6","S7"]},"scores":{"meaningful_impact":{"score":3,"rationale":"The underlying timing and overlap problem is experimentally visible and can impair participation, but population prevalence and downstream organizational harm for the proposed multimodal scope are unknown.","source_ids":["S1","S2","S5"]},"stakeholder_pull":{"score":3,"rationale":"Major platforms already operate raised-hand controls and W3C records accessibility needs, showing a real demand category. There is no committed adopter or expressed request for an independent cross-tool broker.","source_ids":["S4","S5","S7"]},"incremental_advantage":{"score":2,"rationale":"Cross-tool and cross-modality coordination could add value, but the core floor-control cycle is standardized and deployed. Comparative advantage over native queues and skilled moderators is entirely untested.","source_ids":["S3","S4","S6"]},"distinctiveness_plausibility":{"score":2,"rationale":"The remaining cross-vendor, multimodal, accessibility-preserving claim is contrastive and testable, but it is a narrow systems-integration and outcome claim within an established practice rather than a demonstrated new mechanism.","source_ids":["S3","S4","S6"]},"technical_implementability":{"score":3,"rationale":"A bounded standalone prototype is feasible using conventional floor-state primitives. Production feasibility is constrained by administrator permissions, proprietary APIs, adapter parity, latency, and security work.","source_ids":["S3","S6","S7"]},"adoption_authority_feasibility":{"score":3,"rationale":"A meeting host can authorize a low-consequence opt-in sandbox, while platform or tenant administrators are identifiable integration authorizers. Production adoption still depends on vendor permissions and organizational accessibility, privacy, and security review.","source_ids":["S4","S7"]},"evidence_readiness":{"score":4,"rationale":"The candidate specifies a bounded counterbalanced trial, event-level outcomes, comparators, halt rules, and clear falsifiers. Six sessions can establish feasibility and estimate effects, not reliable general efficacy.","source_ids":["S1","S2","S3"]},"safety_net_benefit":{"score":3,"rationale":"Accessible status signals, explicit withdrawal, moderator override, and an immediate return to ordinary controls could protect participation when the broker fails. The same mechanism could formalize exclusion or concentrate control if signals or policy are biased.","source_ids":["S5","S7"]},"scalability":{"score":2,"rationale":"The state machine and protocol can scale computationally, but each collaboration product and modality requires authorization, adapter maintenance, accessibility validation, and policy-specific governance; these costs undermine simple diffusion.","source_ids":["S5","S6","S7"]}},"score_confidence":"MODERATE","costs":{"first_evidence":{"band_2026_usd":"10K_TO_50K","scope":"Preregister and run six low-consequence, counterbalanced sandbox sessions using a prototype with speak and edit bids, three comparator conditions, participant incentives, accessibility review, event instrumentation, analysis, and a manual override.","confidence":"MODERATE","assumptions":["Approximately 150-350 combined engineering, research, facilitation, accessibility, and analysis hours.","A lightweight standalone prototype is used rather than production Teams or Zoom integration.","Resource-equivalent labor is based on BLS software-development and QA wages, adjusted upward for 2026 inflation, benefits, overhead, and specialist time.","No contribution content is recorded."],"source_ids":["S5","S8"]},"initial_deployment_startup":{"band_2026_usd":"250K_TO_1M","scope":"One-platform controlled organizational pilot with production authentication, one voice adapter, one shared-workspace adapter, accessible notifications, audit logging, reliability testing, privacy and security review, incident response, and administrator deployment.","confidence":"LOW","assumptions":["Roughly 2-5 full-time-equivalent years across software, QA, HCI research, accessibility, privacy, security, and program management.","Existing platform APIs expose sufficient event and participant controls.","The organization supplies a host, identity system, test tenant, and legal review.","This excludes broad multi-vendor compatibility."],"source_ids":["S5","S7","S8"]},"operational_launch":{"band_2026_usd":"1M_TO_5M","scope":"Launch a supported service across at least two conferencing products and one shared canvas, including assistive-input paths, policy administration, observability, high availability, penetration and accessibility testing, support, documentation, and rollback operations.","confidence":"LOW","assumptions":["Approximately 8-20 full-time-equivalent years plus cloud, security-audit, support, and certification expenses.","Platform vendors authorize the necessary controls and do not require bespoke commercial agreements beyond the band.","No safety-critical or rights-affecting use is included.","Cross-vendor semantics require substantial adapter and conformance work."],"source_ids":["S5","S6","S7","S8"]},"annual_recurring":{"band_2026_usd":"1M_TO_5M","scope":"Operate and maintain the multi-tool service, including connector updates, on-call reliability, security and privacy review, accessibility regression testing, policy audits, customer support, incident response, and periodic effectiveness evaluation.","confidence":"LOW","assumptions":["A 6-15 person equivalent multidisciplinary operations and product team is retained.","Vendor API change and accessibility-regression work are continuous.","Infrastructure cost remains secondary to skilled labor.","The estimate excludes litigation, regulated-sector certification, and global localization."],"source_ids":["S5","S6","S7","S8"]}},"verified_pipeline_gates":{"externally_supported_problem":{"status":"YES","reason":"Two primary studies directly observe remote turn-transition delay, overlap, and repair problems, and a major platform documents a feature intended to let users contribute without interruption.","source_ids":["S1","S2","S4"]},"externally_credible_adopter_or_authorizer":{"status":"YES","reason":"Collaboration-platform owners and tenant administrators are credible authorizers: Microsoft operates an analogous feature and requires administrator-granted communications permissions. W3C supplies an external accessibility stakeholder mandate. No pilot commitment has been obtained.","source_ids":["S4","S5","S7"]},"distinct_testable_incremental_claim":{"status":"YES","reason":"The remaining claim specifies cross-modality brokerage and measurable outcomes against native raised-hand, ordinary-coordination, and human-moderator comparators while holding participation rules constant.","source_ids":["S3","S4","S6"]},"bounded_next_evidence_step":{"status":"YES","reason":"A six-session, low-consequence, counterbalanced sandbox with content-free telemetry, manual override, immediate rollback, explicit comparators, and preregistered falsifiers is bounded in people, task, duration, data, and risk.","source_ids":["S1","S2","S5"]},"no_unresolved_safety_or_authority_stop":{"status":"YES","reason":"For the sandbox only, consenting adults, a non-consequential task, no content capture, an accessible manual path, human override, and immediate restoration of ordinary controls bound the risks. Production platform authorization, privacy interpretation, and accessibility conformance remain unresolved but do not prevent the standalone evidence step.","source_ids":["S5","S7"]},"credible_cost_scope_and_range":{"status":"YES","reason":"Each band states a distinct scope and staffing assumptions anchored to official BLS wages. Confidence is low beyond the first experiment because vendor access and adapter complexity are unknown.","source_ids":["S7","S8"]}},"next_evidence_step":"Preregister and conduct six low-consequence collaborative-whiteboard sessions using a counterbalanced within-group design with three conditions: the proposed broker, the platform's native chronological raised-hand workflow with host control, and ordinary coordination with a human moderator. Use only explicit content-free speak and edit bids; retain moderator override and an always-visible return to ordinary controls. Measure bid acknowledgment, bid-to-grant and bid-to-completion time, overlap or overwrite events per eligible bid, abandonment, wrong grants, false duplicate rejection, moderator interventions and active work time, repeat-turn concentration, outcomes by input modality, policy comprehension and contestability, side-channel migration, adapter faults, and post-release reset integrity. Treat the probe as falsified or stopped if any simultaneous or unauthorized grant occurs, any participant cannot perceive or return a token, override or reset fails, contribution content is exposed, the queue deadlocks, more than 20% of ordinary cycles require moderator repair, or the broker fails to show at least a 20% directional reduction in both collision-adjusted median bid-to-completion time and moderator active work without increasing abandonment by more than 5 percentage points or producing a modality outcome gap greater than 10 percentage points. The small sample may establish feasibility and estimate effects only; a subsequent powered study would be required for efficacy.","blocking_evidence":["No representative prevalence estimate shows how often cross-modality contribution contention, rather than voice latency alone, materially harms synchronous collaboration.","No named adopter has committed participants, a test tenant, staff time, or platform permissions.","No live comparison establishes advantage over native raised-hand ordering, ordinary social coordination, or a skilled human moderator.","Cross-vendor access to real-time raised-hand, microphone, shared-edit, pointer, presence, and assistive-input controls has not been verified; Microsoft documentation covers only part of the proposed surface.","No jurisdiction-specific legal determination establishes that the proposed identity-linked timing and allocation telemetry complies with applicable privacy, employment, and communications rules or every platform's terms.","No evidence establishes acceptable accessibility parity, participant comfort, policy contestability, strategic-bid resistance, side-channel displacement, or downstream group absorption.","No load, outage, state-reconciliation, adapter-drift, or long-run operational evidence supports production scalability."],"research_disposition":"PARTNERED_RESEARCH_PROGRAM","world_novelty_boundary":"World novelty, patentability, freedom to operate, market size, and realized impact were not measured. The search establishes only that the core request-grant-release floor-control mechanism is standardized and deployed, while the remaining cross-vendor multimodal accessibility claim is unvalidated; it does not establish that no equivalent implementation, publication, patent, or private system exists.","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 platform-owner or tenant-administrator partner and written authorization for the exact sandbox controls and telemetry.","Verify an adapter-capability matrix for at least two conferencing products, one shared canvas, and the intended assistive-input paths, including permissions, event latency, control limits, data-retention terms, and failure behavior.","Complete accessibility, privacy, security, and platform-terms review for the bounded protocol before collecting participant data.","Preregister the three-condition counterbalanced probe, thresholds, exclusions, event definitions, analysis, halt rules, and content-deletion procedure.","Demonstrate zero critical grant, override, content-exposure, and reset failures; moderator repair below 20% of ordinary cycles; and no material modality disparity.","Estimate whether the broker achieves the preregistered directional improvement over both native raised-hand and human-moderated comparators without greater abandonment, discomfort, or side-channel displacement.","If feasibility thresholds are met, design a powered multi-group study that separates voice-latency problems from recognition, ordering, editing, and assistive-modality contention."],"reason":"Bounded web research verifies the problem category, credible authorizers, implementable protocol primitives, close standards and product prior art, and a costable next experiment. It cannot determine the remaining incremental claim because that requires live participant behavior, proprietary platform capability checks, accessibility testing, and comparative performance data. Under the controller rule, the candidate therefore stops for empirical research rather than further general web search."},"proposal_index":2}