{"schema_version":1,"experiment_id":"eoa_inverse_innovation_exp06_four_proposal_generalization60_20260803","cell_id":"catalytic_pathway_enablement__human_computer_interaction","arm":"COMPLETE_PROPOSAL_PORTFOLIO","candidate_id":"hci_multimodal_floor_broker","proposal_index":2,"version":0,"title":"Multimodal Contribution-Floor Broker","problem":"In synchronous computer-mediated collaboration, participants can attempt to speak, edit a shared surface, present a pointer, or request clarification at the same time. Each transition from an intention to contribute to an acknowledged contribution then requires repeated negotiation through modality-specific cues, which may arrive late, conflict, or be invisible to participants using another channel. The group already permits each eligible contribution, but contention at the shared interaction boundary repeatedly delays or disrupts delivery.","actors":["Participants in synchronous collaborative sessions","Meeting or workspace host","Human session moderator","Collaboration-platform owner","Accessibility and interaction designers","Privacy and security reviewers","Floor-broker service steward"],"observable_state":"Two or more eligible contribution attempts overlap; edits overwrite or interrupt one another; participants repeat bids because acknowledgment is unclear; a raised-hand, chat, voice, keyboard, or assistive-switch signal is not visible across modalities; the moderator repeatedly reconstructs bid order; or an acknowledged participant loses the floor before completing the contribution.","consequence":"The group spends interaction time negotiating access to the shared channel, contributions can be abandoned or duplicated, and modality-dependent collisions can distort who is able to complete an otherwise permitted contribution.","affected_objective":"Reduce recurring contention between an eligible contribution bid and its completed delivery in synchronous collaborative interfaces while preserving participation rights, content neutrality, accessibility, participant control, and human authority over exceptional situations.","intervention":"Deploy an independently operated, cross-tool floor-broker service with adapters for supported voice, chat, shared-canvas, keyboard, and assistive-input controls. An opted-in group selects a published allocation policy before the session. Participants submit explicit bids without exposing contribution content; the reusable broker validates identity, modality availability, duplicate or stale bids, and current floor state, then acknowledges eligible bids and grants one visible, time-bounded contribution token. The token coordinates only the shared interaction channel and does not evaluate content. It releases when the participant finishes, returns it, becomes inactive, or a moderator invokes a declared exception. The broker then clears turn-specific state, reconciles the remaining queue, verifies channel health, and returns ready for the next turnover. Saturation, wait, abandoned bids, incorrect grants, overrides, modality-specific failures, and downstream group capacity govern metering, refresh, and shutdown.","structural_mapping":[{"archetype_element":"Target transformation specification","domain_realization":"Transform an eligible, acknowledged attempt to contribute in a synchronous digital workspace into a completed contribution with the temporary floor released for reuse."},{"archetype_element":"Activation barrier","domain_realization":"Participants must repeatedly negotiate recognition, ordering, channel ownership, and release through inconsistent social and interface cues whenever contribution attempts collide."},{"archetype_element":"Permitted pathway boundary","domain_realization":"The broker may coordinate timing and channel access among already authorized participants; it may not judge contribution content, determine membership, change substantive priority, suppress disagreement, or create permission to participate."},{"archetype_element":"Eligible substrate","domain_realization":"An explicit contribution bid from an authenticated session participant, submitted through a supported modality while the group has opted into a declared allocation policy."},{"archetype_element":"Reusable facilitator","domain_realization":"A standing multimodal floor-broker service that repeatedly translates bids into temporary, observable contribution grants and becomes available again after each release."},{"archetype_element":"Facilitator–substrate interface","domain_realization":"Adapters expose a common bid, acknowledgment, queue-position, grant, completion, withdrawal, exception, and release contract across supported collaboration channels."},{"archetype_element":"Selectivity rule","domain_realization":"The broker admits attributable, current, single-participant bids and rejects or reroutes duplicates, stale presence, unsupported modalities, conflicting grants, and exception classes requiring a human moderator."},{"archetype_element":"Cofactor system","domain_realization":"Authenticated presence, functioning modality adapters, participant-visible signals, an agreed allocation policy, and a human exception path enable the broker but remain distinct from the reusable mediation service."},{"archetype_element":"Release condition","domain_realization":"A grant ends through explicit completion or return, a visible inactivity timeout, participant departure, or an authorized moderator action; the broker cannot retain control silently."},{"archetype_element":"Regeneration cycle","domain_realization":"After every release, the service clears the expired token, reconciles pending bids and presence, checks adapter health, records bounded cycle outcomes, and restores a clean ready state."},{"archetype_element":"Turnover capacity","domain_realization":"Capacity is represented by completed contribution cycles per broker instance and time window, measured with acknowledgment latency, wait, floor duration, queue depth, abandoned bids, overrides, recovery time, and downstream absorption."},{"archetype_element":"Saturation and interference","domain_realization":"Simultaneous bid bursts, repeated bids, long-held floors, network delay, stale presence, adapter failure, and strategic use of exception classes can occupy or distort the broker."},{"archetype_element":"Byproduct and side-pathway guardrail","domain_realization":"The service monitors incorrect grants, inaccessible or missed signals, excessive repeat-turn concentration, forced interruption, false duplicate rejection, unresolved deadlock, and transfer of discussion into ungoverned side channels."},{"archetype_element":"Downstream recipient","domain_realization":"The collaborating group receives each completed contribution. Admission and floor duration are paced to the group's ability to perceive and respond rather than maximizing turn count."},{"archetype_element":"Equilibrium neutrality","domain_realization":"The broker accelerates orderly exercise of participation opportunities the group already permits; it does not resolve power differences, objective conflict, insufficient meeting time, inaccessible content, or lack of substantive authority."},{"archetype_element":"Accountable steward and deactivation","domain_realization":"A named service steward owns adapter integrity, policy transparency, accessibility, capacity, audit sampling, incident response, reset, and succession, while participants, hosts, and privacy or security owners retain disable paths."}],"mechanism_mapping":[{"mechanism_slug":"reusable_broker_or_convener_service","role":"Provides the standing intermediary that translates modality-specific bids into a common, governed floor state across repeated contribution cycles.","counterfactual_removal":"Without the reusable broker, participants or a moderator must renegotiate recognition and ordering whenever bids collide."},{"mechanism_slug":"interface_contract_design","role":"Defines bid, acknowledgment, grant, withdrawal, exception, completion, and release semantics consistently across collaboration tools and input modalities.","counterfactual_removal":"Without the contract, channel adapters can interpret the same action differently and recreate cross-modality coordination friction."},{"mechanism_slug":"catalyst_cofactor_system","role":"Requires authenticated presence, healthy adapters, accessible notifications, a selected allocation policy, and moderator coverage before brokerage begins.","counterfactual_removal":"Without explicit cofactor checks, the broker may appear ready while a participant cannot receive grants, return the floor, or invoke an exception."},{"mechanism_slug":"fast_track_with_eligibility_rules","role":"Routes ordinary attributable bids through automatic allocation while sending contested, urgent, inaccessible, or policy-ambiguous situations to the human moderator without weakening participation rights.","counterfactual_removal":"Without selective routing, either every bid consumes moderator attention or exceptional conflicts are forced through an unsuitable automatic rule."},{"mechanism_slug":"inhibitor_and_poison_screen","role":"Screens duplicate submissions, stale identities, automation spam, conflicting tokens, unsupported adapters, and unhealthy channel conditions before granting the floor.","counterfactual_removal":"Without screening, invalid bids can occupy the queue, create simultaneous grants, or prevent legitimate turnover."},{"mechanism_slug":"active_site_capacity_dashboard","role":"Shows active floor state, eligible queue depth, acknowledgment and wait times, abandoned bids, adapter degradation, overrides, and moderator exception load.","counterfactual_removal":"Without capacity visibility, hosts can mistake a poisoned adapter or stuck token for ordinary demand and continue adding bids to a blocked pathway."},{"mechanism_slug":"catalyst_regeneration_protocol","role":"Revokes expired tokens, clears turn-specific state, reconciles presence and queues, restores adapters, and retires a policy or service version that cannot recover safe turnover.","counterfactual_removal":"Without regeneration, stale grants and residual queue state can contaminate later turns and make the broker a persistent bottleneck."},{"mechanism_slug":"turnover_and_selectivity_assay","role":"Compares completed eligible contribution cycles with unbrokered collaboration while counting collisions, wrong grants, false rejections, abandoned bids, overrides, modality-specific outcomes, and degradation.","counterfactual_removal":"Without the assay, a high turn count could be credited as catalytic success despite inaccessible allocation, concentrated capture, or increased moderator work."},{"mechanism_slug":"small_safe_to_fail_probe","role":"Tests the service in low-consequence sandbox collaboration with bounded sessions, participant opt-in, a manual override, and predeclared stop conditions.","counterfactual_removal":"Without a contained counterfactual probe, deadlock, inaccessible signaling, or unfair allocation could shape consequential group activity before the pathway is validated."}],"causal_chain":["A collaborating group explicitly opts into the broker and selects a published allocation policy before beginning a bounded session.","Authenticated participants submit content-free contribution bids through supported voice, chat, canvas, keyboard, or assistive-input controls.","The inhibitor screen checks current presence, duplicates, adapter health, conflicting grants, and exception conditions.","The reusable broker acknowledges eligible bids through perceptible multimodal signals and exposes the current floor and queue state.","The broker grants a visible, temporary contribution token according to the selected policy; contested or exceptional cases route to the human moderator.","The participant uses the shared channel and then completes, returns, or visibly times out the token without the broker judging the contribution's content.","The group absorbs the contribution before the service advances at the permitted rate.","The broker clears turn-specific state, reconciles presence and pending bids, verifies adapter health, and returns ready for another cycle.","Wait, collisions, abandonment, wrong grants, overrides, modality failures, queueing, and downstream load update capacity and selectivity records.","Comparison with unbrokered and human-moderated sessions tests whether the reusable broker lowers recurrent access negotiation without changing participation rights or merely shifting contention elsewhere."],"baseline":"Participants coordinate contribution timing through conversational cues, platform-specific raised hands, chat messages, cursor activity, and host intervention. A moderator notices collisions, reconstructs ordering, invites contributors, and resolves exceptions manually. The comparison holds group membership, task, communication channels, substantive participation rules, and moderator authority constant.","nearest_rivals":["An unmediated raised-hand or chronological queue, which displays ordering but may not coordinate grants and releases across speaking, editing, pointing, chat, and assistive modalities.","A human meeting moderator, who can interpret context and power dynamics but expends attention on every collision and may become the throughput constraint.","Round-robin participation, which guarantees a fixed sequence but can force empty turns, delay timely clarification, and ignore changing contribution readiness.","A chat backchannel, which permits parallel input but transfers attention and unresolved contention into another stream rather than governing the shared floor.","Shared-editing conflict resolution, which repairs or merges concurrent changes after collision rather than selectively coordinating access before contribution.","Generic meeting controls such as mute and spotlight, which can enforce channel ownership but do not provide a complete bid-to-grant-to-release cycle with transparent access rules."],"remaining_contrastive_claim":"The proposal tests a real-time catalytic coordination cycle: a standing intermediary repeatedly receives content-free bids, selectively grants a temporary shared floor, releases it after contribution, regenerates its state, and serves the next bid. Its contrast is not a renamed queue or moderator role; it depends on cross-modality brokerage, observable turnover, bounded exceptions, regeneration, and governance without content judgment.","authority_safety":{"decision_authority":"The collaboration-platform owner may authorize the sandbox integration after accessibility, privacy, and security review. Each group must opt in and select the allocation policy. Participants control their bids and may return the floor; the human host retains visible override and shutdown authority. The broker has no authority over membership, content, evaluation, or substantive decisions.","authorized_first_step":"Run at most six counterbalanced sandbox sessions with consenting adults completing the same low-consequence collaborative whiteboard task under broker-on and ordinary coordination conditions. Limit the broker to explicit speak and edit bids, keep a human moderator present, collect timing and allocation events rather than contribution content, and restore ordinary controls immediately on request.","excluded_actions":["Covert or mandatory activation","Analysis, ranking, transcription, or moderation of contribution content","Inferring disability, status, sentiment, importance, or participant worth","Changing group membership, voting rights, agenda authority, or substantive decision rules","Using the broker in legal, employment, disciplinary, health, emergency, or rights-affecting proceedings during the probe","Muting, removing, or suppressing a participant without a visible rule and human override","Making a supported input modality the only path to contribution","Using allocation telemetry for individual performance evaluation","Expanding beyond the bounded sandbox without separate authorization"],"halt_rollback":"Disable automatic grants and restore ordinary platform controls if simultaneous tokens occur, a participant cannot perceive or return a grant, the broker grants to an absent or unauthorized identity, an adapter leaks contribution content, the queue deadlocks, override fails, or turn-specific state survives reset. Preserve only authorized incident evidence, discard affected timing data when consent requires it, quarantine the service version, and require accessibility, privacy, security, and host review before restart."},"negative_tests":{"strongest_counterevidence":"In matched sessions, ordinary social coordination or a human moderator yields equal or lower collision, wait, abandonment, and moderator work with no worse modality-specific access, while the broker adds formality, strategic bidding, incorrect grants, inaccessible signals, or queueing.","problem_falsifier":"Observation shows that disrupted participation is caused mainly by disagreement, status differences, inaccessible content, missing facilitation norms, or insufficient meeting time rather than recurrent recognition and floor-transition contention among already eligible contributors.","intervention_falsifier":"The broker does not reduce bid-to-completed-contribution friction under held-constant participation rules, requires moderator intervention for a substantial share of ordinary cycles, produces unacceptable wrong grants or false rejections, shifts contention into side channels, or cannot reliably release and regenerate after each turn.","risks":["Formal floor allocation may make fluid conversation slower or less socially comfortable.","Network latency or adapter differences may systematically alter bid order.","Participants may split, repeat, or misclassify bids to gain earlier access.","An urgent-clarification class may become a privileged side path.","Multimodal notifications may remain imperceptible or cognitively burdensome for some participants.","A visible queue may pressure participants to speak when their turn arrives or discourage withdrawal.","The broker may concentrate interaction power in its allocation policy or steward.","Timing telemetry may enable surveillance or interpersonal performance judgments.","A stuck token or common-mode outage may block all supported contribution channels.","Higher turn turnover may exceed the group's ability to understand and respond to contributions.","Participants may move discussion into ungoverned side channels when they dislike the allocation policy."]},"next_evidence_step":"Before the six-session probe, preregister the allocation policy, supported bid modalities, acknowledgment and release contract, accessibility checks, manual override, comparison order, measures, and stop rules. Record eligible bids, acknowledgment latency, bid-to-grant wait, overlapping contributions, overwrites, completed and abandoned bids, token returns and timeouts, wrong grants, false duplicate rejection, overrides, queue depth, repeat-turn concentration, outcomes by input modality, participant comprehension and contestability of the policy, moderator work, adapter failures, reset verification, and the group's ability to absorb each contribution. Treat the findings only as evidence about the tested task, participants, adapters, policy, and service version.","prior_art_status":"UNSEARCHED","diversity_from_prior_proposals":"Proposal 1 is retrospective and evidentiary: it compiles a user's interaction breakdown into a replay capsule for product investigation. Replacement Proposal 2 is synchronous and coordinative: it mediates live contention among already authorized contributors by granting and regenerating a temporary shared floor, producing completed contributions rather than a reconstructed artifact. Proposal 3 is individual and configurational: it translates functional preferences into reversible application settings, whereas Proposal 2 does not alter a participant's application configuration. Proposal 4 is pre-implementation research infrastructure: it instantiates hypothetical interface behavior for controlled study, whereas Proposal 2 governs real contribution timing in an existing collaborative session. The replacement also abandons the retired Proposal 2's state-reconstruction compiler and verified-packet causal path identified by the auditor.","revision_record":{"parent_version":null,"progress_targets_addressed":["Replace the auditor-rejected state-reconstruction proposal while preserving proposal index 2 and version 0","Eliminate the shared screened-state-compiler-to-verified-artifact pathway with proposal 1","Create an independently adoptable real-time CSCW coordination service distinct from proposals 1, 3, and 4","Retain explicit catalytic turnover, selectivity, capacity, regeneration, neutrality, authority, safeguards, falsifiers, and bounded evidence"],"conceptual_changes":["Replaced asynchronous digital-state reconstruction with synchronous multimodal contribution-floor mediation.","Changed the transformed substrate from case history to live content-free contribution bids.","Changed the facilitator from a state compiler to a standing cross-tool broker that grants and releases temporary interaction access.","Changed the output from a verification packet to a completed contribution and regenerated shared floor."],"operational_changes":["Introduced participant opt-in, group-selected allocation policy, multimodal adapters, visible grants, token release, moderator exceptions, and per-turn reset.","Added live saturation, deadlock, wrong-grant, accessibility, side-channel, and policy-capture safeguards."],"evidence_changes":["Replaced handoff-artifact comparison with counterbalanced collaborative sessions measuring bid-to-contribution cycles, collisions, access outcomes, moderator work, and reset integrity."],"claim_changes":["Removed the claim that compiled state reduces resumption work.","Limited the replacement claim to lowering recurrent contribution-access negotiation without judging content or changing participation rights."]}}