{"actors":["Scholars studying born-digital and interactive literature","Readers or operators whose sessions instantiate variable textual states","A digital-text curator who validates and promotes reproducible witnesses"],"affected_objective":"Preserve reconstructable evidence of transient literary states—generated passages, link paths, timing, configuration, and software context—without making readers wait for full archival description during each encounter.","arm":"ORDINARY_DIVERSE_P2","authority_safety":{"authorized_first_step":"Run a reversible capture-and-consolidation pilot on one public or permission-controlled born-digital work in a disposable local environment, without altering the work or its public archive.","decision_authority":"Readers decide whether their session traces may be retained; the digital-text curator controls promotion and correction of archival witnesses; the work's authorized files and documented execution environment remain authoritative over reconstructed outputs.","excluded_actions":["Circumventing access controls or digital-rights protections","Capturing reader identities, credentials, or unrelated device activity","Treating a generated session state as the work's single definitive text","Replacing authorized source files with reconstructed outputs","Publishing copyrighted content or session traces beyond the permissions granted","Silently repairing failed replays to resemble an expected result"],"halt_rollback":"Stop capture if consent boundaries fail, unrelated personal data enters the fast store, required execution context cannot be recorded, or the validation backlog exceeds the pilot limit; disable collection, quarantine affected traces, delete unauthorized personal data, and restore the unchanged pre-pilot environment."},"baseline":"A reader or scholar may preserve an interactive encounter through screenshots, quotations, screen recordings, or retrospective description. These records can document visible output but may omit the event sequence, configuration, random seed, asset version, or runtime conditions needed to distinguish and reconstruct the encountered textual state.","candidate_id":"fast_slow_store_coupling__literature_literary_theory__ORDINARY_DIVERSE_P2","causal_chain":["During an authorized encounter with a variable or interactive literary work, a lightweight session recorder appends locally generated textual states, reader actions, timestamps, configuration values, and declared environment identifiers to a bounded fast event store.","The recorder routes sensitive input and unrelated system activity away from capture, while preservation-critical events such as an external asset change or runtime error receive protected status.","At session end or when a protected event occurs, a transfer manifest freezes the relevant event interval before log rotation or software-state loss can erase it.","Consolidation reconstructs the event order, normalizes time and environment metadata, hashes referenced files, separates work output from reader input, and attempts a replay without silently correcting discrepancies.","A trace is promoted to the slow witness archive only when its permissions, dependencies, provenance, and replay result are recorded; non-replayable but evidentially useful traces may be retained with an explicit limitation marker.","Routine traces that duplicate an existing witness or lack preservation value expire from the fast store after review, while unresolved dependency failures remain protected only for a bounded investigation window.","Scheduled reconciliation reruns selected witnesses against their recorded environment and compares outputs, marking divergence as an environment, dependency, capture, or work-version conflict rather than overwriting the earlier witness.","Fast event capture therefore avoids interrupting the literary encounter, while governed transformation produces durable, qualified evidence from which a transient textual state can be studied or reconstructed."],"cell_id":"fast_slow_store_coupling__literature_literary_theory","consequence":"A transient textual state may survive only as an incomplete representation whose relation to reader action and execution conditions cannot be established, limiting later analysis of sequence, variability, material form, and textual boundaries.","diversity_from_prior_proposals":"Unlike P1's passage-level capture and promotion of provisional human interpretations into a claim dossier, this opportunity preserves machine- and reader-produced execution states of variable literary works in a replay-qualified witness archive. Its problem is disappearance of instantiated texts, its intervention records event and environment state, and its causal path runs through freezing, dependency normalization, replay, and divergence classification rather than interpretive validation.","experiment_id":"eoa_inverse_innovation_exp13_second_slot_policy60_20260806","intervention":"Couple a consent-bounded session event store to a durable archive of replay-qualified textual witnesses. The fast store records only the events and environment references needed to preserve an authorized interactive encounter, using protected triggers for preservation-critical changes and bounded rotation for routine data. At session end or a trigger, selected intervals undergo permission review, event reconstruction, dependency hashing, metadata normalization, reader-input separation, and replay testing before promotion. The slow archive is authoritative about preserved witnesses and their qualifications; the live event store is authoritative only about provisionally captured session events. Versioned reconciliation retains divergent outcomes instead of collapsing the work's variability into one canonical text.","mechanism_mapping":[{"counterfactual_removal":"Without asymmetric stores, each live event would require immediate archival description or the encounter would depend on fragile, unintegrated recordings.","mechanism_slug":"two_store_architecture","role":"Separates low-overhead session-event capture from the slower construction of durable textual witnesses."},{"counterfactual_removal":"Without protected transfer before rotation or environment change, the event sequence and execution context needed to reconstruct a transient state could disappear.","mechanism_slug":"temporal_decay_and_degradation","role":"Motivates session-end and preservation-critical transfers before volatile traces or dependencies are lost."},{"counterfactual_removal":"Without replay, the archive could retain plausible-looking outputs that cannot be linked reliably to the recorded event sequence and environment.","mechanism_slug":"replay","role":"Tests whether a promoted witness can reproduce or explicitly account for the captured textual state."},{"counterfactual_removal":"Without integrity checks, mismatched assets, truncated events, or altered dependencies could be mistaken for properties of the literary work.","mechanism_slug":"data_integrity","role":"Uses manifests, hashes, sequence checks, and dependency validation during consolidation."},{"counterfactual_removal":"Without versioned reconciliation, later executions could overwrite earlier witnesses or conceal meaningful divergence among work states.","mechanism_slug":"versioning","role":"Preserves work, environment, witness, and replay versions together with classified conflicts."}],"nearest_rivals":["Screenshots or copied passages: preserve visible fragments quickly but generally omit the event sequence and execution context needed for reconstruction.","Continuous screen recording: preserves audiovisual presentation but does not inherently yield structured events, dependency provenance, or a validated replay witness.","A static archival copy of source files: preserves an object version but may not preserve which conditional path, generated state, or runtime behavior a reader encountered.","A full virtual-machine image for every session: may retain extensive context but imposes heavier capture and review costs and does not by itself distinguish reader data from preservation-relevant state."],"negative_tests":{"intervention_falsifier":"The intervention is undermined if promoted witnesses are no more reconstructable than matched screenshots or recordings, if structured capture materially disrupts the encounter, or if replay qualification cannot distinguish capture faults from legitimate textual variability.","problem_falsifier":"The problem is weakened if the selected work produces a fixed, fully represented text; if an authorized static package already reconstructs every analytically relevant state without session events; or if the research question concerns only visible output and not sequence, configuration, or material execution.","risks":["Session capture may collect personal or unrelated interaction data.","Replay may create false confidence when external dependencies remain uncontrolled.","Storage and validation work may exceed the value of preserved states.","A curator may privilege reproducible states and marginalize meaningful failures, glitches, or unreproducible encounters.","Environment packaging may conflict with licenses or preservation permissions.","Event recording may alter timing-sensitive behavior.","Duplicate or low-value traces may pollute the slow archive.","Version labels may obscure rather than clarify boundaries among work, platform, and performance."],"strongest_counterevidence":"For some variable literary works, the non-reconstructable singularity of an encounter may be part of the object being studied; converting sessions into replay-qualified witnesses could misrepresent that ephemerality and add little beyond an explicitly partial recording."},"next_evidence_step":"Using one permission-controlled interactive work, define three short, repeatable reading paths and capture matched evidence with screenshots alone and with the dual-store workflow. After consolidation, ask a separate operator to reconstruct each encountered state using only the retained materials. Record capture interruption time, protected intervals transferred before rotation, metadata completeness, replay agreement and divergence classifications, personal-data exclusions, review minutes, duplicate dispositions, and backlog. End after one delayed reconciliation run and treat the results only as pilot process evidence.","observable_state":"In a bounded study of a variable literary work, successive authorized sessions can yield different passages, orders, or presentations. Immediate records show outputs but inconsistently expose the actions, dependency versions, timing, configuration, and runtime failures associated with each state; after the original environment changes, some recorded encounters cannot be confidently reconstructed or compared.","prior_art_status":"UNSEARCHED","problem":"Interactive, procedural, and environment-dependent literary works can instantiate transient textual states faster than a scholar can document their generating conditions. Full archival description during reading would interrupt or alter the encounter, while delayed reliance on screenshots or memory can lose the event sequence, dependencies, and environment needed to interpret the instantiated text as evidence.","proposal_index":2,"remaining_contrastive_claim":"Relative to screenshots, screen recordings, or static source packages, the proposal claims that a bounded event store coupled to a replay-qualified witness archive better preserves the relation among textual output, reader path, and execution environment because consolidation reconstructs sequences, validates dependencies, records permissions, and retains divergent replays. The contrast fails if simpler records support equivalent reconstruction and qualification with less intrusion and review burden.","revision_record":{"claim_changes":[],"conceptual_changes":["Initial formulation of session-event capture coupled to a durable archive of replay-qualified witnesses for transient born-digital literary states."],"evidence_changes":["No external evidence or prior-art search used; all empirical propositions remain untested."],"operational_changes":["Specified consent-based routing, protected transfer triggers, event and dependency consolidation, replay qualification, bounded expiration, versioned reconciliation, rollback, and a matched pilot."],"parent_version":null,"progress_targets_addressed":["Materially independent problem and intervention","Observable state and consequence","Complete fast-store and slow-store routing","Transfer, consolidation, eviction, authority, and reconciliation rules","Actors, permissions, safeguards, and rollback","Rivals, falsifiers, risks, and bounded evidence step"]},"schema_version":1,"structural_mapping":[{"archetype_element":"Fast Volatile Store","domain_realization":"A consent-bounded, session-local append-only store containing literary outputs, reader actions, timestamps, configuration values, runtime events, and references to active dependencies."},{"archetype_element":"Slow Integrated Store","domain_realization":"A versioned archive of textual witnesses whose event sequence, permissions, dependencies, environment, replay result, and known limitations have been integrated."},{"archetype_element":"Read/Write Routing Rule","domain_realization":"Authorized session events write to the fast store; credentials and unrelated activity bypass capture; validated manifests and witness packages write to the slow archive; archival claims require slow-store confirmation."},{"archetype_element":"Transfer Trigger or Cadence","domain_realization":"Routine transfer begins at session end, while runtime errors, external asset changes, rare branches, and impending log rotation freeze and transfer the affected interval earlier."},{"archetype_element":"Consolidation Transform","domain_realization":"Reconstruct event order, normalize metadata, hash dependencies, separate reader input from work output, check permissions, package required context, run replay, and label unresolved divergence."},{"archetype_element":"Promotion and Eviction Rule","domain_realization":"Promote traces with sufficient provenance or explicit evidentiary limitations; deduplicate equivalent witnesses; expire unprotected routine events after review; bound retention of unresolved traces."},{"archetype_element":"Interference and Overwrite Guard","domain_realization":"Protected intervals and immutable transfer manifests prevent log rotation, later sessions, or dependency updates from overwriting preservation-critical evidence."},{"archetype_element":"Freshness and Authority Marker","domain_realization":"Items are labeled live, frozen, under review, replay-qualified, divergent, superseded, restricted, or expired; only the slow archive authorizes claims about preserved witnesses."},{"archetype_element":"Consistency Reconciliation Check","domain_realization":"Scheduled replays compare stored witnesses with recorded work and environment versions, preserving earlier outcomes while classifying dependency, capture, environment, or work-version conflicts."}],"title":"Session Event Store to Replay-Qualified Literary Witness Archive","version":0}