Uncommon Reality - Hiatus Improvement and Independent Continuation Programme

Disclosure, evidence basis, and authority

This programme is project-authored. Its expert and advisor voices are simulated intellectual standpoints used to expose competing judgments. No named thinker participated, consented, spoke, reviewed the project, or endorsed this programme. Simulated convergence is not evidence, peer review, affected-party participation, accessibility review, or authority to release anything.

Sources are the Forum Improvement report, RPT63, RPT67, RPT68, the Roadmap, *Where We Stopped*, *Continuing the Work*, *Governance*, and the Project Intensity and Diminishing Returns evaluation. The released baseline is v1.9.2; inspected v2.0.0 records remain candidates. Later verified GitHub and release evidence supersedes this draft.

This document recommends a posture. It does not authorize a merge, deployment, repository creation, visibility change, correction window, archival action, research claim, or expenditure. Those remain separate owner or independent-fork decisions.

Executive judgment: the hiatus should begin intellectually now

Uncommon Reality should enter an intellectual hiatus immediately, while permitting only the finite work required to close, audit, publish, and safely hand off the v2.0 author-led reference. “Hiatus” does not mean that the project was solved, abandoned in shame, or declared permanently complete. It means that the current production loop (more synthesis, more simulated deliberation, more reports, more versions, more schemas, and more internal verification) has ceased to be the best route to knowledge.

The proposed final status remains:

Author-completed / research-open / independently continuable / not validated.

The hiatus separates four things that the project previously allowed to blur:

1. finite reference closure: complete only the bounded v2.0 release and public-safe continuation surface;
2. custodial care: preserve bytes, links, rights, security, privacy, and correction pathways;
3. independent research: allow others to test, narrow, split, rename, reject, or replace the framework in forks;
4. future author reactivation: possible, but not promised and not triggered by the mere availability of models, tools, or new language.

The project’s governance, publication, and correction machinery became more mature than its prospective evidence. The website is a useful static reference but not a demonstrated public experience. The theories remain provisional and cannot yet support confident cross-domain application or policy use. The next legitimate gains would come from external reality: a locked test, a serious rival, a domain critic, an affected-party challenge, an accessibility finding, or observed reader behaviour. The author-led system should not simulate those gains internally.

The aim is narrow: leave a safe, intelligible reference, reduce maintenance nearly to zero, and enable continuation without making the original author its permanent coordinator.

1. The hiatus posture

1.1 Effective immediately: stop intellectual expansion

From the point this posture is adopted, the author-led project should stop:

• adding master theories, modules, umbrella diagnoses, conceptual domains, or memorable labels;
• generating further general forums, persona panels, simulated reader studies, or advisor rounds about the project as a whole;
• creating reports whose principal subject is the existence, management, maturity, or improvement of other project reports;
• adding versions merely because a bounded change can be published;
• expanding schemas, registries, skills, dashboards, metrics, diagrams, or routes without a demonstrated external need;
• deepening T001, cases, or claims as an implicit obligation of finalization;
• treating more sources, passing tests, polished PDFs, repository visibility, or model consensus as research progress;
• redesigning the website in the absence of observed reader problems and a bounded product brief; and
• reopening settled naming or architecture unless new evidence makes the current structure actively misleading.

This freeze does not block safety work. A credential, personal-information exposure, rights problem, serious accessibility defect, corrupt immutable artifact, or broken preservation path may require immediate action. Such work is custodial correction, not a resumed theory programme.

1.2 Bounded closure work may continue

Only the following author-led work remains in scope before formal archival hiatus:

1. complete the v2.0 candidate’s exact source, workflow, dependency, privacy, security, rights, binary, website, and artifact gates;
2. obtain explicit authorization before any merge and deployment;
3. verify the deployed stopping, continuation, edition, identity, sharing, and private-boundary surfaces;
4. reconcile mutable release-state records without changing the theory or overwriting immutable editions;
5. create a new clean-history successor repository as private, never by changing the visibility of the permanent operating repository;
6. audit the successor’s exact tree, entire new history, metadata, settings, permissions, release surface, and anonymous view while it remains private;
7. obtain a separate explicit authorization before successor visibility changes;
8. run the actual dated 30-day correction window only after public visibility is verified; and
9. archive the reference only through a separate recorded action after the correction window.

The closure list is exhaustive. A new attractive idea does not enter it. If closure discovers a substantive theory problem, record it as an open issue for forks; do not expand the author-led release to solve it unless leaving it uncorrected would materially mislead readers.

1.3 Formal archival hiatus

After the correction window and a separately verified archival action, the official author-led reference should become read-only in ordinary circumstances. The website may remain available. The clean successor may remain forkable. The private canonical repository remains sealed as the source, audit, and provenance record. Reopening should be exceptional and limited to:

• a serious security or personal-data exposure;
• a rights, consent, attribution, or legal preservation problem;
• corruption or disappearance of official public artifacts;
• a high-severity accessibility defect whose repair can be made without rewriting substantive claims; or
• an owner decision to issue a clearly governed successor reference in response to genuinely new external evidence.

Routine dependency updates, fashionable redesign, new model capability, link cosmetics, or another internal critique are not sufficient reactivation triggers.

2. Simulated standpoint review and dissent

The following positions are reasoning functions, not biographical claims about the named people.

Donella Meadows standpoint: interrupt the reinforcing loop

The simulated Meadows function sees a reinforcing loop: more project surface creates more state to reconcile; more reconciliation creates more tests and records; more records enlarge context; larger context creates more drift risk; drift risk then justifies more governance. The intervention is not another dashboard. It is a rule change: no new author-led output without an external learning event and a predeclared stop condition.

This standpoint supports a small public continuation corpus and one bounded empirical test as the most valuable possible future work. It opposes treating the whole private build system as necessary for continuation.

Hannah Arendt standpoint: preserve a public record without administering the future

The simulated Arendt function supports a clear stopping statement, public access, plurality of forks, and a refusal to turn “shared reality” into centralized truth authority. It warns that an official maintainer could continue to govern what counts as legitimate inquiry after the author-led phase ends. The reference should therefore state its limits and make dissent possible, while independent continuations remain free to reject its vocabulary.

This standpoint supports an archive with visible provenance. It opposes any wording that makes the author-completed edition the final authority over a living political or social reality.

Karl Marx standpoint: do not let procedural repair conceal power

The simulated Marx function argues that the project’s strongest rival remains underdeveloped. Delay may be profitable, organized, and imposed rather than a system’s incapacity to repair. A future continuation that merely fills Repair-Lag fields will reproduce the framework’s assumptions. At least one independent track should begin from ownership, accumulation, class, rents, and organized obstruction, then ask whether Repair-Lag adds anything.

This standpoint dissents from making T001 the automatic flagship. It would permit retirement or splitting of the theory before further infrastructure is built.

Elinor Ostrom standpoint: make continuation possible with minimum rules

The simulated Ostrom function supports clear boundaries, licenses, attribution, correction rules, and a defined custodian. It rejects both extremes: an ungoverned repository that implies endorsement of every contribution, and a centrally managed commons that requires the original author to arbitrate all future research. Forks are the appropriate unit for substantive work; the official reference needs only a small correction ferry for safety, rights, preservation, and reproducible accessibility defects.

Ursula K. Le Guin and Donna Haraway standpoints: a technically clean archive can still exclude

These simulated functions warn that a polished handoff may privilege those comfortable with repositories, English-language theory, dense PDFs, and formal schemas. A public repo is not the same as meaningful access. Independent continuation should welcome narrative, situated, multilingual, community-led, and non-technical forms, provided modifications and authority claims are honest.

They dissent from archiving immediately if doing so freezes avoidable access barriers. Their preferred condition is one concise orientation, usable HTML, a plainly disclosed PDF status, and a route by which affected people can contest framing without learning the internal ontology.

Michel Foucault standpoint: audit can become its own regime of truth

The simulated Foucault function accepts privacy and security gates but warns that audit volume can make the project appear authoritative. What is easiest to verify (checksums, fields, routes, and test counts) can displace what matters but remains unsettled: who defines harm, whose knowledge counts, and which institutions benefit. The hiatus should preserve the audits as bounded technical evidence, not as proof of intellectual legitimacy.

Security and preservation advisor function: do not confuse urgency with publicity

This simulated technical function insists that the existing private repository must never be made public simply to save time. Clean current files do not clean Git history, metadata, issues, releases, settings, integrations, or access records. A successor must be created privately from the fail-closed export and inspected anonymously only after every blocking finding is fixed.

Recorded disagreements and dispositions

No dissent is resolved by vote. Each identifies a cost that future custodians or contributors should revisit when real evidence exists.

3. What must be preserved

The hiatus should reduce activity without erasing what was learned. Preserve:

3.1 Intellectual record

• the bounded thesis and the distinction among Theory D, lens E, modules F/G, and the A/B/C ancestors;
• the status of every claim, case, evidence edge, gap, rival, and not_yet_defined field;
• the record that T001 is design_stub / infrastructure_only / not_inspected;
• rejected, narrowed, split, or superseded claims and the reasons for those changes;
• the ethical boundaries concerning reproductive agency, defensive synthetic-media analysis, political dissent, pluralism, rights, and non-restorative repair; and
• the explicit possibility that Repair-Lag adds little, applies only in narrow domains, or should be retired.

3.2 Public and archival record

• stable website routes and truthful orientation;
• immutable published editions, checksums, lineage, and current/superseded/withdrawn dispositions;
• the report/narration distinction and accessibility disclosures;
• applicable source citations, pinpoints, limitations, licenses, attribution, and third-party notices;
• a tombstone rather than silent deletion if a public artifact must be withdrawn for safety or rights; and
• the clean public manifest without private source paths or sealed lineage.

3.3 Private record

• the private canonical repository, complete source history, build machinery, audit evidence, and private crosswalk;
• credential-safe incident and correction records;
• exact release and deployment evidence; and
• the ability to reproduce a bounded public correction without importing untrusted fork history.

Private preservation is not public entitlement. The private record remains sealed unless the owner later approves a separately audited disclosure.

4. Minimum viable independent continuation tracks

No continuation needs to adopt all tracks. A credible cycle should choose one primary track and at most one supporting track. Each track must begin with a question, boundary, resource cap, evidence need, and stop rule.

Track A - Prospective discrimination

Purpose: determine whether Repair-Lag predicts anything different from a serious rival in one bounded episode.

Minimum viable work:

• choose one claim and one domain before inspecting the outcome used for evaluation;
• define unit, period, pressure, repair process, lag stage/type, authority, adequacy criterion, indicators, threshold, missing-data rule, and revision consequence;
• state the strongest rival in terms its advocate could accept;
• obtain at least one real domain or methods review of the design;
• lock the design and amendment process before outcome inspection; and
• separate design lock, execution, result, evidence transition, and claim-state decision.

Success: the test discriminates, even if the result weakens or rejects Repair-Lag. Stop: no accessible data, no defensible criterion, outcome already inspected without a recoverable prospective boundary, unresolved rights risk, or no observation that could favour the rival.

Track B - Rival-first case triangle

Purpose: test whether the framework survives comparison with failure, success, and boundary cases selected without fitting the vocabulary after the fact.

Minimum viable work:

• one domain only;
• predeclared selection rules for one failure, one successful repair, and one boundary or ambiguous case;
• source pinpoints, dates, contradiction notes, independence clusters, and freshness status;
• a political-economy or domain-specific account written before the Repair-Lag comparison; and
• an explicit outcome: incremental value, narrowing, split, or retirement.

Success: the comparison changes at least one substantive conclusion or demonstrates why it should not. Stop: cases are incomparable, selected because outcomes fit, or the work becomes another descriptive dossier without a decision rule.

Track C - Real review and legitimacy

Purpose: replace simulated absence checks with bounded real scrutiny.

Minimum viable work:

• define the review question and what authority the reviewer does and does not receive;
• recruit one relevant domain/method reviewer and, where the work concerns people or rights, one affected-party or rights review function;
• record consent, attribution preference, conflicts, compensation where appropriate, response rights, and dissent;
• publish the project’s response without converting review into endorsement; and
• change a claim, criterion, method, or stopping decision if the criticism warrants it.

Success: a real criticism produces a traceable decision, including rejection of the project’s framing. Stop: recruitment is decorative, the requested labour is extractive, confidentiality cannot be protected, or the project has no resources to respond responsibly.

Track D - Reader and accessibility evidence

Purpose: learn how people actually encounter the website and editions.

Minimum viable work:

• five bounded tasks across a small, relevant participant set rather than another simulated audience;
• at least one real assistive-technology session if accessibility is in scope;
• narrow-mobile, keyboard, 200%/400% zoom, forced-colors, native-sharing, and operating-system print checks where relevant;
• observed task completion, errors, comprehension, and participant language; and
• a ranked repair list limited to demonstrated barriers.

Success: readers can find the thesis, inspect one claim and rival, locate a limitation, and identify how to continue or reject the work; alternatively, the project learns precisely why they cannot. Stop: testing is used to market the theory, participant diversity is claimed from a tiny convenience sample, or redesign begins before findings are recorded.

Track E - Accessible successor edition

Purpose: create an edition that improves access without overwriting immutable predecessors.

Minimum viable work:

• start from Track D findings or a documented accessibility-professional review;
• choose accessible HTML, tagged PDF/PDF-UA work, EPUB, structured audio, or another justified format;
• preserve heading structure, reading order, tables, figures, alternatives, citations, and navigation;
• issue a new immutable identity and lineage row; and
• test with actual assistive technology before making a conformance claim.

Success: a defined group can access materially equivalent content more reliably. Stop: the format is called equivalent or certified based only on automated checks or selectable text.

Track F - Rival, split, or retirement

Purpose: make rejection a first-class research result.

Minimum viable work:

• restate the strongest contribution without project branding;
• identify propositions that are metaphor, lens, mechanism, or empirical claim;
• compare them with at least two rivals without allowing Repair-Lag to absorb those rivals by definition;
• specify which observations make the framework redundant; and
• publish a retirement, split, or narrowed-scope proposal if warranted.

Success: the intellectual map becomes clearer, even if the master frame disappears. Stop: the exercise only renames existing parts or preserves the brand regardless of outcome.

Track G - Preservation and security custody

Purpose: keep the reference safe and available without resuming research.

Minimum viable work is limited to reproducible security, privacy, rights, attribution, preservation, broken-link, checksum, and accessibility corrections allowed by Governance. It may not alter theory claims, evidence status, or case verdicts.

5. Priority order during hiatus

When resources are scarce, use this order:

1. P0 - Safety, privacy, personal data, rights, and preservation. Fix or withdraw harmful exposure first.
2. P1 - Empirical discrimination and rival explanation. Prefer one result capable of changing a conclusion over more conceptual coverage.
3. P1 - Real legitimacy review. High-risk claims cannot mature through simulated scrutiny.
4. P2 - Reader and accessibility evidence. Observe barriers before redesigning or certifying formats.
5. P3 - Source pinpoints and case depth tied to a selected test. Do not expand the whole bibliography.
6. P4 - Presentation and convenience. Improve only in response to observed need.
7. Do not prioritize: more process infrastructure, panels, versions, metrics, or publication machinery.

Evidence, legitimacy, and reader access are not interchangeable. A usability success does not validate the theory. A domain review does not establish accessibility. A secure repository does not confer rights legitimacy.

6. Resource caps and anti-recurrence controls

These defaults are intentionally restrictive. An independent fork may choose different limits, but it should declare them before work begins.

6.1 Author hiatus cap

• no scheduled author research cycle;
• no standing model-agent loop;
• custodial triage only for a reproducible P0 publication issue;
• one decision note per incident, not a new forum or report family; and
• any substantive reactivation requires a new explicit brief, budget, and evidence trigger.

6.2 Independent pilot cap

For one continuation cycle:

• one primary question, one domain, one main claim, and at most one supporting track;
• six weeks or forty human-hours, whichever limit is reached first, unless extended by a recorded decision;
• one compact orientation packet plus the track-specific evidence packet, not the entire private project history;
• at most two parallel model agents and one synthesis pass;
• at most one simulated adversarial review after the first draft;
• at most one new schema, and only if an existing plain table cannot represent a required decision;
• one release candidate and one correction pass, not a sequence of micro-versions; and
• no new public route unless a real-reader or access need requires it.

Model-token or monetary ceilings vary by platform, so each cycle should state its own hard ceiling before starting. Reaching that ceiling is a stop event, not a reason to open an unbounded autonomous loop.

6.3 Evidence-to-overhead rule

At least half of a continuation cycle’s effort should produce or inspect external evidence, real review, or observed reader/accessibility behaviour. If governance, formatting, context recovery, and verification exceed half for two consecutive checkpoints, suspend the cycle and reduce scope.

6.4 Output cap

One cycle may produce:

• one decision-ready research or review report;
• one small public-facing summary if needed; and
• one data or method appendix.

Narration, PDF, HTML, and repository forms should be created only where a real audience needs them. Format multiplication is not completeness.

7. Decision gates

Gate H0 - Hiatus adoption

Record the intellectual freeze, permitted closure scope, owner authority, and non-commitment to future research. No public claim changes.

Gate H1 - v2 reference closure

Require the exact reviewed head, complete required CI, source and artifact audits, immutable-edition verification, explicit merge/deployment authorization, deployment success, scoped live checks, and release-state reconciliation. If a gate fails, fix only the defect; do not broaden the release.

Gate H2 - Private clean successor

Create a new empty repository as private. Resolve release placeholders from verified events. Import only the approved fail-closed export with new clean history. Audit files, binary payloads, Git objects, refs, identities, messages, settings, integrations, access grants, security features, releases, and disabled/inactive execution surfaces. The intended successor is a continuation corpus, not a public copy of the private build system: current plans exclude workflows, executable package, generator, tests, dependency lock, private crosswalk, and private ancestry.

Gate H3 - Visibility decision

Present the exact private successor and residual risks to the owner. Visibility requires a separate explicit authorization. After the change, verify anonymous browse, clone, generated archive, release download, links, licenses, security reporting, and fork surface. A pre-visibility audit cannot substitute for this post-change observation.

Gate H4 - Correction and archive

Begin the 30-day correction window from the actual verified public date. Accept only eligible bounded corrections. Use no more than two release batches unless a P0 issue requires another. At the end, record unresolved items, close the window, and archive only through a separate verified action.

Gate C0 - Independent continuation proposal

State question, track, scope, resources, evidence, ethics, authority, strongest rival, outputs, success, and stop criteria. A fork may proceed independently without owner approval, but it may not claim official status or endorsement.

Gate C1 - Design lock or review protocol

For empirical work, lock the design before outcome inspection. For review work, lock participant scope, consent, data handling, and response rights before recruitment. For accessibility work, lock tasks and severity rules before testing.

Gate C2 - Result and adversarial check

Separate data/results from interpretation. Give the rival a fair reading. Record missing data, deviations, dissent, and negative outcomes. Do not add a framework-saving explanation after seeing the result without labelling it exploratory.

Gate C3 - Claim transition

No result automatically promotes a claim. A separate decision must consider study quality, independence, generalizability, affected-party implications, and replication needs. Split, downgrade, or retirement are valid outcomes.

8. Success and stop criteria

Hiatus success

The hiatus succeeds when:

• the v2 reference is either safely released and boundedly verified or honestly left unreleased with reasons;
• the permanent private repository remains private;
• any public successor contains only the audited clean-history corpus;
• readers can understand what was built, what remains unproved, and how forks differ from the official reference;
• correction and security paths exist without reopening general authorship;
• maintenance approaches zero after the correction window; and
• no new author-led output is produced merely to demonstrate activity.

Continuation success

A continuation succeeds if it changes the evidence state, sharpens a rival, reveals an access or legitimacy failure, improves a bounded reader outcome, or supplies a credible reason to narrow or reject the framework. Success does not require a positive result for Repair-Lag.

Mandatory stop criteria

Stop or rescope when any of the following occurs:

• the predeclared time, human-effort, model, or money cap is reached;
• two consecutive checkpoints add no external evidence, real review, observed user result, or substantive decision change;
• the research question cannot identify an observation that favours a rival;
• required data, consent, rights, domain competence, or affected-party safeguards are unavailable;
• governance and formatting consume more than half the cycle twice;
• the work requires exposing private operating material;
• contributors begin treating the public reference as an endorsement authority;
• a result is being protected by adding post hoc concepts or changing thresholds; or
• the principal motive becomes preserving the project’s name rather than learning.

Stopping under these conditions is not failure. It is the programme working as intended.

9. Public-repository succession boundary

The succession boundary must remain unambiguous:

1. The permanent canonical repository stays private. A clean current tree is not proof that its history, metadata, issue surfaces, settings, or integrations are public-safe.
2. The public candidate is a new repository with clean history. It begins private and receives only the approved export.
3. The public manifest is deliberately minimal. It may disclose public paths, sizes, checksums, media types, license scopes, profile, and edition, but not private source paths, baseline identifiers, internal dispositions, or the sealed crosswalk.
4. The successor is not the complete build environment. It is an independently readable and forkable continuation corpus. The private generator and audit machinery are not promised as public reproducibility unless separately approved and audited later.
5. Substantive work happens in forks. The official successor accepts only bounded correction-window work. Fork maintainers choose their own methods and governance.
6. Forks must identify modification and independence. They may rename, split, reject, or replace the framework, but may not imply Hulki Okan Tabak’s endorsement or real participation by simulated standpoints.
7. The correction ferry transfers fixes, not history. An accepted public correction is independently reproduced in private source, rebuilt, and re-audited. Untrusted fork commits are not pulled into the sealed repository.
8. Public visibility is an observed state, not a line in a document. Repository metadata and anonymous verification govern that fact.

This design sacrifices full build transparency to protect private history and reduce the inherited maintenance burden. That trade-off should be stated plainly. Independent researchers can continue the ideas and public data; they are not receiving a turnkey clone of the author’s complete operating system.

10. What the hiatus and successor must not claim

Do not claim that:

• the project solved the repair crisis or reached definitive closure;
• Repair-Lag, Theory D, lens E, or modules F/G are validated general theories;
• the framework is ready for policy application merely because it is ethically constrained;
• T001 is preregistered, executed, inspected, or evidentially meaningful;
• simulated experts, advisors, readers, or named historical figures participated or endorsed the work;
• source volume, typed edges, case completeness, test counts, schemas, dashboards, version history, or public visibility demonstrate truth;
• the website has been proven usable, accessible, comprehensible, or effective for real audiences;
• selectable text, narration, or HTML companions establish tagged-PDF or equivalent accessibility;
• technical scans constitute a penetration test, privacy certification, legal clearance, or complete supply-chain assurance;
• the public successor is the original private repository, contains its complete history, or reproduces the full private build;
• a fork is official, endorsed, or obliged to preserve the framework;
• author completion creates scientific authority over independent work; or
• archival status prevents future criticism, correction, rejection, or rediscovery.

The strongest truthful claim is smaller: the author-led project seriously formulated a problem, built a provisional and inspectable approach, documented important failures and limits, published a bounded reference, and left conditions under which others may continue or reject it.

11. Minimum public hiatus package

The final reader should not need to master the project’s operating history. The minimum package is:

1. a concise homepage stating the thesis, provisional status, and stopping point;
2. *Where We Stopped*;
3. *Continuing the Work*;
4. the v2 stopping report and distinct narration companion, with accessibility status;
5. one clear route to claims, rivals, cases, evidence gaps, and revisions;
6. licenses, citation, provenance, governance, security-reporting, and attribution information;
7. the clean manifest and checksums necessary to inspect the public corpus; and
8. a visible statement that substantive continuation occurs independently in forks.

Everything else may remain available as historical depth, but it should not compete with this orientation. A repository is not the homepage, and an archive is not a reading path.

12. Final programme disposition

The simulated forum recommends:

1. adopt the intellectual hiatus now;
2. complete only the finite v2 closure and safe-succession work already authorized through explicit future gates;
3. preserve the private canonical repository and immutable public record;
4. make the clean successor public only after private audit and separate owner authorization;
5. limit the correction window to safety, rights, preservation, reproducibility, and concrete accessibility defects;
6. move all substantive theory and research into independent forks;
7. treat prospective discrimination, rival-first comparison, real review, reader/accessibility evidence, and possible theory retirement as the highest-value continuation tracks;
8. impose small, predeclared resource and output caps on every future cycle;
9. reactivate author-led work only in response to a genuine external learning event or serious custodial need; and
10. treat a well-kept hiatus, rather than continuing production, as the final proof that the project learned from its diminishing returns.

Closing position

The earlier Improvement Programme argued that ambition must pay rent. The hiatus is the point at which that principle becomes operational. Uncommon Reality has already extracted most of the value available from internally generated synthesis, governance, and self-correction. More of the same would likely improve the apparatus while leaving applicability, legitimacy, and reader experience substantially unchanged.

The appropriate next move is quiet but not empty: finish the reference honestly, protect people and private material, preserve what is useful, make independent continuation possible, and stop manufacturing internal momentum. If future work returns, it should return because the world answered back, not because the project found another way to talk to itself.