Back to explore
Autorite / Open-ended research program

What does a system preserve as it changes?

Autorite is an open-ended program studying representation, realized history, preservation and reachability through explicitly scoped research packages. Follow the path from questions and methods to experiment records and cycle publications.

Work statusTwo cycles published · new challenge in preparationSources checked
Open laboratory
01

The starting question

Can two apparently identical states be treated alike if they come from different histories? Which distinctions must a representation preserve to answer a question? Autorite begins before a ready-made architecture: it specifies the scope, the observations that could test the question and what a result would change.

Question and scope first.
Preservation, representation and history are not explanations by themselves. Specify the system, transformation and inquiry being discussed.
A result carries its limits.
Assumptions, counterexamples, negative and indeterminate outcomes remain on record. Completing an experiment is not acceptance of a broad claim.
An open program, closable cycles.
A cycle closes by publishing its results, synthesis, open questions and frozen knowledge state. New questions emerge from that record.

Six connected questions

Q0

Preservation and continuation

Which properties of states, paths or lineages must survive for specified continuation criteria?

Q1

Revision

What happens when rules, criteria or representations themselves change?

Q2

Reachability

Which targets are reachable under specified rules, resources and horizons?

Q3

Representation

Which distinctions must remain visible for a particular inquiry or action?

Q4

Realized history

Which distinctions in past transformation structure must remain recoverable?

Q5

Relational dependence

When does relational information add explanatory or functional value beyond local properties?

02

From question to method

How a question becomes a research record

  1. 01

    Bound the question

    Declare the system, transformations, inquiries and assumptions. Keep the question distinct from a testable package.

  2. 02

    Specify methods and tests

    Record the protocol, comparisons, controls, source version and execution conditions before seeing outcomes.

  3. 03

    Recheck the result

    Connect artifacts, references and replay records. Validating the record does not automatically establish scientific truth.

  4. 04

    Decide and publish

    Keep result, interpretation, decision and cycle closure distinct. Publish deviations and unresolved issues too.

This is a reading guide; it starts no experiment. Nor does an edge in the research graph automatically transmit scientific support through other edges.

When questions become concrete

If every small step is acceptable?
Change accumulated through nearby states can still exceed the starting criterion. RP003 separates local checks from checks over an entire path.
Same state, different lineage
Copying, branching and merging can yield the same present appearance. RP004 and RP006 examine which lineage distinctions particular inquiries require.
Can a constraint protect future options?
Some short-horizon decisions can reduce access to viable targets. RP005A separates conditional benefits of constraints from claims of expanded raw reachability.
03

Packages and cycles

The first cycle combines four finite research packages. The second examines lineage representations relative to inquiries. These summaries lead to the public package records; each result retains its own scope.

Five research packages

RP002AClosed within declared scope

When history helps

Compares current observation, finite history, learned recurrent state and an oracle with generator access.

Bound to fixed algorithms and generator settings; does not establish universal superiority or necessity of history.

RP003Closed within declared scope

From local to path-wide preservation

Compares local continuation criteria with endpoint, prefix and accumulated-variation criteria.

Does not establish a metaphysical identity theory beyond its finite scope.

RP004Closed within declared scope

Branching lineages

Tests which inquiries representations can distinguish under copies, forks, independent duplicates and merges.

Present-state equivalence does not determine lineage; known-example results are not universal sufficiency.

RP005AClosed within declared scope

Constraints and viable futures

Examines the relation between decision restrictions and target access over declared horizons.

Pure edge removal never expands structural reachability. Observed benefits are conditional, and negative results remain.

RP006Closed within declared scope

Query-relative lineage sufficiency

Classifies eight whole-field masks for two exact inquiry families on six known constructions.

Reproduces known finite examples; establishes neither unseen-domain sufficiency nor bit-optimal representation.

Two completed cycles

Cycle 0113 / 13

Foundations under test

Results from four packages, deviations, synthesis and open questions remain together in the published close record.

Cycle 029 / 9

Query-relative lineage sufficiency

The finite RP006 result and related reviews enter the close record; foundational questions are not thereby resolved.

These ratios track publication and review obligations, not accuracy percentages or completion of a general theory.

04

Related writing

Open the series trail

Autorite: question, philosophy, method, application

A four-chapter reading series, from a changing system to an investigation that tests the question.

05

Docs and links

06

Status and open issues

As checked on 22 October 2026, the public publication and laboratory records show Cycle 01 published at 13/13 obligations and Cycle 02 at 9/9. Five packages are closed within their declared scopes. The program remains open-ended; closing a package does not resolve its foundational questions.

RP006 is limited to six previously known constructions and the declared field-mask class. Cycle 03 has not opened. Preparation, synthetic engineering checks and supplemental design review for a new finite-lineage challenge are not candidate research data or new scientific results. The public record also leaves the external reviewer’s identity and institutional verification unresolved.

Package closed ≠ foundational question resolved. Software check passed ≠ scientific claim confirmed. Sufficient on finite examples ≠ sufficient in every domain.

This summary reads both official sites as accessed on 22 October together with the pinned public source version. Their latest change entry is dated 21 October. The current shared record and cycle publications take precedence over the older README introduction. No new experiment was run. Source documents are in English.

Existing project text and dated sources

The editable project record below preserves the earlier introduction dated 13 October. Current cycle and preparation status is explained above; older source links retain their historical context.

What does a system preserve as it changes? Can two apparently identical states be treated alike when they come from different histories? Which distinctions can a representation lose before it stops being sufficient?

Autorite is an open-ended research program addressing these questions through explicit scopes, comparable research packages and traceable evidence records. Its starting point is to determine which question can be tested under which conditions, before committing to an architecture.

Lines of inquiry

  • Preservation and continuation: Which feature, relationship or lineage must survive a specified transformation? Does local preservation suffice over an entire path?
  • Representation: Which distinctions matter for a particular inquiry or action? The same representation may be sufficient for one question and insufficient for another.
  • Reachability: Which targets remain reachable from a given source under specified resources, rules and horizons?
  • Realized history: When is the present state sufficient, and when do distinctions between past transformations add information?
  • Revision and relational dependence: What happens when rules themselves become objects of change? When do relationships add explanatory value beyond local properties?

The independence and completeness of this question set are not assumed. Nor is an ontology such as “everything is relational” protected in advance.

The first cycle: foundations under test

Cycle 01 brings together four research packages. RP002A examines the boundaries of history usefulness; RP003 studies local and path-wide preservation; RP004 addresses lineage distinctions such as copying, branching and merging; RP005A examines constraints, reachability and viable targets.

RP002A compares current observation, finite history, a learned recursive state and an oracle with generator access. The aim is not to establish that history-using systems always win. Conditions in which the present suffices, history adds information, or variable history does not inform the future are constructed separately.

How a result enters the record

A claim retains its scope, assumptions, evidence, counterevidence and observations that could change it. Executing an experiment and accepting a scientific result are separate events. Negative outcomes and unresolved issues also belong in the research record.

The program is open-ended; its cycles are finite and closable. A closed cycle publishes results, synthesis, open questions and a frozen knowledge-state snapshot. Evidence from that cycle generates questions for the next one.

Status and limits

The current public record reviewed on 13 October 2026 shows Cycle 01 published at 13/13 and Cycle 02 at 9/9 obligations. Cycle 02 examines query-relative lineage sufficiency through RP006 within six known constructions and a declared field-mask class. This closure does not establish sufficiency in unseen domains or a general theory. Cycle 03 has not opened; the new finite-lineage challenge remains a preparation draft.

Explore the sources

Sources and revision

Current status checked against the public research record on 13 October 2026. Historical sources below preserve the basis of the original introduction.

10 October 2026 · historical introductory sources

Public repository revision inspected on 10 October 2026. These links are pinned to the same commit; later research developments are outside this dated account.

07

Atlas connections