Publication field 01
Counted unit
Define exactly what constitutes one result so scenarios, trials, and judged episodes cannot be interchanged.
Evidence and accountability
The current stage
and its boundary.
Current work has produced internal simulation evidence. No quantified public performance result is currently published.
This page sets the standards a public result must meet and separates the current internal record from stronger future gates.
Current public evidence state
No public result
is published yet.
Current work has produced internal simulation evidence. No quantified public performance result is currently published.
Current public evidence state
Boundary visibleInternal simulation
What may be said now
Current work has produced internal simulation evidence. No quantified public performance result is currently published.
Quantified-result truth lock
Quantified results will be published only after the public evidence register, counting rules, baselines, and boundaries are reconciled and approved.
What this does not establish
Internal simulation does not establish independent review, representative-system validation, paid customer proof, deployment, field performance, customer economics, or recurring revenue.
Quantified-result truth lockQuantified results will be published only after the public evidence register, counting rules, baselines, and boundaries are reconciled and approved.
Mandatory public boundaryInternal simulation does not establish independent review, representative-system validation, paid customer proof, deployment, field performance, customer economics, or recurring revenue.
Publication contract
A result appears only
when one record resolves.
A result will appear here only when its counted unit, success criterion, held-out condition, baseline, source, version, date, and evidence boundary resolve to one approved public record.
Publication field 02
Locked success criterion
State the pass condition before reporting the result, including how abstentions, exceptions, and corrections are treated.
Publication field 03
Held-out condition
Identify what was withheld from development and how that separation was preserved.
Publication field 04
Baseline class
Name the comparison logic, its information access, and whether the comparison is like-for-like.
Publication field 05
Source, version, and date
Attach the artifact, code or configuration state, run receipt, and review date needed to find the same record again.
Publication field 06
Internal simulation
Label the evidence level directly and keep stronger gates visibly separate.
Record integrityAll publication fields must support the same claim under the same stated conditions.
Proof gates
One gate does not prove
the next.
Each state supports a different public statement. Progress requires new evidence; it is not inherited from the gate before it.
Evidence gate
Internal simulation
A defined method produced a scoped result in a simulated environment. This does not establish representative-system or field performance.
Future gate
Independent review
A named independent process examines the scoped record and states what it reviewed. This is not yet established.
Future gate
Representative-system validation
A relevant system and context produce their own bounded evidence. This is not yet established.
Future gates
Customer, deployment, and commercial proof
Paid work, live operation, field performance, customer economics, and recurring revenue each require separate evidence. They are not yet established.
Status disciplineEvery public statement stops at the strongest gate its record actually supports.
Publication readiness ledger
Open truth locks stay
open in public.
These are the work items that must close before quantified results can be promoted into public-facing claims. Their presence is not a claim that the underlying result has passed publication review.
Open truth lock
Counting contract
Reconcile what is counted.
Align scenario-level wins, trials, and judged episodes so the public unit cannot be misread.
Publication gate
Source-backed definition approved.
Required before a portfolio total or representative ratio appears publicly.
Open truth lock
Comparison contract
Reconcile baselines and success rules.
Confirm information access, frozen criteria, exception handling, and like-for-like comparison.
Publication gate
Comparison method approved.
Required before a relative performance statement appears publicly.
Open truth lock
Source record
Attach provenance and caveats.
Connect every proposed number to its source, version, date, conditions, and current review state.
Publication gate
Public evidence register approved.
Required before the site presents a quantified evidence ledger.
How to read the evidence
Read what is shown.
Stop at the boundary.
Evidence supports a specific statement under specific conditions. A desirable conclusion cannot silently promote it into a stronger state.
Evidence state
Internal simulation
A result produced in a defined simulated environment.
≠
Separate proof required
Representative-system or field performance
Relevant hardware, integration, and live operation require their own evidence.
Evidence state
Internal reproduction
A defined result reproduced by an internal process.
≠
Separate proof required
Independent review
Independence requires a named outside process and a stated review scope.
Evidence state
Recorded outcome
A record of what occurred under stated conditions.
≠
Separate proof required
Proof the decision was correct
A trace does not, by itself, establish the quality of the choice.
Evidence state
One scoped result
A finding bounded to its system, method, baseline, and conditions.
≠
Separate proof required
General performance
Other systems, conditions, and domains require separate evaluation.
The reading ruleState the strongest claim directly supported by the record—then stop.
Bring the exact claim
Discuss the standards.
Then the next gate.
An evidence discussion begins with the intended claim, the source record, the evaluation rules, and the next gate that matters.
Discuss evidence requirementsBegin with the system you have.