Research

How the simulation is checked

Three levels of checking, and what would count against each.

Illustrative simulation of the CRIS continuity framework. Not an established scientific law.

Model 1.0 · Illustration. Runs on this page. The data file, build tests and run links are in place, awaiting sign-off as a tested illustration. The model has not yet been reviewed by people outside CRIS.

In short

There are three questions. Does the code do what the page says? Is the model a fair picture of the formula? Does the idea hold against the world? This page answers the first, sets out how the second will be answered, and says why the third is a research question.

Level 1 · CodeDoes the code do what the page says?The worked runs and edge cases are build testsin place: tests pass on every buildAgainst it: A test failsLevel 2 · ModelIs the model a fair picture of the formula?Outside review; sensitivity runs on each assumptionplanned for phase P3; reviewers to name (Q6)Against it: One assumption decides most outcomesLevel 3 · IdeaDoes continuity, measured this way, predict how real systems fare?Predictions written before data; a published methodnot claimed on this pageAgainst it: Predictions fail against real data
Validation: three levels of checking, and what would count against each.

Where we are

  • Level 1, code: in place. The worked runs and edge cases are build tests, and they pass on every build. See the changelog.
  • Level 2, model: planned for phase P3, before the research pages open, with outside reviewers to be named; listed under Coming on the papers page.
  • Level 3, the idea: not claimed on this page. Any such claim belongs in a paper, with its method and data.

Sensitivity runs (level 2)

For each assumption, the review will vary it across a reasonable range and record how often the outcome changes, across all ten conditions and a grid of intelligence values: the match band at 0, 2, 5 and 10 points; the support range at ±25%, ±50% and ±75%; each modifier at double weight in turn; and no cap, to see how many runs hide a surplus. The results will go in a short note linked here. If one assumption decides most outcomes, this page will say so.

What would count against it

  • Level 1: any test case fails to reproduce.
  • Level 2: one hidden assumption decides most outcomes, or reviewers find the model misrepresents the formula.
  • Level 3: predictions written down before data is collected fail against real data.

How a claim reaches the page

1 DraftResearch andEngineering2 Tag its leveldescriptionto standing3 Team checkand build tests4 Outside reviewmodel claims only5 Sign-offFounder; counselfor standing wordsDescriptionteam checkCode claimteam check, tests passModel claimoutside review, note, FounderFramework claimFounder sign-offStanding claim (law, proven, scientific, predicts)Founder and counsel; usually a paper
Review path: five steps from a drafted claim to the page, and the five levels of claim.

Research and Engineering drafts each sentence and tags its level. Outside reviewers review model claims and publish a short note. The Founder and Chief Architect signs off model, framework and standing claims, as Steward of the Fourth Law under the proposed bylaws. Counsel reviews standing words before they appear anywhere on the site.

Back to the simulation

Owner: Research and EngineeringNext review: 2027-03-01