# Project 01 materials: synthetic check

Illustrative resources from the Intuitionist Edu Project 01 demonstration
(https://intuitionist-edu.lampbotics.com/demo-project1/).

| File | What it is |
|---|---|
| `protocol.md` | An example protocol written for public reading |
| `fixtures.json` | Five made-up sets of ordered-outcome model parameters, each with its known true values |
| `check.py` | Three hand-written readers and the check they run (Python 3, standard library only) |

Run:

    python3 check.py

Expected: the positional reader raises a false alarm on correct fits stored as
log-gaps and passes a reordered export with the wrong slopes; the always-convert
reader distorts values that were already cut points and misses a real ordering
problem; the representation-aware reader reads every case correctly, catches
the real problem, and stops when the stored form is not stated.

## What this is not

- Not an experiment on AI agents. The readers are fixed code written by hand.
- Not study data. Every value is made up to reproduce one methodological point.
- Not evidence that the practice improves research quality. That is the
  question the planned evaluation asks.

The browser demonstration runs the same logic (`../check.js`). Checked against
statsmodels 0.15.0, `OrderedModel.transform_threshold_params`, October 2026.
