> For the complete documentation index, see [llms.txt](https://tonify-1.gitbook.io/tonify-docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://tonify-1.gitbook.io/tonify-docs/simulations/sim4.md).

# sim4 — the play matrix

The bipartite substrate — one graph, three intensity regimes, the lemmas checked as unit tests.

**The question.** The lemmas hold for any play matrix. Do they hold on a *large* one, built to match the anchors of the real distribution — and what happens to the crossover when the intensity–size coupling takes its measured value?

**The world.** 20 000 artists, 1 101 981 listeners, 21 820 434 active pairs, 462.4M streams. Artist popularity is drawn piecewise (lognormal body, log-bridge, Pareto tail) and calibrated to three anchors; listeners pick artists by exact weighted sampling without replacement — preferential attachment.

**Three regimes on one graph**, so the comparison is ceteris paribus: an ergodic control where every listener has identical intensity, the heterogeneous canonical substrate, and the coupled family where pair intensity scales with audience size by the measured `γ`.

## Gates — printed before any conclusion

| Gate | Check                                   | Result    | Tolerance |
| ---- | --------------------------------------- | --------- | --------- |
| G1   | control regime: max abs(UC/PR − 1)      | 0.000e+00 | ≤ 1e-12   |
| G2.1 | share of artists below 1000 streams     | 86.6%     | 86–88%    |
| G2.2 | share above 225 734                     | 2.60%     | 2.4–2.8%  |
| G2.3 | share of streams held by the top 0.28%  | 40.3%     | 40–55%    |
| G2.4 | the Theorem 1 identity across all pairs | 7.105e-15 | ≤ 1e-9    |
| G2.5 | zero sum: abs(ΣUC − ΣPR)/ΣPR            | 1.838e-14 | ≤ 1e-12   |
| G3   | crossover sign violations               | 0         | 0         |

The control gate is the sharp one: with identical listener intensities the two pool rules must coincide *exactly*, and they do — to the last bit.

## The crossover at measured coupling

| Coupling γ               | Artists with UC > PR | Among the top 0.28% |
| ------------------------ | -------------------- | ------------------- |
| 0 — independence control | 37.1%                | 33.9%               |
| +0.12 — measured centre  | 45.6%                | 8.9%                |
| +0.28 — measured head    | 52.8%                | 3.6%                |

![L1 identity and the measured coupling](https://raw.githubusercontent.com/ProximaCA/tonify-sims/main/figures/fig16_L1_crossover.png)

## Three mechanisms on one substrate

Pool rules pay every stream something; the direct mechanism pays 75% of artists exactly nothing, while 7% earn zero under every mechanism because their audience is empty.

![Three mechanisms](https://raw.githubusercontent.com/ProximaCA/tonify-sims/main/figures/fig15_three_mechanisms.png)

## Listener composition beats artist size

A companion run gives every listener their own coupling — measured separately for mainstream and beyond-mainstream listeners — and asks who gains. Controlling for artist size, moving an audience from all-mainstream to all-beyond multiplies the user-centric gain by roughly three (coefficient +1.123 ± 0.048, against a size coefficient of −0.018). And the share of artists better off peaks at the *most mixed* population — 62.4% at a 50/50 split against 49.5% and 39.3% at the extremes. That peak is the dispersion premium of corollary (d) showing up in a cut the model was never built to test.

![Group-level coupling](https://raw.githubusercontent.com/ProximaCA/tonify-sims/main/figures/fig18_group_gamma.png)

## Run it

```bash
python3 sim4/bipartite_gen.py    # gates G1–G3, exports, fig15–16  (~7 min)
python3 sim4/group_gamma.py      # gates H1–H4, fig18              (~8 min)
```

Specification, changelog and the red-team ledger: [sim4/SPEC.md](https://github.com/ProximaCA/tonify-sims/blob/main/sim4/SPEC.md), [sim4/README.md](https://github.com/ProximaCA/tonify-sims/blob/main/sim4/README.md).


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://tonify-1.gitbook.io/tonify-docs/simulations/sim4.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.
