# Study 02 — Regime dynamics: pre-registration
*Filed on 13 October 2026, before any analysis · author: Bran Brière · [version française](02-regimes-preenregistrement.md)*

The date of this file in the repository history is authoritative: the rule below was set before looking at the data.

## Question

Which observable states of the country precede a change of power (revolution, palace coup, electoral defeat, vacancy)?

## Data-splitting rule

| Period | Use |
| --- | --- |
| 9 October 2026 – 8 January 2027 (era 1, first 3 months) | **Exploration:** the data may be examined and hypotheses formulated. |
| From 9 January 2027 (at least 3 months) | **Test:** the hypotheses, frozen before that date, are evaluated on months never examined to build them. |

## Commitments

1. The hypotheses and their success criterion will be added to this folder, in a dated file, **before 9 January 2027**. No hypothesis added afterwards may be presented as tested.
2. Data from the test period will be neither plotted nor summarised before the hypotheses are frozen.
3. Firmware changes during both periods will be listed; their possible effect will be discussed in the limitations.
4. If the country changes era (RESET), the rule applies separately to each era.
5. The result will be published whatever it is, including if negative or inconclusive.

## Limitations known now

A single real country: the citizens are not independent experiments. The political rules are written in the code; a regularity that follows directly from a rule is not a discovery. Computer simulations will serve as a reference.