Working papers from HyperPrior Research. Five are in preparation. https://hyperprior.net/papers/ 2026-10-14T00:00:00Z HyperPrior Researchhttps://hyperprior.net/ https://hyperprior.net/favicon.png https://hyperprior.net/papers/the-prior-is-the-product/ 2026-10-14T00:00:00Z HyperPrior Research

Pretraining data is a prior, whether or not anyone meant it to be. We argue for designing that prior deliberately, and for measuring what a model actually absorbs from it. In preparation.

https://hyperprior.net/papers/forgetting-on-purpose/ 2026-10-14T00:00:00Z HyperPrior Research

A model that learns from a stream has to keep deciding how much of the past still applies. We make the case for models that forget on purpose, and for measuring when and why they let go. In preparation.

https://hyperprior.net/papers/scoring-the-future/ 2026-10-14T00:00:00Z HyperPrior Research

The only benchmark that cannot leak is one that has not resolved yet. We set out a protocol for scoring forecasts committed before their outcomes exist, including when the forecaster can move the outcome. In preparation.

https://hyperprior.net/papers/where-beliefs-live/ 2026-10-14T00:00:00Z HyperPrior Research

A model that predicts well has to track the hidden state of whatever produced its data. We look at reading those beliefs off its activations, and at what their geometry says about how the model represents that process. In preparation.

https://hyperprior.net/papers/other-peoples-world-models/ 2026-10-14T00:00:00Z HyperPrior Research

In a market, most of the environment is other agents. We ask what a world model has to learn there, and how to plan with a model that goes stale once the other agents adapt to it. In preparation.