Shared data and tools

A model organism needs a record and tools that researchers can use again. This repository supplies both. The dataset is frozen; the instrument can be revised and its results compared with earlier accounts.

The dataset

The collection recipes identify Reddit posts and comments, pixel placements and the artwork atlas. They reach beyond the 705 communities selected for the current analysis. Coded community records add codes, themes and case accounts. Start with the data guide or use the coded communities.

The instrument

Explicit contracts govern reading, coding, comparison and handback. Agents keep notes and retrieve evidence as needed. One community's record need not be discarded to make room for another. Researchers direct the inquiry and judge its theoretical contribution.

Understand the stages or run the study. For a new run, open three handler chats in order: Collection, community coding, aggregation. Full agent instructions retain the detailed rules.

Evidence you can inspect

Read the completed model, form accounts and audits. A checkpoint interview lets you question a restored analyst and compare its new explanation with saved outputs. The full reference holds source documents and hashes.

Reading pages are available here. GitHub downloads remain private, and hosted reviewer interviews are not enabled. Each checkpoint states what has been verified and released.

A baseline others can improve

Later work can revise the instrument, test another explanation or broaden the population. Its contribution is what it adds to, corrects or qualifies in the existing account. File checks establish what was preserved; they do not establish that an explanation is true.