Questions, answered.
Everything people usually ask about the library, the labs, the mentor and the API. Can't find yours? Ask on GitHub.
About datamodellib
What is datamodellib?
A free, open library of 38 mental models across 10 families: distributions, networks, diffusion, game theory, dynamics and more. Each model has a plain-language summary, a diagram, when to use it, how it gets misused, free sources, and in many cases an interactive lab you can run in your browser.
Is this the book The Model Thinker?
No. The library is inspired by Scott E. Page's The Model Thinker, but every summary is original writing and no pages or figures are reproduced. We are not affiliated with the author or the publisher. If you find the library useful, buy the book.
What counts as a "model" here?
A simplified, explicit representation of how something works: a distribution, a network, a game, a feedback loop. Writing assumptions down makes them testable, which is what separates a model from a story. The many-model thinking post explains why using several at once beats relying on one.
Is it free?
Yes. There are no accounts, paywalls or ads. The site and its code are MIT-licensed.
Using the library
Where should I start?
Take the guided tour on the home page, or open the library and filter by the Starter level. The Schelling segregation and SIR labs are good first experiments because small parameter changes produce visibly different outcomes.
What is REDCAPE?
A way to choose a model by the job you need it to do. The seven jobs are Reason, Explain, Design, Communicate, Act, Predict and Explore. The playbooks page walks through each one with worked cases.
Do the interactive labs send my data anywhere?
No. The labs are simulations that run entirely in your browser. Nothing you change in a lab is uploaded or stored.
Can I link to a specific filter or model?
Yes. Every model has its own page at /model/<id>, and library filters live in the URL, for example /library?family=Diffusion or /library?diff=Starter.
Mentor, API and agents
When will the mentor launch?
The AI mentor is built and being tested. It will look up models in this library, search the web when needed and show its reasoning. Until it launches, the mentor page shows what's coming, and every model, lab and playbook is free to use.
Is there an API?
Yes, a free, read-only JSON API with no key required. GET /api/models returns the full catalog and supports family, job, diff and id filters. See the API docs.
Can AI agents read this site?
Yes. There is an llms.txt index, every page returns Markdown when requested with Accept: text/markdown, and the agent skills page has drop-in skill specs for model selection and reasoning.
Licensing, privacy and feedback
Can I reuse the content?
The site's original text and code are MIT-licensed, so you can reuse them with attribution and the licence notice. Third-party sources linked from each model keep their own licences. See the terms of use.
Do you track me?
We use cookie-free Cloudflare Web Analytics for aggregate page-view counts. There are no accounts, no advertising trackers and no cross-site tracking. The privacy policy has the full details.
Why do I see a CreditBase pop-up?
We feature CreditBase, a directory of free credits, grants and investors for founders. It appears once per page load and closes with the × button, the Escape key or a click anywhere outside it. It sets no cookies and stores nothing.
I found a mistake. How do I report it?
Please open an issue on GitHub with the page link and what looks wrong. Suggestions for new models are welcome there too.
Still curious?
The fastest way to understand a model is to break it in a lab.