Data usage

Nothing you write here is used to train an AI model, by us or by anyone we send it to. We do not sell content to model developers and we do not licence the forum as a training corpus.

Your question is sent to the language model that runs the desk, which reads it and decides what to search for. The searches run on our server against the Internet Archive and OpenAlex. The catalogue records that come back are sent to the model, which writes the reply you see.

The model is used for two things only: understanding what you asked, and explaining what the search found.

This is the guarantee the whole feature is built around, so it is worth stating exactly.

Every document shown to you came back from a real catalogue search in that request. The model can only refer to records the server retrieved and handed to it. If it writes a reference to something that was not retrieved, that reference is discarded before the page renders, rather than being shown as a broken link.

What kind of document something is, and how close it sits to the events it describes, are computed from the catalogue record by our own code. The model is never asked, because those are claims about the world.

Catalogue results are cached on our server and reused for a short period when someone searches for something similar. The cache holds catalogue records, which are public bibliographic facts, not anyone's question or account.

This is a cost measure and a speed measure. Every reference desk question costs real money to answer, and the cache is what keeps the desk open without a paywall.

Accounts are limited to a number of reference desk questions per hour, and to a rate of posting and account changes. The counters are tied to your account and expire on their own.

The limits exist because the desk calls a paid API. They are not a judgement about you.

The model provider receives your question text and the catalogue records retrieved for it. It does not receive your email address, your username, or your account identifier.

The Internet Archive and OpenAlex receive search strings and our server's requests. They do not receive your identity.

Forum posts, community descriptions, and profiles are public documents. They can be read, quoted, linked, archived by third parties, and indexed by search engines. Assume that anything you post is permanent in practice even after you delete it here.

If something about you needs to come down, write to contact@centuriesuntold.com and say what and why.