In the UK we have the laws around how government documents of historical value go into the national archives after 20 years, documents with personal information are held longer (up to 100 years). I could see a good argument for the Snowden archive to be automatically decrypted and published after 100 years on a similar basis for it's historcial value.

I think the answer to what happened is really straight forward though - they published as much of it as they could, what they published is probably pretty representative of other stuff in the archives and there's no news value to publishing 100 variations of the same story. If they published 1% of the archive, but the other 99% is just a million details about the same basic decisions then I don't think it matters much. It's also a really slow and painful process to make sure you're protecting the identities of people.

Having said that, I could see some value in a team taking a privately hosted open source AI model, feeding into it the archive and any related news articles into it and asking it if there's anything newsworthy that we missed.