When I look at, for example, technical updates from the Facebook Engineering teams, it's really a little textbook, a distilled version of the knowledge at that point in time.
Same if you bring new team members up to speed by talking to that person.
There is typically no single "historical" document that does the job. Old documents (e-mails, descriptions, comments, ...) tend to be outdated. New documents tend to be increments, so they don't really make sense if you don't know what is written in the old documents.
I expect that in the near future, automatic summaries based on NLP would be able to do all the heavy lifting, so the head of engineering (or whoever is in charge) will only edit the doc, not write it from scratch.
Now it you want to create a product/company out of that practice, then you would build a system that would let people reviewing messages be able to 'vote them up' based on their importance to the over all work product. That sort of 'auto curation' would then let you scan just the list with a 'curation threshold' and just read the important bits rather than everything in the list.
That might be a good compromise between a dedicated historian/textbook writer and a list which had all the data but it was inter spersed with the occasional "lets gets the group together for a movie / which movie" discussion.