I find this kind of usage frightening. I honestly think that a message history of longer than a week encourages poor organisation to such a degree that daily operation is significantly and silently hampered.
To the degree that teams fall into the habit of just searching slack history for info, I think those are the two key factors:
- we remember bits of conversation, and who said something, or in which channel, can be a good way to rediscover the full details; note-taking and aggregating stuff into a knowledge base or docs repository can strip out that info making it harder to rediscover. (The obvious caveat here is that there can be sharp difference between re-finding a conversation you participated in e.g. in your team, vs finding information about a system you depend on from a conversation you didn't participate in.)
- the process of getting stuff shared in slack into a longer-term knowledge base can be slow and tedious, and may require someone to spend time deciding which pieces are potentially valuable, generalizing from the specific case discussed to a form that makes sense out of context, condensing what was said over many messages, arbitrating between views which are valuable but conflicting, etc.
I would like to see an NLP tool for summarizing threads and extracting the information which is most likely to be reusable.