My few cents:
I agree with John Swinbank's analogy; that is the problem that HipChat was introduced to solve. It was not meant to be a place to permanently store information.
I think the intent/desire from the start was to use HipChat for ephemeral discussions, and mailing lists for more persistent/consequential conversations (including summarizing HipChat discussions). This unfortunately didn't happen; because HipChat retained history, summarizing to mailing list turned into an "when I get to it" task (which was, typically, never). That doomed the mailing lists (traffic is now practically zero), forced everyone onto HipChat, created anxiety about having to follow all chats instead, etc.
While theoretically we could just decide to "do it right" from now on, I am doubtful that a culture change can happen w/o a bit of help from a policy change. This is based on the fact that we've been trying to change the culture basically from the moment HipChat was introduced. In retrospect, I think it was a mistake to turn the history on in the first place: if a tool allows an anti-pattern to get amplified, it will get amplified. And not having history would also clearly differentiate chat from mailing lists (and made it intuitively clear what is a good use case for one or the other).
That's why support the idea (and would even advocate for an even shorter retention period, but am willing to settle for a month to achieve a compromise ). And, as K-T said, nothing that already exists will be deleted – this is about retention going forward.
PS: I'm using "mailing lists" above, as
RFC-85 hasn't been adopted yet.