Ironically, I think there is a connection between the first part and the last part: Google usually has well-announced shutdowns with good shutdown plans with long migration-out periods, which results in Google shutdowns spending a lot of time in the public eye (when they are first announced, when various milestones hit, and as they are in their final period), which magnifies the attention around Google product shutdowns and the impression of their significance.
OTOH, its also one of the reasons I'm very comfortable using Google products -- any product can get shutdown if its not doing well enough (and a company that won't voluntarily shutdown products that aren't contributing will find itself shutting down along with its products), but at least Google usually does a good job of handling shut downs.
Would you mind elaborating why it was shut down if it had accelerating usage? seems like a good reason to keep it going instead of shutting it down to the public...
I think mobile definitely played a key role, as the old UI didn't work there at all.
Also, Moderator actually started out as an early appengine app written in python, then was ported to normal Java.
If you want the algorithms, they are mostly already public.
Given the way UI/etc frameworks are today, it would take you about 1000x less time to write moderator again from scratch than it would to try to massage an open source version into working.
I was mostly interested for pedagogical reasons, not actual reuse. I'll probably just try to simulate a session with it myself if I can still signup, and record a screencast. Be my own test user for a UX experiment.
It is unlikely, for example, you would ever be able to discern the voting and question scheduling algorithms from reading the source code, despite the fact that i could write them for you in about 1-2 pages of java.
Externally, we mainly used it for civics related stuff, though it saw some serious popularity in niche areas (like people using it to do youtube videos + questions), and was, in fact, in the accelerating growth phase when it was deprecated years ago, and we essentially cut off new external usage.
Because of how it was being used, despite being deprecated, it was kept running for many years to make sure people could find or build alternatives.
(Despite what people may think of Google here or there, at least all of my projects have had good shutdown plans :P)