Orkut, however, is a rare misfire for the survival model: because Orkut is so old and almost everything from that bygone era of Google has survived indefinitely, it gave Orkut a 95% chance of surviving to 2018. Of course, everyone knew Orkut was a dead social network walking (even before Google+ was a gleam in Page's eye), but how do you tell the model that...?

Groups is in a similar situation, at 90%. I think that's also too high, but I'm not sure what I'd personally estimate the risk at. On the one hand, it feels like Google should be loathe to kill off dozens of thousands of mailing lists and its Usenet archive and decades of history & material; on the other hand, they sure act in every way like they want Groups to die and go away.

In general, the model seems to be accurate when it predicts something is very likely to die, but not so accurate when it's predicting something's very likely to survive. It'll be interesting to followup in 2018 and see how the overall set of predictions did and what an updated model looks like.

Groups is a terrible service.

Strike that. It works well on the groups I use, but the group that I founded and abandoned (Rochester Arabic Discussion Group) is quite literally now unrecoverable. I can ban spammers from my group, but I can't delete their whole post history with prejudice. Had I taken timely action against all spammers as they arrived, it is possible the community would have survived, but now it's in "Warning - Blocked for Adult Content" mode and the actions required to prepare for a review in order to be reopened for posting would take at least a month of nonstop clicking to clean it up. I guess it's my fault for trying to start a community in a language I was studying at a time when I was likely to be too busy with getting my degree to devote proper time to fighting spam and helping to focus the discussions.

There is no way to perform bulk actions of any kind from the admin area. This is my biggest gripe. Groups is a service that needs an overhaul, it does not need to be killed, but only time will tell what Google does with this one.

Groups was an awesome service when it launched back in 2001. It was the search engine to go to if you wanted to search UseNet, which (in the pre-Friendster/MySpace/Facebook days) was the largest social network on the Internet. Google slapped a much nicer UI on it than DejaNews ever had, rivaling desktop Usenet readers but without the need to download anything or run another program.

It started sucking when they took UseNet out of it - then it became just another mailing-list site in an already crowded field. (Granted, UseNet itself was nearly dead at this point.) And the field was shrinking, since more affinity groups started moving to FaceBook, or MeetUp, or getting their own individual sites on the Internet with stock forum software. And IMHO, it really died when it went to the AJAX UI - there's no real reason for a content site like Groups (or Blogger, for that matter) to go heavily AJAX, it just makes it feel clunky.

Seconded. The AJAX UI sucks the most of all. Try loading up 100 pages of threads on one screen, selecting 19,850 spam threads to delete, and finding out an hour and a half later that your request is too large and all your work is lost.

Thirded. The AJAX UI is precisely when I stopped using Groups. It's not that AJAX is bad (obviously), it's that this is a really poor implementation.

I agree. I've never had a use for Groups other than stumbling upon a thread within Groups while researching a bug, or trying to get something new done on linux that I've never done before. I don't see this less-than-absolutely-massive userbase aligning with Google's current interests. Especially while trying to take over the social-discussion world with Plus.