This is because it fails to follow its own specification for "escaped_fragment" URLs.
http://developers.google.com/webmasters/ajax-crawling/docs/s...
http://groups.google.com/robots.txt
Google wants webmasters to "opt in" and let Google access #! URLs without using Javascript (they want to be able to use a bot). This is done by providing an alternative "escaped_fragment" URL.
But Google themselves will not allow users to access Google Groups #! URLs without using Javascript. They will not provide alternative "escaped_fragment" URLs. Why not?
Search for the newsgroup comp.unix.soures and you will see an error message that Googlebot was unable to access Google Groups because of its robots.txt
Of course, you will also see comp.unix.sources has (fortunately) been mirrored on numerous servers elsewhere, probably well before the Google acquisition of Deja.
I think Google should give these old newsgroups back to the community in their original format.
One great thing is that USENET posts were dated, so you could look for the first occurrence of a word, phrase, rumor, etc. That's hard to do on the web.
Then, a few years ago, the search function started to get worse with every release, with reduced functionality and more and more unreliable results. You might search for a term and see a result from 2005. Then, when you limited results to before 2006, you would get nothing! Around the time they were trying to force Google-plus integration, it became nearly unusable.
There used to be an advanced search page[1], but that's gone. When you search within a group, there's a pull-down for advanced search options, but that's missing from the main cross-group search. It seems that search operators[2] may still work.
The same kind of degradation happened with Google News Archive and, to a lesser extent, Google Books.
Some history: Google acquired DejaNews and the contents of other usenet archives, and has largely let all of that data languish, with what can sometimes be very weak search abilities of the archives via Google Groups (no hits for XYZ in an active group for XYZ, for instance), and where Google doesn't make the Usenet archives available and visible via the main search Google engine, and has generally become somewhat of problem.
Then there are the folks that dredge up a decades-old usenet thread — possibly having no idea what Usenet is — and post to it, and with the usual hilarity that ensues. "Hey, is PDQ still available?" to a post offering PDQ that originally posted in 1997, etc.
By some appearances, Google Groups is headed in the same direction as Google Reader.
I've never really used Google Groups, but then this might be a good thing in this case. Just trying it quickly, there doesn't seem to be any way to subscribe to a group without getting a Google account, I guess after doing so you might be able to just forward it all to whatever email you want. But to an outsider it certainly doesn't look like "an email listserv", and it's not entirely clear how to use it as such.
Whether this makes Google Groups "shitty", I'm not sure. I guess a large part of it is about expectations. Does Google still claim to "do no evil"? I'm sure if it was Microsoft that was running the service then no one would be surprised.
(Regarding the actual interface, atleast I find for example gmane to have a much better interface, Google Groups seems very bloated with lots of dead space. One additional thing, as a none native English speaker, Google Groups shows for each message, a rather large box with a question asking if I'd like a translation of that post. I have a hard time believing this feature is used enough to warrant the amount of space that's spent on it.)
Microsoft had a bunch of newsgroups running onicrosoft servers. Some of those newsgroups propagated outside MS.
It was a nice experience. There was the Microsoft MVP (most valued professional) programme where people with in-depth knowledge of a product and reasonable people-skills would provide peer support. Including the term MVP in a web search would return the websites for those people. For years (and probably still) those pages were sources of xcellent information about MS products.
Here's one example. Search terms [excel fill handle MVP]
http://dmcritchie.mvps.org/excel/fillhand.htm
This page loads instantly. It tells me what the fill handle is, what it does, how to use it, how to trouble shoot it, how to get more information about it, how to use the existing MS Excel documentation to get this information. It covers some advanced fill handle use, and some gotchas. It alsotells you whih parts of this information are not found in the help files.
In this case: MS did good[1] things for Usenet. Google fucked it.
[1] ignoring the massive amounts of sub-optimality caused by OE having some Usenet functionality but being buggy and insecure.
Unfortunately, "Do No Evil" does not include, "Don't Be A Dick."
Google has killed 15-20 years worth of information that's not only of interest for it's history, but for the still useful conversations on damned near any topic you can think of.
Ok, so not everything is still useful. I can't argue that since there were plenty of sports and entertainment groups, but there were also many groups whose information doesn't really become stale. In hiding this data, Google has fucked over anyone who might have benefited from finding it.
Hell, the very damned first time I ever heard of Google was an announcement on USENET. I thought they were pretty slick then and I started using Google instead of DogPile(? I think it was dogpile. That was a long time ago.).