Well I must disagree here.
Problem is, it's something pretty hard to prove straight, since the only way (expect if you are working at Google for example) would be to compare two pages with the exact same ranking (more than PR) and content with two different urls. Experimenting on this directly is tricky because, two pages likes this would most likely suffer from some penalization since they would have the same content, pretty much the same dilemma as per quantum physics actually, when the way we measure things alters the experiment.
But, there are good signs about that. For example, the fact Google will as well highlight the matching keywords in the url of the results proposed in the SERP (if any of course) shows that the information is used, and thus most likely is participating to the weight of the keyword in the page. From this, it's far from crazy to think that the weight given to the matching keywords in the URL could be more important than the same keyword being present once in the page's content, just like for page titles.
Would actually be pretty logical. But, since all this is dealing with optimization, thus optimization of something (here it's content), there is just now way that implementing such optimization could be enough in itself.
Just like for the page title, it's not enough to build serp in itself, and a page title can even be counter productive if the keywords used are not descriptive of the content, but it does matter a lot.
To me, it's essentially the same with keywords in url, can be bad or wrong mostly depending on the relevancy. The main difference is that it would really not be a good idea to use title without any human keywords, like for an unrewritten url.
Url with good and relevant keywords will have a greater impact, I'm convinced about it, and a web site running a good quality keywords (in average) in all urls will benefit from all this tiny bits grabbed through the url of each page and will most likely go up in SERP faster.
Another thing that can matter is, every time the url will be displayed on a page, it will have keywords. It's even more obvious in a forum, when user will post URLs directly without building a nice bbcode link. There is no doubt in my mind that : http://www.example.com/talking-about-this-txx.html
is more efficient than http://www.example.com/viewtopic.php?t=xx
Users will know more about the topic linked, and so will bots, since the href link will contain keywords. Keywords that will be present in the page where the link is posted and associated to the target page. Of course, it goes both ways, keywords can be very bad and mean nothing both for bots and human, but this is another matter.
So again, it's all a matter of quality and relevancy, so I agree that keyword in url are far from always good, but, yes they make some difference.
With a forum, if topic title is not always good to use in url in many cases, it's not necessarily the case with forum ones. Using them as virtual directories in url can really have a positive impact : http://www.example.com/phpbb2-support/topicxx.html
can really be handy for user and SEO useful IMHO.
But http://www.example.com/the-super-nice-f ... picxx.html
would of course not be comparable, it's all matter of equilibrium and common sens.
Now talking about mod authors and the long run, well, that's why I though a whole community would be better than just a thread or too for SEO, but, we're running of topic.