Perhaps a better metric is the most recent commit combined with the number of open issues. Or maybe just the number of open issues, judged by their severity.
Do Ruby/Rails/whoever often have releases that break backwards compatibility?
Although sometimes it's more an issue with the documentation as opposed to code issues.
Sometimes, such changes do need to be made, though; sometimes, you have at least part of a point. On the other hand, those changes might consist of a flurry of commits every time the underlying technology's version bumps, and the project may appear dormant the rest of the time. Some might think it's a "sick" project because it only gets updates in groups every now and then, with three, six, twelve, or even eighteen months between such flurries of updates (depending on the underlying technology), but they very well might be wrong.
[Basing your impression of a given project's quality and health on project activity is dangerous.](http://blogstrapping.com/?page=2011.065.16.43.41)
It's unlikely you'll find it in the software management system of your Linux distribution of choice, and (being not very experienced with C) I haven't managed to get it to compile on Debian -- but it's in FreeBSD ports, and has no problems at all on that platform. Given that FreeBSD is my platform of choice, that means it works for me.
Your mileage may vary, I suppose.
For most software, though, it's true that documentation can always use improvement.
---
edit: Actually, my window manager of choice has basically been abandoned by its developer several years ago, and it's still bug-free, stable, and lacking basically nothing. I finally came up with a feature enhancement I'd like it to have today, after using it for five years with no complaints or wants -- and that enhancement is really just an improvement of an existing feature.
More to the point, it's an enhancement that no window manager has, so I'm likely to need to write the code for this feature enhancement myself if I want it badly enough. I may pick up maintainership for the project to do just that.
Think about that a moment: one feature enhancement requested in have a decade, merely an extension of an already existing feature, and it's something entirely new to window managers as far as I'm aware. This thing has had no need for additional code in all that time, and it has been better as a result of all that.