1. Worse is Better: Linux was at first a very simple implementation and famously eschewed excessively modular micro-kernel designs that were popular in academia at the time in favor of a pragmatic monolithic architecture.<p><a href="http://www.dreamsongs.com/WorseIsBetter.html" rel="nofollow">http://www.dreamsongs.com/WorseIsBetter.html</a><p>2. Cathedral vs. Bazaar: Linux was very successful in attracting a large and diverse developer community, whereas the rather top-down management approach of many GNU projects at the time ended up not attracting many HURD contributors.<p><a href="http://www.catb.org/~esr/writings/cathedral-bazaar/cathedral-bazaar/" rel="nofollow">http://www.catb.org/~esr/writings/cathedral-bazaar/cathedral...</a>
Linux was useful to enough people to gain the momentum necessary to get better and keep up with developments. Also, it mostly cares if stuff works, not if it is the theoretical perfect thing.<p>Hurd didn't get to the starting level of usefulness and had no other motivator to gain the necessary momentum. The project also seems to aim for the "perfect OS" instead of getting something working out the door. And now there is this other open-source OS called Linux around which is better by nearly every measure...
I think it's because Hurd is too ambitious. Overly ambitious project almost always fail. By the time Hurd was done, everybody has already forgotten what Hurd is.