The reacTable is pretty big news in my research field (one name for that field is new interfaces for musical expression). It was very well-received by researchers; but then it got really big by breaking out of academia and into the real world when Björk included it live onstage in her recent performances (example video here).
Her use of it is a tiny little bit gimmicky - it's mostly used to add some squeaky noises over the top of existing tracks, rather than being an integral part of arrangements, or being used with any precision - but in the mode it's being used it works fine, adding expressive daubs of sound with a really impressive expressive freedom. Watch whoever-it-is using the thing and you can see they have really fast and intuitive access to lots of effects and processing units, thanks to the very nice way the reacTable works with the little blocks that you put on it. The graphical display on the reacTable works well for the musician as well as being interesting enough to display to the audience.
This is exactly the kind of thing that should come out of the research field. The research group at UPF (in Barcelona) has a good standing in the field and has produced a lot of research. Not all of the research is theoretically 100% solid, since I remember reading a couple of papers that hurried too quickly towards the implementation, skipping over unsolved issues which really should have been solved first to make the solution worthwhile. But as with the reacTable you can really see the benefit of moving towards real implementations: putting aside commercial considerations (university research departments shouldn't have commercial considerations), the fact that the reacTable is out in the world being used and noticed means that it's now in an authentic context, used by musicians who have to tour and perform and impress a load of people. We have conferences where new interfaces are shown and demonstrated; but when the inventor shows his/her invention to a load of other inventors that's not an authentic context, it's rarified and only a shadow of what should be the real eventual target for these technologies. Open-ended research is fine, but in the Digital Music field even open-ended research should make observations, take data from real-world musical contexts. It would be nice to know how the reacTable group planned/managed the move from research to real instrument... and how much luck vs strategy was involved.