I was invited a while back to be a part of the testing of the new natural language search engine, Powerset. For a while now, I've had high hopes that this might be THE killer search engine we've all been waiting for. It might be...but at this point, it's very difficult for me to know.
I've only played with a few of the the demo sets that they have available in the lab. I won't go into details because frankly, I'm not sure what details I'm allowed to share. But I assume I'm safe to give general impressions.
It's interesting. The concept is one that I'd never considered before, and it definitely has potential. Currently, though, it tends to fail at the implementation level as often as it succeeds. Again, let me stress that this is all very very alpha stage, and the whole point of having us play with the system is that the feedback is used to make it better. There are users there who are totally involved in making that happen. I honestly haven't had the time to get very involved, so I've certainly not been an asset to them (sorry).
So, at this point, I'm really uncertain about whether or not this will ever be the ultimate search engine or not. It's just too hard to make any real judgment call at this stage of its development. So, although you aren't getting anything really useful out of me, I'll leave you with this. If they can take the concept they've got going, and refine it to work consistently, it'll be very cool.
If I have more thoughts or impressions later on, I'll let you know.