I will be hard to try it in a unbiased way after reading NKS… I actually studied with one of the people that wolfram forgot to cite in his book and was pretty pissed about the lack of credit to people like Konrad Zuse.

And the issue about Matthew Cook and Rule 110 proof, to me the main result in the book (together with the equivalence between one-dimensional CA and Turing Machines), just made more disappointed.

Based on Doug Lenat’s artilce, it looks like Wolfram alpha will mostly rely on a large internal database and some ad-hoc real time web services. This gives me more confidence regarding the accuracy of the data, but less hope when it comes to the revolutionary aspect of this service. That said, I’ll wait to see before making any judgement calls. If their database is large enough, this could still become an extremely useful service. I share your skepticism, but truly hope it will be unjustified.

Perhaps some when in the future but as for before and as I understand, it won’t access public web data, but instead private databases. I guess the same ones that Mathematica 6+ can access under the name of data on demand, that’s my guess.

I hope for the best, but I have serious doubts that it will be a revolutionary thing. So I hope I am mistaken.