This is awesome. I haven't been terribly optimistic about Wolfram Alpha (I feel like it's a huge nerdgasm that regular people just won't get) -- but an API makes it really interesting to me. My mind is swimming with ideas of apps people could build given something like Wolfram's engine. I don't know why I assumed there wouldn't be an API, but I'm sure glad I was wrong.
It's feeling more and more like a natural language interface for Mathematica. Which, I think, means that an API is a way to get around the layer of Alpha they've wrapped around Mathematica.
It'll be most interesting if you can supply your own knowledge representations. Imagine being able to tie up your own information against the huge repository of data that (appears!) to be present in Wolfram Alpha.
If you search for site:wolframalpha.com , you'll see that Google even crawled some search results. I guess when it was briefly opened earlier in the week, Google had a look.
Also, I got one of the TC invites to try W|A a couple hours before the main launch. It is much less useful than you'd expect. No data on: crime rates, test scores, alcohol content (though it does tell you the LD50 of alcohol), etc.
Fun stuff: side-by-side comparisons. Compare Apple and IBM, or Harvard and University of Phoenix.
Also, for some of the stocks it shows a "simulated log-normal random walks based on historical parameters" -- a couple charts of ways the stock could theoretically perform in the future. Not too useful, but fun.
Yeah, they had to send a follow-up email. It looks like the beta was kind of last-minute; my email from the Wolfram folks misspelled "Wolfram" and "Password" -- and mixed up my password and my username.
But they caught the relevant mistakes pretty fast.
I'm curious to see the sort of services that people build with this. This could be a opportunity to get in at the beginning of something huge. Anyone have any cool ideas?
I dream of an automated fact-checker. Of course, to do it right you need Turing Test-passing AI. But if you start small, with the capability to parse a few statements of a few forms, that would still be useful.
I saw an argument on HN recently where people made these statements:
1. The UK is a much more violent society than the US, statistically.
2. There are dozens of U.S. cities with higher per capita murder rates than London or any other city in the UK.
3. Murder rates are higher in the US, but murder is a small fraction of violent crime. All other violent crime is much more common in the UK than in the US.
No-one provided sources; people asked for them later. Suppose HN ran an automated fact-checker over every sentence ever posted, searching for statements of the form:
"[country] is (much|a lot)? more [quality] than [country]" - or a more sophisticated regex that captures more applicable cases.
And then, for the sentences that it understands, it checks the claim. Maybe it highlights incorrect statements in red, or just adds footnotes. This might make online debates more rigorous. Not only would a few statements be fact-checked, but the knowledge that a fact-checking program will look at your comment might incentivize people to be more careful with facts.
Newspapers and other media would find automated fact-checkers useful too.
Taking a longer view (like decades), with good speech-to-text technology, it could fact-check TV broadcasts in real-time. This should improve the debate quality not just of obscure internet forums, but also nationally-broadcast TV. Then, maybe, we'd see an improvement in the accuracy of all political debate, which should improve democracy.
'Wolfram' is a terrible name for a search engine. Many languages around the world do not even use a letter akin to 'w', often using 'v' in place of both 'v' and 'w'. This is one reason for Yahoo's dominance in non-native English speaking countries.
'Wolfram' will be unpronounceable and unmemorable for a large percentage of world's population.
Wolfram is the existing brand they have already long established which already is well known in Universities through Mathematica product. Since this product will appeal to a similar set of users, it wouldn't be a good idea to start a fresh with a new brand.
Just because they won't be able to pronounce it like an old English man doesn't make the service any less useful or powerful. And I don't buy Yahoo's name is the reason for its success.
