| Bluejack ( @ 2009-05-20 07:22:00 |
Things Wolfram Alpha can't calculate
Nobody can deny the success of the Wolfram Alpha hype machine, and I've heard first hand accounts of how awesome the product can be when you have a question that it was designed to answer. However, this "Computational Knowledge Engine" is falling short of my more mundane, practical needs.
For example, can it "compute" when the next bus will arrive? No. Nor can it give me a timetable. Can it "compute" the postage required to send a first class letter to Canada? Apparently not. It does, however know about both Seattle (although not Seattle Metro), and Canada.
I'm not being snarky: I think they've done an amazing job in the domain of structured data, but here's the rub. If you can't predict whether or not they will be able to answer your question, then there's very little value in going there. I could have saved myself some steps and gone straight to the USPS web site (unweildy and annoying as it is), or similarly gone straight to the Seattle Metro site (which is actually pretty awesome anyway).
Now, if the alternative is sitting down with pile of census documents and plotting my own graph of population trends of major U.S. cities, then sure, Wolfram Alpha is a hell of a product. I expect we will see a new citation style for Wolfram Alpha showing up in academic papers any day now. But for every day needs?
Meh.
Nobody can deny the success of the Wolfram Alpha hype machine, and I've heard first hand accounts of how awesome the product can be when you have a question that it was designed to answer. However, this "Computational Knowledge Engine" is falling short of my more mundane, practical needs.
For example, can it "compute" when the next bus will arrive? No. Nor can it give me a timetable. Can it "compute" the postage required to send a first class letter to Canada? Apparently not. It does, however know about both Seattle (although not Seattle Metro), and Canada.
I'm not being snarky: I think they've done an amazing job in the domain of structured data, but here's the rub. If you can't predict whether or not they will be able to answer your question, then there's very little value in going there. I could have saved myself some steps and gone straight to the USPS web site (unweildy and annoying as it is), or similarly gone straight to the Seattle Metro site (which is actually pretty awesome anyway).
Now, if the alternative is sitting down with pile of census documents and plotting my own graph of population trends of major U.S. cities, then sure, Wolfram Alpha is a hell of a product. I expect we will see a new citation style for Wolfram Alpha showing up in academic papers any day now. But for every day needs?
Meh.