- it nails every common query every time in a useful way (i.e. stuff people want like: where can I see X film in Y city?)
- it expands beyond the scientific borders. So far it appears only statistical queries and some science stuff works really well. I know they say pop stuff is "coming soon" but I dont buy it. The task is too big to organise that kind of data in the way they seem to be doing.
Basically I think Wolfram has ignored the process which has got Google to the top. Google apprach search from the "index and link everything and then analyse words to find good matches". Frequently it is wrong, but often that can eb overcome by adding or tweaking the wording once or twice. Wolfram is going to analyse your questions - which is fine but suggests that if you dont get the right data small tweaks to your wording are simply going to result in the same data. They are also trying to provide a nice page of data on all sorts of topics. Which is nice (and for advanced statistics it appears to beat anything else on offer) BUT Wikipedia has huge archives of data which can be recovered just as easily. Wolfram perhaps covers the complex data relationships base fairly well (though this article suggests far from perfectly) but for basic data (which one assumes is it's main target) it is probably pointless...
AKA I think there is a lot of Hype. But no delivery.
We should see this for what it will most probably be. One step on the path to creating a search engine which understands and computes your question based on the sum of all knowledge on the internet.
This highlights a problem, which is that if I want to ask about cars, I know people who like cars, and if I want to know about recipes I know people who can cook.
And if I want useless help, I ask someone who gives me a completely confident reply that may have been made up on the spot.
If Wolfram Alpha tells me an answer about cars, should I trust it? If it then tells me another answer about cars, I can't build on my feeling of trust because it may have used a different source. If I ask it about constellations, then what? Is it bluffing me? Is it pulling irrelevant information? Without knowing the answer already, I can't know.
At least a classic search engine can point me to a source, and I can decide on a source-by-source basis how much trust to give, and with lots of results I can compare answers between them.
- it nails every common query every time in a useful way (i.e. stuff people want like: where can I see X film in Y city?)
- it expands beyond the scientific borders. So far it appears only statistical queries and some science stuff works really well. I know they say pop stuff is "coming soon" but I dont buy it. The task is too big to organise that kind of data in the way they seem to be doing.
Basically I think Wolfram has ignored the process which has got Google to the top. Google apprach search from the "index and link everything and then analyse words to find good matches". Frequently it is wrong, but often that can eb overcome by adding or tweaking the wording once or twice. Wolfram is going to analyse your questions - which is fine but suggests that if you dont get the right data small tweaks to your wording are simply going to result in the same data. They are also trying to provide a nice page of data on all sorts of topics. Which is nice (and for advanced statistics it appears to beat anything else on offer) BUT Wikipedia has huge archives of data which can be recovered just as easily. Wolfram perhaps covers the complex data relationships base fairly well (though this article suggests far from perfectly) but for basic data (which one assumes is it's main target) it is probably pointless...
AKA I think there is a lot of Hype. But no delivery.