These are two separate tools.

However, I have heard some rumours recently that WolframAlpha may be in discussions to acquire Yauba (another impressive new search engine in alpha)

If this is true, the combination WILL be a tremendous threat to Google.

I hope that this is my first comment on this topic here, but I want to make it clear to readers who may comment after me, to stop comparing apple & oranges. WolframAlpha is not the same as Google.

Really great coverage of Wolfram/Alpha here Fredric.

Thanks

It may be that I haven't followed the coverage closely enough. Where is the data sourced from? Is it grabbing it from standard Web-based sources such as Wikipedia and the CIA World factbook etc.? What happens if two sources have contradictory data?

Or is there a large team of data inputter and massagers behind the scenes at Wolfram HQ?

A person (a learner) can grab a book on Calculus, which he can look up on the table of content of index on different sections of the book on how to solve different calculus problems. Looking at a calculus book with all the details on it, doesn't mean that the reader will have a smooth read and be able to comprehend calculus concepts on his own. This is different from when the learner sits with a professor of calculus and they go thru the techniques of how to solve various calculus problems. The professor will teach the learner about various techniques and how they're being solved. He also teaches the learner of how to use inference from memory to be able to infer the right technique for use in certain calculus problems. Mind you that the calculus problems that students sit at the final exams are not necessary the same problems that have been taught to them during the semester. Those different problems that the students see at the final exam, can be solved by them when they infer from their calculus knowledge memory in their brain (knowledge base database) aboaut the correct technique to be applied. If their inference (or reasoning) processing is wrong, because they misplaced those knowledge in their memory or didn't store them, then of course, he will not answer/solve the calculus problem or perhaps retrieve a wrong technique to use.

So, Google is a quick reference (or a pocket book guide) for the user while WA is more like a human knowlegeable person that guides the user. It doesn't mean that the human knowlegeable person can answer everything, because you can't ask a doctor to advise you on MBS (mortgage-backed-security), since MBS is a different domain from medicine. There are various knowledge-based in existence today, but there is no universal one, ie, one that is knowlegeable about medicine, law, politics, science, blah, blah, blah,... I suspect that this is what WA is target to grow over time, ie, a multi-domain (working to become a universal one as time goes on) knowledge database. As knowledge grows over time (via knowledge acquisition), the system becomes more smarter.

WA is no Cuil, because Cuil is in the same domain as Google. WA is in a different domain.

BTW, I have developed FIS (Fuzzy Inference System) for various applications in the past and I know the capability of such systems. Today it is limited to specific domain, but if someone can make it multi-domain in one framework, that would be awesome.

Convergence is bound to happen. Just as twitter, friendfeed, facebook, etc are starting to converge with similar features, it is a matter of time that Wolfram Alpha starts to converge with a traditonal search engine or search engines start to converge with Wolfram Alpha.

Companies are not static entities. Google 2009 is very different from Google 1999. As will be Google 2019. As competition intensifies for information, each side will add more features.

Either way, Wolfram Alpha is in a good position, as they can either use their stock backed by the amount of excitement about the service to buy a search engine like Sonny above suggested, or a search engine will try to buy them, as others have suggested. All I will say is that this would be a much better move for Microsoft than wasting $100 million plus on a skin for wikipedia (like for Powerset)

@Sonny Where did you hear such a rumor?

@Falafulu Fisi Fabulous analogy! Thanks!

Wow...I need more drugs.

Susan Boyle had the number one YouTube video of all time at roughly 56 million hits except for this site suddenly telling people there are YouTube videos with over 100 million hits.

What a pile of crap.

Are we really all this stupid just for advertising bucks?

Patrick

Would you care to elaborate what did you search for?

yet...

yeah, right

Good post. Yes, people do forget that products evolve and that companies often buy other companies when the combination becomes even more compelling.

The issue with WolframAlpha is that the queries it handles cannot really be monetised. How many advertisers are really going to bid on a query like "the average body temperature of the fruit fly"? And unfortunately a subscription model tends not to work well on the public Internet.

It would make perfect sense for Wolfram Alpha to either build its own search engine or buy one of the new ones that are out there. I think the best ones would be www.hakia.com, www.yauba.com, or www.kosmix.com. All of them seem to have solid technologies and would work very well with WA. (Hakia and Yauba already seem to have an advertising program in place, whereas kosmix still seems to be pre-revenue.) Another option would be for them to integrate Yahoo BOSS which is an API to access Yahoo's search index. The downside is that it may be quite expensive if you send a lot of querires to Yahoo, since it is not free. A third option would be for them to simply build their own search engine. It is not easy, but, come on, if you developed mathematica, I am sure it cannot be that hard for someone like SW!

I am also curious about WA's "special sauce" - how easily can it be imitated by some other company? What is really distinguishing about it?

I think search business models require getting two things right - 1) the usefulness to users, i.e. the search algorithm and 2) usefulness to advertisers. The second depends very much on the type of economics involved in ad buying. Google success depended just as much on its auction-style of selling ads as it did on superior search results. Google got the data crunching and the economics right - I think that's why they became dominant.

Any next gen search businesses will have to hit both of those nails.

WA met my lowest expectation: a giant calculator. I think they (like most semantic (mostly deterministic inference stuff) guys) don't have a clue to do IR yet. They got the presentation layer mostly right though.

I'm waiting for a search company with Geoffrey Hinton as Chief/Principal Scientist.

For instance, I typed "caffeine ld50" and got nothing.
But typing in "caffeine molecule ld50" brings me all the available data on caffeine, including ld50.

Strange.