Wolfram Alpha search engine
http://blogoscoped.com/forum/151457.html
has announced its launch day, it's 18th May:

http://blog.wolframalpha.com/2009/05/08/so-much-for-a-quiet-launch/

It will be interesting to see... I like to think of it as a bare-bones Wikipedia with only the hard facts outlined in it's answer...

It's more than a fact search engine, it also derives data and presents it in interesting ways.

Sergey Brin interned for Wolfram at one point and has also been given a preview of it. Wolfram says that it could make sense having it run behind the scenes of Google searches. Google acquired Trendalyzer over 2 years ago and they recently released public data for population and unemployment on it. Google's charts are interactive while Wolfram display a static image with results. I'm guessing that Google has converted the code base from Trendalyzer over to their own servers/platform by now. So, it will be interesting to see how quickly they start filling it with public data from search queries now that WA is getting so much attention. It will also be interesting to see if the two companies decide to collaborate and allow WA results on Google searches. However, I think the ultimate presentation would be a combination of WA results with Googles interactive charting and trends. The ability to mine, crunch and display this data over time with drill-down capability would be incredible.

I guess that this screencast is pretty new and long (more than 13 minutes):
http://www.wolframalpha.com/screencast/introducingwolframalpha.html

A Wolfram Alpha API document can be found in the Google cache.

http://74.125.95.132/search?q=cache:1yOPEKSvoeUJ:www.wolframalpha.com/WolframAlphaAPI.pdf+site:http://www.wolframalpha.com/&cd=20&hl=en&ct=clnk&gl=us

Daniel Tunkelang at The Noisy Channel has preview access to Alpha.
He has written about it and offered to try searches suggested to him.
http://thenoisychannel.com/2009/05/07/playing-with-wolfram-alpha/

I know that Wolfram Alpha is not a straight Google competitor. Still I think that it gives us the opportunity to look at Google from a different perspective. To boot, I see two interesting questions risen in the web in these days:

- In the last 10/15 years, we have made an habit to ask (search) engines for things that already are in web pages (indeed, Google tells us as a basic search tip that we have to "think how the page you are looking for will be written" [1]). With Wolfram Alpha, we'll be able to ask a question that nobody has thought about, and still have an answer. So Google is just for already-solved problems? Google seems no more unviersal...

- While Wolfram Alpha is about using (already existing) structured data, Google has born to organize the unstructured chaos that are HTML web pages. And seemingly this has always been Google's philosophy: e.g. Google Desktop was born to let you search instead of structuring your files in folders, Google Squared is making a spreadsheet from unstructured data, etc.. Indeed, Google already works with structured data: to keep it simple, let's say it's the case of Google OneBoxes [2]. But now Wolfram Alpha does *a lot* more than telling what's the weather in San Francisco or the geographical location of a telephone area code... So what should Google do? Should they center on unstructured data alone, or should they expand OneBoxes a lot?

(Well, maybe it's the same question, from two different point of views.)

I'd like to read what Blogocopeders have to say about this and similar questions. And of course I'd love a post by Philipp about it!

[1] http://www.google.com/support/websearch/bin/answer.py?hl=en&answer=134479

[2]http://www.google.com/intl/en/help/features.html

What is Google Maps, from this point of view? What Google Translate? Aren't they, at least in some sense, knowledge engine like Wolfram Alpha? In fact they all start from a fixed set of structured data to give an answer to infinite possible questions [1].

It seems to me that Google (I'm basically referring to Google Web Search only) has not only accomplished its basic mission, i.e. to give you *links* to the unstructured data – web pages, but also books – that are more relevant for your query. In fact, Google has also tried to do some basic computation out of structured data, to give you an *answer* to your query: see OneBoxes, Google Maps, Google Translate. But now it seems that Wolfram Alpha has gone a lot more far in this direction.

And: what is the difference between (a) organizing all the pages in the web, ranking them according to the relevance to your query and (b) compute an answer to your query? Philipp once showed us that (a) can be the first step to (b) [2].

[1] Here is what Wikipedia says (in this moment) about Wolfram Alpha:
"Wolfram Alpha is not a search engine, as it does not look up answers to queries on an index of web pages or documents. Queries and computations are similarly posed to it via a text field, but it computes answers and relevant visualizations on the fly from a knowledge base of curated, structured data."

[2] http://blogoscoped.com/archive/2007-02-02-n25.html

Its live now, have been trying out Wolfram alpha all morning, conclusion, its absolutely not a search engine, even though its giving pretty interesting results. For now my conclusion
   – Web searches: Google (=web)
   – Fact searches: True Knowledge (=fact database by web and people)
   – Computable searches: Wolfram alpha (=more less arbitary it seems, the query gets translated by hand written rules into a mathematica query apparantly)

David, what you wirte is interesting. I have two questions for you:

- So what do you think about all the answers that Google tries to give that are not web pages? e.g. OneBoxes, maps, translations?

- What is the difference between "fact searches" and "computable searches"? (Of course I know that every distinction is fuzzy.)

Their blog coverage:
http://blog.wolframalpha.com/2009/05/15/live-from-champaign/

I agree with David: Wolfram Alpha is really useful not when you are looking for a fact (you can find more information about more topics asking Google or looking in Wikipedia), but when you are looking for a computation about one or more facts.

E.g. I would not use WA to search for [Alfred Hitchcock], but I would use it to search for:
- [movies by Alfred Hitchcok with Cary Grant after 1954]
- [American movies by English director that grossed more than 100,000$]
- relationship, if any, between [GDP of director's birth nation, American movies' box-office grossing]

(Interestingly, WA in this moment can't answer such questions, not even the simplest ones like [movies by Alfred Hitchcock]. There is a lot of improvement in the music/books/movies section!)

Sidenote: WA actually convert currencies among different years:
http://www.wolframalpha.com/input/?i=%245+2008+dollars+in+1908
a feature that I wished to see in Google and Haochi kindly realized in iGoogle:
http://blogoscoped.com/forum/135359.html

Google's indexing their results as they have no robots.txt (via Waxy.org):
http://www.google.com/#q=site:wolframalpha.com+inurl:input&hl=en&filter=0&fp=-dx2DTn7Pl4

Unfortunately, there's nothing to index since the site uses a search-engine-unfriendly combination of images output and Ajax.

I have checked the wolframalpha.... it is interesting, but technically i can say that it is not a search engine. but the way of obtaining information is excellent. i guess i m gonna visit it once a day... nice way of computing..

Interested that you see True Knowledge as mainly for facts not inference. It uses computation a lot for questions like this: http://www.trueknowledge.com/q/is_george_w_bush_older_than_britney_spears

"Wolfram alpha" chokes.....It's total lack of prescience left me thirsty.

We're talking....tempted to carjack a gatorade truck level of thirst here!

"Wolfram alpha"

might be more accurately termed, at least for now,

"Incandescent filament beta"

Buddum bum

Thanks folks, I'll be here all week.

(insert cymbal clash here)

TS

Time to read
http://www.theregister.co.uk/2009/05/19/dziuba_wolfram/