Comments on: Wolfram Alpha: Next major search breakthrough?

Stephen Wolfram has come up with a new and potentially revolutionary search paradigm for finding answers via the Web.

[carlg113]

lol

[ddhboy]

So how long until google buys him out?

[knowles2]

I give it a year.

[ericanew]

Microsoft already has there hands in contracts with Wolfa (for future via Bing service displaying WolframAlpha search results). If there is going to be a buyout it will be from Microsoft. I don't see Wolfram selling nevertheless

[Daniel_Tunkelang]

I wish for his own sake that he had the sense to hold off on the hype until launching it. This space is riddled with the corpses of over-hyped solutions that fail on launch, and Wolfram's reputation for hubris doesn't exactly instill confidence.

[Xhorba]

LOL what kind of suggestion is that? Don't hype until you launch because it might fail? Are there businesses on your planet?

[jtara]

I really welcome this development, and hope he can attract the kind of money to be able to index the web as deeply as Google has. Unfortunately, the latter has been the sticking point with all Google alternatives so far, no matter how good they work - they simply don't have the money to index as broadly or in as timely a fashion as Google.

Google has fought tooth-and-nail AGAINST natural language input, usually citing "unnecessary noise words!" Instead, we have the tail wagging the dog, actually changing the way we use language today. More and more, people are speaking in indecipherable keywords, instead of in real sentences and paragraphs. I believe this is a direct result of our dependence on Google, and so we have the Googlization of language. It is a sad development, and perhaps we can start to reverse the trend..

[Xhorba]

Very interesting about google and NLS! I didn't know they had taken a declarative position on the subject until I read your comment (and googled it ;) The problem with NLS is that people expect the search engine to have unreasonable levels of common sense. We need more dog wagging to train people to think of the nls engines context before it will work. Wolfram avoids this problem by giving users the context up front -- science questions.

You lost me with the whole "people talking in keywords" thing, though. I am a software developer with kids... I've seen txt msg speak, but I've never heard people verbalize google interrogatives like, "Angelina Jolie kids" or "tainted peanuts products"... What did you mean, exactly?

[damiandennison]

I guess at this point is it not how good the technology is but how much money you have to put it to use. This is a sad case because we never get to have much of choice when this happens. I think that is bad for all of us.

[flickrz]

This won't go too far. People are impatient - they won't want to wait for more than 2 seconds for results to come back. As I understand, this would require huge amount of processing and would certainly be very slow compare to standards of yahoo, google, live and other keyword based search engines.

[Super2online]

I think your confusing the purposes though. Google is a search for documents that might give you the answer after reading one to many pages/docs to confirm the outcome. This will give you precise answers. Plus you can use natural language. I don't know about you, but I would much rather wait 10-30 seconds knowing the result will provide exactly what I need, rather than the poke and hope that is the current situation with all search engines, including Google. I wish him all the luck in the world!

[jonathan0766]

Wolfram must be looking for some capital. He knows full well that if his system is good, Google will toss a nice chunk of money at him just to remove it from the market. The Google brain trust is very aware of his mathematical capabilities. If there's an elite group of math gods out there that Brin and Page worship, Wolfram is in their company.

[4thDDev]

Many mathematicians appreciate using Mathematica and how it handles computational abstract algebra. HOWEVER, most mathematicians in that area are aware of how Wolfram likes to take algorithms that have been around for years (or 100s of years) and imply that WR created them with their genius. I wouldn't be surprised if Alpha is based on some open source project or a Google API and renting rack space with another company. WR ego aside, they do make solid applications and are good at creating UIs for a target audience.

[rrod182]

yawn

"Google simply retrieves documents based on keyword searches. Google doesn't understand the question or the answer, and doesn't compute answers based on models of various fields of human knowledge."

Surely Mr Spivack has got it wrong. Google has an implicit model of human knowledge - links represent human judgements about semantic relationships.

[markl1984]

I look forward to trying this. Ask google a meaningful question and it will respond with links to advertisements and blogs filled with vapid comments. Maybe with Wolfram alpha we will be able to use the internet and it's vast but diffuse bodies of information to immediately answer real questions and efficiently solve real life problems. To those who found this article a yawn - don't worry - the internet will always have ample supply of pointless games, pornography and illegal music downloads.

[envirolab]

So this is like an automated wikipedia that generates its own answers. Cool, now we get unauthoritative answers from a system with potentially no context behind why the specific answer was given. How is this any better than a magic eight ball?

[scottthesculptor]

As time goes by and the internet gets ever dumber it's harder to sort the useful information that you are searching for from the ever growing pile of useless information.

There really does need to be a better way to get a useful answer.
Hope this one helps.

remember gopher?
worked well when most people on the 'net were computer science professionals.
Didn't need an AltaVista or a Google until the brainless masses found the net through AOL and the WWW.

The valid_info to noise ratio is ever widening.
Just coming up with an answer by analyzing the question instead of searching for one seems like a good idea to me.

[summershoe]

Sounds very high minded but I hope it works as advertised. When the web really got rolling in the 90's I had hopes it would heighten democracy by letting anyone easily know the facts. But it quickly became clogged with so much propaganda you can google your heart out and still believe Al Gore claimed to invent the internet and Donald Rumsfeld was the first to utter his "known knowns, known unknowns, and unknown unknowns" quote. Maybe this search engine will sift through the hype better?

[bbneo2]

Skynet!!!! Run for your lives!!!!

[Palmetto_72]

In any event, I will finally be able to ask the question, "42? What the hell kind of answer is that!"

[gmhendo]

A really interesting development huh? i was taken by the comments that implied or looked at the social effects of the system. On reflection it is easy to see how internet search has changed learning, you can now attend univerity or college from your home, just as an example.
So what are the medium/long term possibilities? How will it affect the way people start to think? Is it another brick in the wall of artificial intelligence? Are these thingds good?

It's an interesting world

If this works--and it's a really big IF--this is the next big thing. Hard to say without knowing what he's really thinking--is it real, or is it PowerPoint. (Oh, wait--bad example. PowerPoint is the one Office tool without XML. D'oh!)

But this could be really amazing. If I were a VC, I'd be on the phone.

[sent2null]

www.trueknowledge.com already does this today folks, and have been doing it for a year now. Here is the problem that wolframalpha will face. Accuracy costs learning time for the semantics of a natural language question. meaning must be given to each word in a statement and related, that can happen through automated feeding of definitions (which takes time and effort on the engineering side) or more cheaply by letting the user space do it (exactly what google did with their statistical search algo.) semantic search sites like trueknowledge.com require lots of learning to be able to answer complex natural language questions. However, it has had a head start over wolfram alpha, check it out and join the beta to see the results, some I just received:

I just asked it a question "what is 4 x 5?" and it answered straight away "20" but it choked on "what is 3 plus 9 x 2" but only because it lacks the semantic equivalence of "plus" to the "+" which I can provide (but it is a pain to do so)

I asked trueknowledge another question just now "where was barak obama born?" and the result, "honolulu, hawaii" straight away. More impressive "What is the closest planet to Earth?" , return "Venus". It has gotten smarter since the last time I used it a few months ago, I signed up for the private beta over a year ago and it couldn't answer the "planet" question then. Note how that question requires the system to know what "closest" means in the context of this sentence! very impressive. I switched the question to "what planet is closest to Earth?" and again "Venus" and then finally I asked "Is Venus bigger than Mars?" to which it returned "I don't know" at which point after telling it the answer was "yes" It was then able to answer the question correctly as "yes". Again the learning process (which spanned several pages) is the hump that these systems must get over to become really efficient. TN has been in beta for over a year! I will be interested to see how Wolfram's service compares in its initial beta.

[Manskj]

We often think of math something we have invented to explain the universe based on emperical evidence but in fact if you drop enough matches on a table you will find the number Pi which leads to the calculated answer of a circle. This intersection between math and cellular automata in this way leads to an answer to the circumference of the earth. So by putting a natural language processor on top and grabbing the implied context(s) and deviations you could skip the math part and vary the bottom layer algorithms of the physical universe to calculate the answer. In other words somebody asks for the distance of flight from Madrid to Sydney and instead of calculating the arc via mathemtical formula you start dropping sticks or some reduced mini celluar automata.
Lets say you want to know how strong the TV signal is in a valley. First you figure out the domain which in this case is radio waves and transmission. Youget the relevant input like radio tower locations and terrain but then you dont use Maxwells Equations you use the fact that space is 3 dimensional and that something must spread from here to there. You include the terrain in the model and calculate and calculate and drop lower order terms.

So we can think of the stack the normal way we deal with stuff as:
1)Ideas
2)Language
3)Physics and Empirically Observed Results (Theory)
4)Math
5)Cellular Automata of the Universe
Wolfram Alfa seems to cutout the middle and deal with it this way:
1)Ideas
2)Language
3)Cellular Automata of the Universe

I've summarized some of this at: <a href="http://isontech.blogspot.com/2009/03/wolfram-alfa-search-engine.html"> http://isontech.blogspot.com/2009/03/wolfram-alfa-search-engine.html</a>

[gabrielweinberg]

Hey, also be sure to check out our new search engine, Duck Duck Go: http://www.duckduckgo.com/. We also have some semantic properties, e.g. ambigious keyword detection: http://www.duckduckgo.com/?q=apple, as well as have zero-click info, e.g. http://duckduckgo.com/?q=Futurama.

Take care,

Gabriel Weinberg
Founder & CEO, Duck Duck Go