The Noisy Channel

 

Wolfram Alpha is Live, But Struggling

May 16th, 2009 · 12 Comments · Uncategorized

Wolfram Alpha is live, though it is experiencing some strain under load. Lots of reactions on Techmeme, both commenting on the brief launch delay and expressing mixed reactions to the service itself.

I encourage you all to try it, at least when it recovers from the initial load. If nothing else, I need for everyone here to keep me honest after I’ve been spouting my opinions about Wolfram Alpha for the past few weeks!

12 responses so far ↓

  • 1 Andrew // May 16, 2009 at 8:54 am

    Wow, this is pretty cool. At first, I was inputting search strings from a Google perspective, but then I started to adjust my inputs and expectations, and there is definitely a very cool application for the results returned by this fact/data engine.

  • 2 chris dixon // May 16, 2009 at 9:55 am

    There is definitely some very cool stuff in Wolfram Alpha. I agree with you that the NLP interface just makes it harder. I would also argue that putting a search box there is probably a mistake. Google has done such a good job that they’ve set people’s expectations for search boxes so high. (I would probably have opted for an SEO-able directory and maybe a spotlight like guided search box). If you took a million random queries that people type into Google, what % do you think WA would cover?

  • 3 Daniel Tunkelang // May 16, 2009 at 10:55 am

    That’s an interesting question. But even if they do cover queries that Google handles (and they do hit queries from the head–check out Google’s hot trends–the question is whether they do as well as Google on them. Google often points to Wikipedia.

    For example, the #2 query in Google’s Hot Trends today is Tucker Carlson. Google points to Wikipedia. Compare to Wolfram Alpha. I’d say Google/Wikipedia wins–and that this is representative of the sort of informational queries for which people turn to Google (and for which they are ultimately served by Wikipedia).

  • 4 chris dixon // May 16, 2009 at 11:00 am

    The truth is if there is any threat to Google (which I don’t think there really is) it is growing too dependent on Wikipedia. At some point if the vast majority of searches take you to Wikipedia, people might just bypass Google. I suspect that good sources of information will remain sufficient fragmented that this isn’t a real threat…

  • 5 Daniel Tunkelang // May 16, 2009 at 11:02 am

    Agreed. Google won’t discuss what fraction of searches they send to Wikipedia, but I think it’s substantial–particularly if you don’t count the navigational queries they send directly to home pages or the transactional queries they send to product pages on ecommerce sites. I also agree that Wolfram Alpha poses no threat to Google.

  • 6 Omar // May 16, 2009 at 12:02 pm

    I ran a few queries but I’m getting a lot of “Sorry Dave” messages. Will try later.

  • 7 Daniel Tunkelang // May 16, 2009 at 12:08 pm

    Give it a week–right now, Wolfram Alpha is the top story on Techmeme. They clearly didn’t prepare for the load that, given all of their marketing, they should have anticipated. You’d think they’d have learned from Cuil.

  • 8 Omar Alonso // May 16, 2009 at 4:45 pm

    Traffic is getting better now. First reaction after a couple hours: I like it. There are some issues like poor coverage, no disambiguation, and no results but I’m sure they’ll fix that.

    My impression? This is like using Mathematica for the Web and if you know how to ask questions in the right way, you can get some very interesting results. We can argue if a Mathematica interface is the right approach or not, but I like the bit of fresh air. I’m glad that there is competition out there moving away from the top-10 list of trios (title, url & snippet). Finally I can see nuggets/facts in a single page, which is *good*. No more manual foraging until you find what you are looking for. Granted, for certain queries and topics.

    I don’t think WA is a search engine nor a Q/A system. This is something different and I like that. You can’t compare WA to Google or Yahoo like many people are doing now. This is a good alternative to access information on the Web. It has a lot of shortcomings but I like the overall direction.

    Few examples of the many that I did try:

    Strong in math: http://www.wolframalpha.com/input/?i=3x^2+%2B+7x+-+5
    Like the timelines: http://www.wolframalpha.com/input/?i=nobel+prize+winners
    Poor coverage: http://www.wolframalpha.com/input/?i=turing+award+winners
    Didn’t understand user intent: http://www.wolframalpha.com/input/?i=ibm+ceo

  • 9 Daniel Tunkelang // May 16, 2009 at 6:32 pm

    Omar, I like the idea of it being Mathematica for a collection of curated content–I wouldn’t go so far as to call it “for the web”. And I think you sum up my criticism in the caveat “if you know how to ask questions in the right way”. I think there’s a lot of potential here, if they can deliver it in an appropriate interface. NLP doesn’t cut it–I’ve seen my share of attempts to deliver natural language interfaces to create structured queries and structured content, and it isn’t pretty. Wolfram Alpha can do better–and I hope they will.

  • 10 Omar Alonso // May 16, 2009 at 7:19 pm

    I agree, they need to work on the interface. Lots of stuff for HCIR.

  • 11 ken ellis // May 18, 2009 at 10:37 am

    As you said, brittle interface, horribly over-hyped, and they’ve probably abused the phrase “natural language processing”, but my affection for the site has grown since Friday. Here’s a nice one up your alley, information on the word “thought”: http://www24.wolframalpha.com/input/?i=thought
    spoken+written frequency, narrower terms, broader terms, etymology, I’ll never use dictionary.com again. Although they miss the past participle. There seem to be a lot of great niches like this, if you can find them. The input interpretation links are also quite nice, a good way to help you clarify what you were looking for, and guide you towards one of those well-populated niches. So although the hype was out of line with what they’re doing, I still give them good points for interface design.

  • 12 Daniel Tunkelang // May 18, 2009 at 10:48 am

    Jason also liked them as an etymology resource:

    http://mendicantbug.com/2009/05/16/first-impressions-of-wolframalpha/

    My reaction? It’s cute, though it strikes me as overkill as a GUI for WordNet and the other sources they cite.

    Don’t get me wrong, I want to like these guys. I’m all about structured access to structured information! But they’re trying to do it with their hands tied behind their backs–and it’s frustrating for me to watch.

Clicky Web Analytics