http://innocuous.org/articles/2006/01/26/google-just-isnt-very-good/ Richard Tibbetts on Various Topics Thu, 20 Jan 2011 11:35:50 +0000 http://wordpress.org/?v=2.9.2 hourly 1 http://innocuous.org/articles/2006/01/26/google-just-isnt-very-good/comment-page-1/#comment-17 a m i t t a i Fri, 10 Feb 2006 06:10:08 +0000 http://new.innocuous.org/articles/2006/01/26/google-just-isnt-very-good/#comment-17 Google is presenting itself to the academic research community as a cutting-edge group, like Microsoft, IBM TJWatson, etc used to. at least in natural language processing (NLP). rnrnThe others now are not really big players anymore, they seem to have been diverted to app-specific stuff... which is sort of odd, because they never really gave a whole lot away, but Google's providing all their fancy services for free, (though i guess they sell boxen to corporations). rnrnThe stuff they do (that i care about), though, is fuck-hard. Info retrieval, summarization, categorization, translations? They're good at them, very good at them, but they're still laughably far from 'solved' (compared with, say, speech recognition) so they're not in the diminishing returns zone by any means. rnFurthermore, these are all things where raw computer power *does* get you extremely far. One of the reasons why they're whomping ass is because they have more resources. Other research groups can't compete against one facet of their work, much less a startup against all of them. rnrnHell if i know what all their sekkrit software plans are, but as long as they keep doing unglamorous things better than anyone else, they'll not be short of money or brains. rnrnFrom what I've seen over the past year-ish, Google's _the_ place to be. Having hardware in spades would seem to be their biggest feature: none of this "waiting for a turn on the beowulf cluster" bullshit. However, they aren't skimping on smarts, as you said. They've pulled in pretty much the top guy in my field, plus a hotshot post-doc who won the most recent 'bestest conference paper ever' award. Oh, and last week my advisor started a half-year-ish sabbatical at Google... rnrnIt's looking like a giant sandbox for smart people to play in, and as long as they've got PhDs and hardware it should stay that way for a very long time. Google is presenting itself to the academic research community as a cutting-edge group, like Microsoft, IBM TJWatson, etc used to. at least in natural language processing (NLP). rnrnThe others now are not really big players anymore, they seem to have been diverted to app-specific stuff… which is sort of odd, because they never really gave a whole lot away, but Google’s providing all their fancy services for free, (though i guess they sell boxen to corporations). rnrnThe stuff they do (that i care about), though, is fuck-hard. Info retrieval, summarization, categorization, translations? They’re good at them, very good at them, but they’re still laughably far from ’solved’ (compared with, say, speech recognition) so they’re not in the diminishing returns zone by any means. rnFurthermore, these are all things where raw computer power *does* get you extremely far. One of the reasons why they’re whomping ass is because they have more resources. Other research groups can’t compete against one facet of their work, much less a startup against all of them. rnrnHell if i know what all their sekkrit software plans are, but as long as they keep doing unglamorous things better than anyone else, they’ll not be short of money or brains. rnrnFrom what I’ve seen over the past year-ish, Google’s _the_ place to be. Having hardware in spades would seem to be their biggest feature: none of this “waiting for a turn on the beowulf cluster” bullshit. However, they aren’t skimping on smarts, as you said. They’ve pulled in pretty much the top guy in my field, plus a hotshot post-doc who won the most recent ‘bestest conference paper ever’ award. Oh, and last week my advisor started a half-year-ish sabbatical at Google… rnrnIt’s looking like a giant sandbox for smart people to play in, and as long as they’ve got PhDs and hardware it should stay that way for a very long time.

]]> http://innocuous.org/articles/2006/01/26/google-just-isnt-very-good/comment-page-1/#comment-16 Nathan Williams Fri, 27 Jan 2006 14:38:47 +0000 http://new.innocuous.org/articles/2006/01/26/google-just-isnt-very-good/#comment-16 There's something to this, but there's also something to the idea that Google is investing in the infrastructure and tools needed to produce large-scale web projects (GFS just being the most-frequently-cited example I see). This implies that there's value to scale, since a startup wouldn't have those tools to rely on, and would have to invest a lot more to get to that point. On the other hand, it's easy to think that Google is well past the ideal tradeoff and is in the land of diminishing returns. There’s something to this, but there’s also something to the idea that Google is investing in the infrastructure and tools needed to produce large-scale web projects (GFS just being the most-frequently-cited example I see). This implies that there’s value to scale, since a startup wouldn’t have those tools to rely on, and would have to invest a lot more to get to that point. On the other hand, it’s easy to think that Google is well past the ideal tradeoff and is in the land of diminishing returns.

]]>