You can be too real time: Google vs Bing
Feb. 23rd, 2010 08:59 pm![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png)
Back when blog search was a big thing, we did the rounds of various blog search startups in silicon valley; I was finding that searches on Google were dominated by what was new rather than what was authoritative and useful and I was interested to see if blogs were being seen purely as fodder for ADD-like surfers wanting just novelty or as a lightweight publishing system for building up a corpus of information. Is it, like the teenagers who caught on to text messaging first, just another way of saying 'what are you doing now? now? now?' or will we want to go back and look up recipes and travel diaries on personal blogs, tips and guides on technical blogs? Eventually, Google pushed the blog results back down into the general mix, where they belong - but now it's Twitter dominating the results.
Take 'mike siwek lawyer mi'. Steven Levy got an unusual level of access at Google, and covers the different stages of what Google has used in its search algorithm over at Wired. He does it rather uncritically; it's a piece about Google not an analysis of the state of search and he skips technical analysis of Bing in favour of colourful metaphors. One of the the things he quotes is Google's claim to do better on a specific search as an example of how it's better on names than Bing; search for those four words and Google thinks you want a lawyer in Michigan and puts him at the top of the list, Bing, says Levy, doesn't get him for several pages. But search is a work in progress; Bing now has several ways to find the lawyer in the first few results, including directories of lawyers. What does Google actually find tonight? A lot of references to Levy's article, to tweets about the article - and no link on the first page of results that brings up the man himself.
No single result is a good test of search; I keep both Bing and Google as search providers and maybe one time in ten Bing doesn't give me what I want and I repeat the search on Google. I'm sure in a few days, Mr Siwek will bubble back up on Google. But Bing's results aren't broken by those 'more important because they're recent' stories the way Google's are; there are links to Levy's piece but it doesn't assume that's all you want. I suspect that's a different emphasis in the search tuning and usually, that's more useful. There's a whole range of what makes things interesting and novelty isn't the only measure.
Take 'mike siwek lawyer mi'. Steven Levy got an unusual level of access at Google, and covers the different stages of what Google has used in its search algorithm over at Wired. He does it rather uncritically; it's a piece about Google not an analysis of the state of search and he skips technical analysis of Bing in favour of colourful metaphors. One of the the things he quotes is Google's claim to do better on a specific search as an example of how it's better on names than Bing; search for those four words and Google thinks you want a lawyer in Michigan and puts him at the top of the list, Bing, says Levy, doesn't get him for several pages. But search is a work in progress; Bing now has several ways to find the lawyer in the first few results, including directories of lawyers. What does Google actually find tonight? A lot of references to Levy's article, to tweets about the article - and no link on the first page of results that brings up the man himself.
No single result is a good test of search; I keep both Bing and Google as search providers and maybe one time in ten Bing doesn't give me what I want and I repeat the search on Google. I'm sure in a few days, Mr Siwek will bubble back up on Google. But Bing's results aren't broken by those 'more important because they're recent' stories the way Google's are; there are links to Levy's piece but it doesn't assume that's all you want. I suspect that's a different emphasis in the search tuning and usually, that's more useful. There's a whole range of what makes things interesting and novelty isn't the only measure.