|
In recent months there’s been a robust and apparently well-funded debate about the legal status of search engine results, in particular Google’s search results. On Tuesday, Tim Wu, a well-known law professor at Columbia weighed in with an op-ed in the New York Times, arguing that it’s silly to claim that computer software has free speech rights. Back in April, equally famous UCLA professor Eugene Volokh published a paper, funded by Google, that came to the opposite conclusion, that in some cases they do. (Personally, I think they do to the extent the results reflect the intentions of the humans who wrote the code.)
The reason this is a hot topic, of course, is because some people whose web sites don’t appear as high as they’d like in search results think it’s a monopolistic plot against them, and Google should be required to present search results in a neutral way. It might be, but more likely it’s not, and the cure would be far worse than the problem.
The whole argument about search neutrality is based on a false assumption, that there is such a thing as a neutral search result. Any mechanical definition you can invent, e.g., the page with the most incoming links, or the page with the most incoming links from other domains, will instantly be gamed by SEO spammers and the answers will be useless. Furthermore, a good search engine does a great deal of semantic analysis to get useful results. For example, if you search for key lime pie, Google recognizes that as an idiom, looks for it as a unit, and also realizes that it matches a lot of recipies so it adds decorations to the search page appropriate for a recipe search. It’s a strong enough idiom that many searches, e.g., for “can lime pie” will be redirected to key lime pie. If you happened to name your web site “can lime pie”, too bad, your name will be autocorrected.
How the heck you can make that “neutral” without completely destroying the utility of a search engine? You can’t. The only way to imagine that you can is to completely fail to understand what search engines do.
The only place I can see any possibility of a remedy is in the universal search, where Google adds results from maps or plane schedules or the like. Some decades ago, as part of an antitrust settlement, IBM agreed to document and separate out some of the functions of their mainframe system OS/360. That way, if people wanted to use a competing product for a function, the product could use the defined interface and people could install it and it’d work. In practice, hardly anyone ever did, but the interfaces were there if anyone wanted them. I’d think something like that might be workable for the results other than search, with maps being the prime example. But it’s not the same as making the results “neutral”.
Sponsored byRadix
Sponsored byIPv4.Global
Sponsored byVerisign
Sponsored byWhoisXML API
Sponsored byVerisign
Sponsored byCSC
Sponsored byDNIB.com
John, you might find my 2006 article on this topic interesting. It supports your position. http://ssrn.com/abstract=893892 Eric.