Slashdot Mirror
Google's Search Copying Accusation Called 'Silly'
itwbennett writes "Google's Bing sting, reported in Slashdot just days ago and subsequently denied by Microsoft, is now being called 'silly' and 'petty' by search industry analysts and execs. The reason: it would be impossible for Microsoft to use the copied results to reverse engineer Google's search algorithms. And in fact it is more likely that Microsoft was conducting competitive research. Charlene Li, founder of technology research and advisory firm Altimeter Group, saw Google's actions as a misguided response to a real threat from a competitor in its core search business. 'Google isn't used to having competition. You look at this incident and you wonder why they are doing this. It feels amateurish in a way, a kind of 'they're not playing fair' attitude,' she said."
Microsoft is looking at the user's clickthrough: Given a choice of ten web sites, which one does the user expect to be most relevant to the search term? This is empirical information that is not provided by Google and as such can not be copied from Google.
Suppose a Windows 7 phone user looks up a club through Bing and then meets with his friends there, and the Google Latitude app then counts this in favor of that location. Is Google then copying from Microsoft?
A more apt analogy would be a student attending a lecture on a subject of interest.
Later in that year he takes a class on the subject and it turns out the earlier lecture was taught by a former student. When the final test comes he discovers that the earlier lecture was based on the questions within the test.
Microsoft is correct in their rebuttal. What Google did was equitable to click fraud. It wasn't a proper "honeypot", they simply manufactured a website with a unique keyword and then clicked a link. Unless Microsoft explicitly engineered their program to ignore any activity in the Google.com domain the crawler would simply see a URL with an interesting and unique keyword and an associated link clicked. That's what an indexer is designed to do.
A proper honeypot would have reduced the variables by repeating the test on a non Google.com domain to see if a random website with similar random jibberish and links would also be indexed and whether it would be at a comparable rate. Even with the manufactured websites Google admits that the indexer only indexed 6% of their attempts.
All Google discovered is that Microsoft does exactly what they said they did--improve their result rankings by analyzing their users browsing. Part of their users browsing evidently happened to include the Google.com domain.
"The fact you got modded up just goes to show how far the Microsoft bias runs here."
No really, it's more about being right.
Have you got your LWN subscription yet?