I'm not talking about using just visits to improve search though, I'm talking about using the pattern of interaction on a site to basically replicate a site's data. That is, imagine netflix recommends movie X to you, now Bing infers that Netflix has done so using custom code to parse their IE logs, and then recommend movie X to you.
"Stefan Weitz, director of the Bing search engine at Microsoft, said in an interview the company studies how certain users interact with Google in order to improve Bing. ..."
I'm not sure I get your example (not clear when Bing would recommend a movie to me). So let me give that I do understand.
I go to Netflix and search for the move "Network". I end up clicking on "The Social Network". Later on Bing, if I search for "movie Network" -- I'd hope that one of the movies that comes back is "The Social Network", based on that clickthrough data from Netflix..
In my mind the only thing that is borderline unethical is that we've artificially limited this clickstream data to one company. I'd like to give it out more broadly. It's my data right?
I was considering the case when you go to NetFlix, and they recomment the movie "Social Network" to you, and you click on it. Surely the url will have sufficient info to claim that its a recommendation.
Now bing implements a movie recommendation service, which also magically recommends for you "Social Network", based on clicktracking of Netflix.
IMHO, what Bing is doing with Google is exactly the same scenario as above (extract click data from a competitor's service, and surface the exact same data in their own product).
Why wouldn't I want Bing to recommend the movie? The first thing Bing should do is use Netflix's APIs to suck down all of my ratings of movies. Then it should use my clickstream data for Netflix, IMDB, Hulu, Amazon, Facebook etc... And if I'm using Media Center or SageTV, I want those things to also get consolidated together.
Are you saying you want your Netflix data in one silo, your Hulu data elsewhere, your TV data elsewhere? Don't get me wrong, I want this to all be opt-in, but when I opt-in on MY behavior, I want them to use it.
IMHO, this would be completely unethical.
That's exactly what Bing has admitted to doing with google: http://online.wsj.com/article/SB1000142405274870412450457611...
"Stefan Weitz, director of the Bing search engine at Microsoft, said in an interview the company studies how certain users interact with Google in order to improve Bing. ..."