What?
What?
It's as easy as setting up a lambda with headless chromium and then integrating those results into whatever internal results your system has (if any) on the fly. Google is fast enough that this could be done without any perceptible performance impact, and that's ignoring the possibility of massive caching of common searches. It will look like normal web traffic to google, and the lambda network will result in a diversity of IP addresses, so it might never get flagged.
lol specifically, what Google search results Bing Toolbar users click on and what query they had entered in Google to get those results.
Yes some fraction of this was on google. That's not cheating. No content from google was transmitted back to Bing. Just the next site the users went to. Is Bing supposed to go through the logs looking for visits to google and then delete the rest of the session because it might indirectly reveal something about what existed on the google page?
I should own my records about what pages I go to, not Google. If I share them with Bing it's nobody else's business.
Your theory cannot explain how Bing associated that "next site" with the specific search term the user had entered in Google.
If you had clicked a link on HackerNews, it wouldn't have shown up in Bing under some random search phrase. It's obvious that Microsoft parsed the search query out of the google.com URL, and the only reason why you'd do that is to mine what results were being presented for each search query.