There's some very subtle clickthrough tracking going on at Google. Just before you click on a link on a search-results page, at the "on mousedown" event, Google rewrites the links in its search results with a long redirector URL that is presumably being used to track which search results are being selected most often.
For example, the first search result for a Google search for Boing Boing is listed as "boingboing.net/". If you hover your mouse over the link on the results page, the status-bar in your browser displays the link URL as "http://boingboing.net". However, if you right-click on the link and copy the link location, it is revealed to actually be "http://www.google.com/url?sa=t&ct=res&cd=1& url=http%3A//boingboing.net/&ei=U4gJQ6_fBqKiQevXjYIO" (it will probably be a slightly different URL for you).
I have no doubt that most of Google's intended uses for this are beneficial to Google users. For example, Google can use this to refine its search results based on which links Google users click most often.
However, there is a grave privacy implication here, especially when coupled with Google's never-expiring cookie: this new (?) practice means that Google now has a record not just of all the searches you performed, but potentially of all the links you've clicked through on its site.
It may be that Google is simply tracking click-throughs and not associating them with users, but it sure doesn't look like it. Look at the letter-salad at the end of the real URL there: "U4gJQ6_fBqKiQevXjYIO". That looks like a unique identifier to me — if all Google cares about is which result is most popular with searchers, there'd be no reason to uniquely identify each click.
At the end of the day, the thing about this that bugs me is that it is stealthy and non-transparent to the user. If my search-engine is collecting info on every click I make on its site, I want to know that. Further, I want to know what it's doing with that information.
I hope Google will release more info on this today.
(Thanks, Dylan!)
Update Loadquo sez, "If you check the unique identifier, it is constant for each search, but not constant over searches. Which suggests it is part of ploy to see what people click on for certain searches. Whether the search identifier is linked in any way to your user cookie would deterimine whether Google had broken its 'do no evil' code."
Update 2: Koz sez,
I believe the letter-salad you're seeing is related to google's
personalised search & search-history features. In my case, I have an
option in the top right which is:"Turn OFF Personalized Search for these results"
Which reloads the results without the unique ids.
If I want to turn it off, I can click my account and then Delete
personalised search. I don't know if my experience is different
because I'm logged into gmail, but it doesn't look like anything evil
is happening here.
Update 3: Rev Jeffrey Paul sez, "One of their
features is a 'search history', which includes links to the pages you
clicked through to from each search's result page (You must be logged in to your Google account for this to work.)
"It's great for when you found something useful off of Google then close
the tab/window/whatever and end up needing the information again. There's
direct links to just the pages you found important from the results."
Update 4: Dylan sez, "After deleting all
Google-related cookies /and/ turning off personalized search and the
search history feature, it is /still/ doing click-tracking URL's in
the search results."
Update 5: Dave points out that Google's been experimenting with this since at least last February.
Update 6: Knock me over with a feather. There's a Greasemonkey Script to pull out Google redirects (Thanks, Kap!)
Update 7: The author of the CustomizeGoogle Firefox plugin sez, "Today I've added a new feature to the CustomizeGoogle Firefox extension. The feature is to remove click tracking in the Google search results."
