I just noticed that Google is munging search result URIs. For example, if you run a search on “mozilla”, the first result is http://www.mozilla.org/. However, the URI that they link to on the search results page is:
I suspect that the usg parameter is probably one used to prevent bots from gaming whatever results they’re trying to garner, and possibly also to link clicked search results to a specific user or browser session. I understand why they do this, but the downside for the end user is that the copy link option in the context menu of any browser is no longer useful. One needs to actually follow the link to get the URL in a form that you can copy into another application.
From my perspective, this is a pretty major usability bug, and I hope they revert it.
5 thoughts on “Google Munging Search Result URIs”
I think they’ve been doing this for a while. Randomly, you’ll get munged links, allowing them to track whatever it is that they track and randomly you won’t.
It’s kind of annoying, especially what happens when you try to copy it – and results in people pasting huge links to IRC >:(. As far as I know, it’s just there for tracking. I guess it’s nice to know this sort of thing in Personalised History, if you like that sort of thing, and also for Google Webmaster Tools… but it would be nicer if worked, without having to break Copy Link Location.
What Blake said. They’ve been doing this for years. It’s to track which link you click on. The feature was designed to eliminate the need for this but it was disabled due to misguided privacy concerns.
Usually they’re more tricky. Each link the Google’s search results fires an onclick event which rewrites the ‘href’ into a tracking URL. That means that when you hover over the link, or copy it, you see one URL, but when you click it you go via a tracking URL.
Rich, that’s basically correct and the old approach did not affect usability. The new approach does and makes both the copy link function “broken” and also makes the bottom statusbar of most browsers show a URL that is not easily human parseable.