Misterious.

The IDs or hashes or whatever seem not to be unique.

Take the value from one url, put it into Google and you will find several other urls with the same value. Example:

https://www.google.com/search?q=ivv6tf7rfsr0klq5

I didn't think about trying this. With your example, the results appear to be all from the same web page, but that is an illusion, it's just the search engine returning first the results for which the letter case matches. On the third page of results, the URLs are back to be a bit less related. I am wondering if a specific id refers to a portion of someone's browsing history.

These URL's have apparently existed for over a year [1]. The author of that SO question says the links are external to his site.

If you're reading this and you have access to web traffic logs from a fairly popular website, see if anyone is visiting your URL's with this query parameter, then post the referring URL here.

[1] http://stackoverflow.com/questions/14696073/visitor-using-ur...

Good idea. Checked the logs on a high traffic website. And indeed, there are GET requests with the _sm_byp parameter. No referrers. But the user agent is "Mediapartners-Google". Also the IP is from Google. So either a bot from google added the parameter on its own, or it found it in the wild.

My working theory: Safari Mobile bypass. Google's way of getting around Apple's anti-tracking/anti-3rd party cookies in iOS.

Edit: Nevermind. Looks like an Adsense thing.