Found something unexpected while running logs* on my test site. It's fully roboted-out, and its domain name is pretty generic, so I had to use a site: operator to force a result. For the test phrase "example" (that is, ahem, the site's name as a phrase) I got three hits, each with the "A description is unavailable" boilerplate:
example.com/m/
www.example.com/
example.com/
Now, the with-and-without www is understandable: Since they have never requested anything from the site, they don't know that one version redirects to the other. But where on earth did they get the /m/ let alone think to put it first? Is this form so ubiquitous that a clever search engine will assume it's present even if they have not one shred of evidence to say so? (It doesn't exist, in fact-- and I did the test search on a desktop.)
Option B is that they're putting up an URL they met years ago, on the off chance that it's still valid. But if so, why not some bona fide pages as well? (A further test search, with site: alone and no search term at all, leads to the same three results.) I've had the name for about 3 years, and cursory research** suggests it was parked since at least 2009 before the owner let it lapse. Was the /m/ format already widespread in, let's say, 2008 or earlier?
If anyone else has got a roboted-out site, see what comes up.
* A human visitor gave google.com as referer. This seemed awfully unlikely to me, since google has never been much for EMDs, so I went exploring.
** Couldn't use Wayback Machine because they look at current robots.txt and refused to show me anything.
