Even so, even without doing any work to ameliorate these flaws, it could still be (ab)used. Don't assume that it's only useful if everyone can scan which of the top 100 websites you've visited.

Any site could use this to check which competitors' sites have been visited. It's unlikely anyone else has an interest in checking that information, so the cache is not going to be poisoned by anyone else. With knowledge of which competitors a potential customer has checked out, you could do some effective price discrimination -- the guy looking at the $10 solutions sees your lowest price, while the guy looking at some competing Microsoft Dynamics package enters a more enterprisey sales funnel.

It's also useful for retargeting. Throw the code up on an ad network and you only test for cache hits against domains of current advertisers. If there's a hit, store it in a cookie so you don't need to check the (now filled) cache again. You can now show ads for companies a person has already had an interaction with, without having to cookie every visitor to the advertisers' sites first.

It doesn't take much to come up with (mostly nefarious) uses for this, even without perfect accuracy and even without the ability to have multiple parties check the same URLs.

It also doesn't take much to come up with ways to improve the process. You can ameliorate the problem of overlapping testers by having a large pool of URLs from each site to check. The average top 1000 site probably has dozens and dozens of images and other resources per page, each of which can be used for a cache test.