Anyway, I've never done anything in webcollage to try and filter out porn, my thinking being, it's supposed to show what the web looks like, and if the web looks like throbbing cock, well, there you have it. So I was initially somewhat irritated to learn that the random-link URL was supposed to be content-filtered.
But you know, I'm getting really tired of looking at junkies giving blowjobs. There are less pictures of text recently, but it's still way less interesting than before. Blah!
I never really understood why so little porn had been showing up in webcollage. I had been getting random pages by feeding random words from the dictionary into various search engines, and the percentage of porn was pretty low. But Alastair (the Alta Vista guy who implemented the random link) had a plausible theory: he said, "when people search our site for porn, they don't use big words."
Most of the words in the dictionary? On the big side.
It used to be possible to pull random images out of the various image-hosting services by generating the right random URLs, and I got a lot of good webcollage mileage out of those until they caught on. (Several of them used to compose their URLs of: user=X, album=Y, photo=Z; but all you really needed was Z, which was world-unique. But then they changed it so that it would give you an error if you didn't get a matching Y for the Z, which made it nearly impossible to guess right the first time.)
Do any of you know anyone who works at Google? Apparently Ray is never gonna answer my mail again, so maybe I can track down someone else who has the ability to add a random-link to their site.