> Theo de Raadt <dera...@cvs.openbsd.org> writes: > > > Around 3 months after starting it, the author deleted all the records > > except the FreeBSD ones. > > That's really bizarre behavior. I was not aware of that part. If the > data isn't actually collected or used sensibly, then there is of > course no reason to try submitting data.
Well, back then his code would compute stats for systems which have never existed (OpenBSD/amd64 2.0 being one of the funniest examples). Also, for some reason, there was about ten times more OpenBSD/luna88k numbers (of which we are only aware of a handful installations) than all other OpenBSD platforms together. Of course since the code which computes stats from the gathered data is not available, the only sensible thing to do was to mention bsdstats values can't be trusted, and move on. This indeed caused a purge of some of the most blatantly incorrect data. But since there was no communication about this either, there is no way to know whether the data which had been kept is correct or not. Come to think about it, bsdstats.org is basically a web-2.0 flavour of rand(). I am not sure the world needs one. Miod