[yjftsjthsd-h]

That actually sounds more reasonable, although it does run tiny risk of being trivial to mess with if a malicious client wanted to skew numbers. But I don't think it's possible to defend against that without being horribly invasive, privacy perspective.

I must say, it feels odd to support a Poettering proposal, but this actually does look like a good solution.

[stordoff]

> although it does run tiny risk of being trivial to mess with if a malicious client wanted to skew numbers

Is that not also the case with the UUID solution? Generating the UUIDs in virtual machines, or just replacing the UUIDs in the requests, doesn't seem out of the question

[msla]

> although it does run tiny risk of being trivial to mess with if a malicious client wanted to skew numbers.

I don't doubt it will happen if this becomes well-known. Activism takes many forms.