Simon Kitching <[EMAIL PROTECTED]> said on Mon, 09 Aug 2004 10:05:07 +1200: > I think a better way to measure the number of debian installs would be > for security.debian.org to count unique IP addresses. While lots of > people won't have popularity-contest installed, a large majority of them > will be getting security updates... > > Of course this would not count users of testing or unstable, which don't > have security updates. And it won't properly count people using > apt-proxy, etc. or behind NAT firewalls. But it would be a start.
Or any proxy. I think surveys with known and unquantifyable biases are useless, and as such do not represent a "start". As I said earlier, popcon has one use and one use only - to see relative usage of packages; and even there, it has biases - I also mentioned earlier than laptop users wouldn't use popcon, because they turn off access-time updates on their HD. And as you also say, server users also may be less likely to partake in "fun" surveys such as popcon. But I guess in this use, popcon does represent a start, as long as the people putting the CDs do at least realise the biases. -- TimC -- http://astronomy.swin.edu.au/staff/tconnors/ The Klein-Gordon equation was derived by Schroedinger. Hence its name. -- Peter Robinson, Rel. Quant. Mech Lecturer. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
