From: James Tauber (jtauber@jtauber.com)
Date: Wed Oct 18 2000 - 20:55:00 PDT
I wonder what the algorithm is for "random pictures". It's not evenly
distributed as you get a lot of duplicates in a day. Not many duplicates
between days, though. Doesn't seem to favour people who haven't been voted
for much.
What should they use? A combination of total votes and standard deviation?
I've forgotten my stats, but isn't it the case that you need fewer samples
if the ones you have have a lower SD?
James
This archive was generated by hypermail 2b29 : Wed Oct 18 2000 - 21:08:48 PDT