Why doesn't spam make meaningful comparisons impossible?
There are some estimates that 60% of the articles in Usenet today are spam or spam cancels. (30% articles are spam, followed up by their cancels.) And if you look at the numbers in terms of number of bytes, it might be even more than 60%.
If a heavily spammed newsgroup was compared on two different servers, it is likely that there would be wide variations in performance. Some servers reject spam, some servers cancel it, some don't propagate it. Also what one calls "spam" is not called "spam" elsewhere, etc.
The newsgroups in the sample set were selected deliberately to make comparisons meaningful. So spam is not that much of an issue with the newsrAte anyway. It is very important to note that there are many newsgroups which do not have a lot of spam. In fact, most true "discussion" newsgroups don't have much spam. (By "discussion" I mean newsgroups where people actually read posts, and write a reply, -- maybe even a thoughtful reply.) There are plenty of non-discussion newsgroups on Usenet where people just post lots of pictures, or job listings, etc. Those are not good candidates for being in the sample set. (See Why sample these 200 newsgroups, and not some other set?)
Based on inspection, and historical averages, the newsgroups compared are not heavily spammed newsgroups. There is always an occasional spam which comes into any newsgroup. The "expected" values come from a server which was already spam filtered. This means that if you take in spam, your totals will be more than 100% You will only have lower numbers if you reject MORE spam than the expected server.
In conclusion, the comparisons between servers are very meaningful.
What exactly is being counted/compared?
I spam filter like a madman. Won't this throw things off?
My server has 30,000 newsgroups. You only report 200.
Why sample these 200 newsgroups, and not some other set?
How do I signup?
Up to The newsrAte RKT
Up to newsrAte home
Up to Mib Software home
Copyright 1998, Forrest J. Cavalier III, Mib Software
INN customization and consulting