> Kenneth Porter wrote: >> A recent thread on spam detection suggested that geographical distance >> from sender to recipient correlates with spam, > > I'm really not sure how given that the majority of spam appears to > originate from the USA: > > http://www.spamhaus.org/statistics/countries.lasso > > with nearly as many "issues" as the rest of the top ten combined > (quoting from just one arbitrary source). > > I guess it really depends where the person doing the correlation is based. > > Personally, I think you'd have just about as much success scoring 1 > additional point to any email originating from the US.
It actually works very well with very small and very large distances. Anything that I receive from an IP address located with maybe 50 miles of my location is almost 100% guaranteed Ham. However, I've never received even a single email from China that wasn't spam. This probably has more to do with the absolute location of people I communicate with than the country or distance. What would seem to be really useful is if spamassassin kept the geographic coordinates for all sender IPs and created "hammy" and "spammy" area mappings, and used distance from these as a weighting factor. Terry
