Bryan Rawlins wrote: > This has me curious, I'm going to try and compute a probability that a > messages is/is not spam based on if the sending server uses TLS. > Probablities will be calculated based on results of our existing filters > and will not be influenced by the data collected. > > I'll post results to the list after approximately 24 hours. > Here's the results of around 23 hours worth of data.
Some things to keep in mind: Over this time period we estimate 70%+ of our volume was rejected at the firewall/SMTP conversation level and thus is NOT factored into these numbers. (We don't track the 'low hanging fruit') Not all mail judged by the system is then used as feedback, so these numbers aren't a good indicator of absolute volume, but should be statistically accurate. Our filtering engine is proprietary, written fully in-house, so results with other systems may vary. Total: 122383 Spam 32594 Ham No TLS/SSL: 122383 Spam 27057 Ham TLS/SSL: 1064 Spam 5537 Ham My $0.02, It appears to be a useful as an indicator to influence an overall score (eg. spammassassin), but as with many other things it doesn't appear to be good for making an absolute decision based on just this information. As an aside I also tracked what cipher was being used, albeit for a shorter period of time. Totals: 948 Spam 4525 Ham sslv3:aes256-sha:256 23 Spam 39 Ham tlsv1:aes256-sha:256 871 Spam 3781 Ham tlsv1:des-cbc3-sha:168 33 Spam 273 Ham tlsv1:aes128-sha:128 1 Spam 9 Ham tlsv1:rc4-sha:128 14 Spam 160 Ham sslv3:des-cbc3-sha:168 0 Spam 7 Ham tlsv1:rc4-md5:128 6 Spam 244 Ham sslv3:rc4-md5:128 0 Spam 12 Ham Bryan Rawlins Systems Administrator OnlyMyEmail, Inc. -- ## List details at http://lists.exim.org/mailman/listinfo/exim-users ## Exim details at http://www.exim.org/ ## Please use the Wiki with this list - http://wiki.exim.org/
