SpamBayes for Olde Worlde environments-----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Behalf Of Stuart Moors Sent: Thursday, December 14, 2006 12:01 PM To: [email protected] Subject: [Spambayes] SpamBayes for Olde Worlde environments <...> So, how about - for the email service providers, a facility that: A. filters according to Bayesian principles,
The assumption behind all client-side spam filtering is that the communications costs of transferring messages is lower than the cost of filtering them on the server. This is not very practical for a metered dialup account with a lot of spam. Some ISP's perform user-specific Bayesian filtering on the server, so you may be able to get this by changing your email provider, even if you are restricted as to your connectivity provider. B. retains spam at the server, either indefinitely until the space runs out or until a per-message expiry date occurs Because you are using their expensive server facilities instead of the end-user's inexpensive facilities, most providers who operate for profit and do not support themselves with advertising don't care to do this. Dial-up users today are unfortunately stuck with the worst of both worlds: the spam problem enabled by widely available high-speed connections along with slow, expensive communications, making it unappealing to download messages for CPU-intensive filtering in the mail client. As others have pointed out, a webmail approach may be better for dialup users. Unfortunately, anything based on a web browser that serves advertising through a dialup line may not solve your problem. The good part about webmail is that you can look at headers in both ham and spam folders and decide what to download into your local mail client. You can also identify messages that are misclassified, often without viewing the messages. If you can find a browser solution for your local machine that minimizes the ads it downloads, this may be practical for you. Another possibility is to find a webmail provider that you pay for and does not serve any advertising to conserve your bandwidth and minimize your connection time. If you cannot find anything like this, let us know. C. alerts the user to a SINGLE message (say once per week) of filtered out stats and summaries This certainly could be done. -- Seth Goodman
_______________________________________________ [email protected] http://mail.python.org/mailman/listinfo/spambayes Check the FAQ before asking: http://spambayes.sf.net/faq.html
