i used a CR-system until recently. i got rid of it because 1) the whole bayesian filtering idea came up and seems to work well enough that i can live with it and 2) i got tired of adding extra rules to my .procmailrc everytime i subscribed to a mailing list.
i used it because i felt that typical anti-spam techniques like SpamAssassin and others that work by analysing the content of emails are inherently flawed. you're doomed to an endless cat and mouse game with the spammers. you make your filters more sophisticated and they make their spam sneakier. there was no end in sight. at best, you (or your mail admins) spend hours updating an increasingly large file of filter rules. a simple whitelist + CR system (mine essentially just verified that the email had a valid reply-to address) takes you out of that loop. there just isn't a way for the spammers to get around it. checkmate. the extra 5 seconds that it takes for a stranger wishing to contact me out of the blue to authenticate themself is unfortunate, but worth it if it means that i never have to think about spam again.
i'm a programmer and amateur cryptographer, so i appreciate elegant solutions to difficult problems. CR systems to me, are an elegant solution to the problem of spam. if i'd had a better interface for adding exceptions than editting my .procmailrc by hand, i'd probably have stuck with it. ideally, i'd like to see CR systems integrated with a PGP style web of trust. so if Bob authenticates themself to Alice, and i trust Alice, Bob has implicitly authenticated himself to me as well.
in the meantime though, bayesian filtering is elegant enough and requires even less work on my part as a user, so i use that.
| Year | Number of comments posted |
|---|---|
| 2003 | 2 |
Total: 2 comments. View all these comments on a single page.
The most recent 20 comments posted to Electrolite by anders:
Show all comments by anders.