Home
| Calendar
| Mail Lists
| List Archives
| Desktop SIG
| Hardware Hacking SIG
Wiki | Flickr | PicasaWeb | Video | Maps & Directions | Installfests | Keysignings Linux Cafe | Meeting Notes | Linux Links | Bling | About BLU |
kclark at elbrysnetworks.com (Kevin D. Clark) wrote: > If you're going to consider OCR for spam, you might want to read this: > ... > I'd be curious if systems like fuzzyocr or similar systems (which > I haven't played with at this point) handle this sort of scheme. Looking at that discussion about OCR, I got reminded of the #1 reason you should consider running your own private anti-spam software instead of relying on a big commercial email service: spammers, like virus creators, craft their wares to exploit weaknesses identified in software that serves the top 80% of consumers. (Example, most computer virii are written strictly for MS Windows, few for the Mac or Linux. If Linux were suddenly adopted by 80% of desktop users, we'd face a whole different legion of virus writers.) My evidence is purely anecdotal; having installed SpamAssassin back in 2003, and enhanced it with additional exim configs 18 months ago, I've developed an extensive personal set of rules which are tailored to the specific correspondents I have. When contemplating adding a new rule, I don't have to worry about how the false-positive rate might go up for anyone else but me. If I were running an ISP's mail server, I'd have to remove most of my custom-tailored rules for that very reason. (The OCR example's a good one: if I wanted to, I could block all image attachments, period, from anyone not on my white list.) My one lament about SpamAssassin is that its configuration is annoyingly user-unfriendly, and it's so easy to make mistakes that you have to expend effort to have to re-test it every time. -rich -- This message has been scanned for viruses and dangerous content by MailScanner, and is believed to be clean.
BLU is a member of BostonUserGroups | |
We also thank MIT for the use of their facilities. |