|
mac geekeryGet your geek on. |
How I Stopped Junk EmailSeptember 29, 2004 - 6:31am
There’s a choice quote from Moby Dick that puts to words my heart’s feeling for all non-directed advertising:
There was a time when I got 160 junk emails a day and, at the time, the number was rising quickly. Open relays, cracked servers, bad ISPs, foreign countries that didn’t care, they all conspired to sell me H)T PR#GNANT T33NAGE V!!!!RG!NS!!!! or fake Viagra by the gross. All this to the guy that neglects to buy groceries for the sake of new hardware. I figured I’m not quite the “target market” for that sort of stuff. So, I put some good technology to use and today I get zero junk in my inbox and Apple Mail files away potential junk before I see it. It’s almost always right. Today, I am as junk-free as it gets. How did I do it? How I Did ItI currently get a grand total of two to five pieces of junk email a day. Now, ordinarily, the retort would be “so would I if I, too, had just opened a new email account” but in this case the email account in question has been the same for seven years. I’ve posted the address on:
First things first, my email address in question is codepoet@pobox.com. This address is managed by a wonderful forwarding company that literally does SMTP for a living so a good deal of my protection lies there. For those bits that slip through, the mail is forwarded to Apple’s .Mac mail account where another (mysterious) level of protection takes place. After that, it goes to my local machine with POPFile installed which adds a layer of Bayesian filtering to my mail to sort it into four groups: inbox, lists, ads (legitimate), ads (illegitimate). After that, Apple Mail’s Bayesian filter scans my inbox and list mail again and weeds out the remainder. It’s a lot of firepower, but most of the software has learned how to do its job so at this point in time it’s mostly maintenance-free. PoboxThe first, and most productive, shield against spammers is Pobox. I cannot recommend them highly enough. Using their service I can setup any of the following rules to protect my account from rogue SMTP servers:
The last one is the most interesting. Set the rules you know will provide good results (do you plan on getting mail from Nigeria?) straight to bounce and the mail never even gets forwarded. Some rules, most notably “black-hole” lists, are rather hit-and-miss with regard to their accuracy, so set them to “flag” a message instead. Think of it like Spam Assassin’s points system: for each rule it breaks, it gets a point. You then set the final rule at Pobox to do something to a message that breaks a certain number of rules. Myself, I have the countries set to bounce and the lists to flag and the flag rule to hold questionable mail for review. For those items I tell PoBox to hold on the server, they send me a mail each (hour|day|week), at my option, with a list of the senders and subjects of each mail being held. I have about a week to claim it before it disappears. It never, ever, hits my mailbox. Ever. Who’s Been Reading Your E-Mail?Now this is all based on SMTP rules. This covers the mail servers and the probability that they are sending spam, but it’s not covering the content of the messages themselves. For that, I use POPFile still. It’s well-trained and has hit a 98% reliability rate with me and never deletes a mail (just flags). As such, I’ll two-prong this problem with mail server filtering, and mail message filtering. POPFile is a Perl-based Bayesian filter for email that isn’t entirely geared towards just junk mail protection but towards “smart filtering” of messages. You setup buckets for your mail to go into and for a training period (usually just a few dozen messages to get started) you tell POPFile what message goes where. It looks at the content and headers of the message, scans for words that are similar amongst messages you file in certain buckets, and then performs that action for future messages of that kind. (If I file all my cocoa-dev messages into the Lists bucket then POPFile will associate the phrase cocoa-dev in the header to Lists, for instance — automatically). Setting this up is a whole other article, which I plan on writing soon. I find it essential. When POPFile files a message, it can either add a header to the message indicating which bucket it used, or alter the subject line. I chose the header and simply had mail sort on the header. For those messages classified as “Inbox” or “Lists” I have Apple Mail check them for junk content (by not stopping a rule on them). For messages classified as “Ads” or “Spam” Mail just files them away in the appropriate folder; if Think Geek or Smart Home send me an ad, POPFile sends it to the legitimate ads bucket and Mail exempts it from any checking, filing it into my Ads folder. If someone wants me to care about the size of my nuts and bolt then they’re toast. At the end of the day, all I have to deal with is comment spam from my sites, but MT-Blacklist on the MT sites and a forced preview here have stopped that for now. It’s amazing how much effort went into making this effortless, but I think it’s paid off. About Adam Knight |
|
||||