Come to think of it, what I’d like to see in comment spam detection next is this algorithm: whenever three comments are submitted within an hour, and all three contain the same URL, add the full hostname in the URL to the spam filter list and notify me. Yeah, it’s open to denial of service, but it’s a weak DoS in that anyone who’s denied service can get around it easily by not posting URLs with that hostname. And there are significantly more spammers than there are people carrying out DoS attacks on my comments.

What can I say?

So in the previous post, I debunked the draft rumor going around. I figured I’d do a little more poking and find out who was running Congress.org, the origin of the rumor. They do pretty shoddy research, whoever they are. No big dramatic reveal here, alas. Congress.org is owned by a company by the name of Issue Dynamics Inc.. They’re a political consulting company that focuses on liberal causes; they’re big on grassroots, which explains why they’re running Congress.org. It’s presumably an effective means of encouraging people to generate letters to Congressmen. ...

The talk on the Dean campaign wasn’t all that interesting — Keri Carpenter talked about how the Dean campaign was shaped by the people, and Tom Limoncelli talked about how it was a great experience and touched on the technology some. Nothing deep. Keri Carpenter did say, at the end, that clearly great netroots wasn’t enough but she didn’t really volunteer any ideas on what would have helped. Tom Limoncelli said he thought Dean lost because he was anointed the front-runner early and everyone teamed up to bring him down. That latter seems kind of self-defeating to me, since netroots takes some time to build. You wouldn’t want to use a strategy that puts you ahead early if being the front-runner leads to failure. ...

Sideways is a good movie, but not exactly transcendent. Touching and human and delicate, yes, definitely. But I couldn’t avoid a certain detachment from the characters. Or, no, that’s not right. I couldn’t avoid a certain detachment from the world they inhabit. The characters themselves are sympathetic and interesting, even Thomas Church Hayden’s womanizer, Jack. He is not a particularly good person, but he’s our not particularly good person, and Paul Giamatti is a skilled enough actor to show us why his Miles might be fond of such a man. Even better: when something bad occurs, consequences exist and are not softened. And that makes the characters more believable and brings me closer to them. ...

MUSIC: “3 AM, I’m awakened by a sweet summer rain Distant howling of a passing southbound coal train…” OPEN on ROGER PARKER FOR NEW JERSEY STATE SENATE HEADQUARTERS. MUSIC continues. It is very late at night. It is raining, mildly, not enough to make a statement. The headquarters is in a strip mall plaza, with a big plate glass window opening onto the nearly empty parking lot. Inside, lights are going out one by one. ...

So here’s the thing. I’m sitting in a talk about spam, and the guy giving the talk is running over various HTML tricks spammers use to get spam past mail filters. A guy stands up and says “So obviously the trick is to block all email with HTML in it!” That’s just stupid. First off, it ignores reality. I don’t live in a world in which I can block all HTML email for all my users; neither do most sysadmins. Second, this is very clearly a talk for people who live in that world. If the context of the talk allowed for blocking all HTML email, then there would be an obvious solution and the talk would take about five minutes. ...

Yeah, I pretty much want all of this. Thanks!

cfengine is cool. I dug it. The tutorial was introductory and I was pretty sold on the concepts. If you already know about cfengine there is nothing useful for you in this post. Cheap summary: a host is classified into a number of groups. Lots of classifications are automatic; there’s a linux group (any machine running linux), there’s a 129_120_10 group (any host on the 129.120.10 subnet), there’s a Hr02 group (any host running cfengine between the hours of 2 AM and 3 AM), etc. Why would you want that last? Maybe you only want to do some checks during that hour. Yes, this is yet another way to schedule periodic jobs in a manner that future sysadmins will be unable to find… but I digress. ...