By Theo Emery
The Associated Press

CAMBRIDGE, Mass. - It's going to take the best and the brightest to slam the spammers. Hundreds of programmers gathered last week at the Massachusetts Institute of Technology to apply their collective brainpower to a problem that has evolved from annoyance to menace: the rising flood of unsolicited e-mail. Companies and Internet service providers put up a fight with the latest filtering programs, but spammers quickly bypass their defenses.

"This is the most dastardly thing I've ever seen," said attendee John Graham, expressing grudging admiration for a technique he's encountered in which spam messages are broken up then reassembled in order to elude filters. "This is ingenious. There are some really clever people making spam difficult to filter."

Organizers had expected a small gathering of 40 to 80 programmers, hackers and Internet activists, but several hundred packed an auditorium to hear the latest in spam countermeasures.

For the more clinical, spam simply poses a difficult technical challenge. Others are downright offended by it.

William S. Yerazunis, an MIT computer scientist, compared spam to petty street crime - cheap to carry out, profitable for the offender and enormously expensive to halt.

"It's really theft," said Mr. Yerazunis, 46, a researcher at MIT's Mitsubishi Electric Research Laboratories. "And the theft efficiency ratio is about the same as stealing hubcaps and car radios."

Spam traffic has grown from 8 percent of Internet e-mail in 2001 to as much as 40 percent in 2002, according to Brightmail Inc., which provides filtering products for several major Internet service providers.

Spam is costly for everybody. It costs about $250 to send a million spams, but about $2,800 in lost wages, at the federal minimum wage, for those million spams to be deleted, Mr. Yerazunis estimates.

Spam filtering software looks for patterns that suggest an e-mail is spam. But the spammers are constantly evading them, finding new ways to arrange text to make the messages unrecognizable as spam.

Mr. Yerazunis' presentation on his CRM114 Discriminator language was a centerpiece of the conference. His filtering technique "hashes" the messages, matching short phrases from the incoming text with phrases that the user previously supplied as example text, catching spam that might not exactly match standard spam text. He claims that the system has higher than 99.9 percent effectiveness; it can be downloaded free and is compatible with SpamAssassin or other spam-flagging software.

"This thing is even more accurate than humans," he said.

Jason Rennie of MIT's Artificial Intelligence Lab is writing code he says eventually will help filter spam text that is manipulated to be unrecognizable. For example, his program could recognize "mort!!!!gage" used in the place of mortgage, a common spam word.

But filters must be finely tuned depending on intended recipients, Mr. Rennie said, because one person's spam is another person's "ham," hacker-speak for desirable e-mail.