Experts say spam fight needs to be more strategic

The fight against spam to date has been too tactical and not strategic enough, according to experts at the Next Generation Networks conference.

"There's too much of the thinking, 'I've got a problem. How do I stop it from hurting me?'" says Phillip Hallam-Baker, principal scientist at VeriSign Inc. The thinking ought to be how to stop spam in general, he says.

"It's a public health problem. We have to look for ways to stop the infection from spreading to others," he says.

Three approaches could work, says Paul Judge, CTO of CipherTrust Inc. Filter spam so it never reaches desktops. Train users to recognize and kill spam without responding to it. Outlaw spam and set up stiff penalties as a deterrent.

Spam can be limited by setting thresholds for cutting off e-mail from a single source address when more than a certain amount is sent per set time period, says Dean Drako, CEO of Barracuda Networks Inc.

Similarly, if a machine has been infected to generate spam, it can be cut off if it sends more than a certain amount in a given time period. About half of the machines used to spam are hijacked, says Hallam-Baker. Filters can also weed out spam based on key words.

Phishers, whose e-mails seek to trick credit card and other financial information out of victims, use sets of commandeered machines to send their e-mails, Judge says. CipherTrust finds that phishers use about 1,000 such zombie machines per day, then switching to another battery of machines the next day. These zombie batallions range up to 15,000 in number, he says.

Hallam-Baker suggests ISPs stripping off all executables from their customers' e-mails to prevent creation of zombie computers. "It's a completely irrelevant capability that is only dangerous," he says.

Laws against spam throw potential legal hassles in front of spammers as well as the threat of financial penalties and the result of having all their e-mail blocked, Hallam-Baker says.

The downside is that anti-spam laws have had little effect. California passed an anti-spam law last year, Barracuda Networks CEO Dean Drako says. "There was no significant impact on spam being sent on the Internet," he says.

Hallam-Baker suggests a three tiered registration system to stop spammers. Creating bonded senders could help by establishing a group of bulk e-mail senders who are likely not spammers. They would post a bond they would forfeit if they are caught spamming.

Senders would be authenticated via a light-weight DNS mechanism that would link the sender address to a small set of e-mail servers. If the address was lifted to send mail from a different server, it would be dropped.

Domain names would be authenticated at the owners expense to be certified as being linked to a legitimate business, giving ISPs and businesses another way of determining whether e-mail is from a spammer or not, he says.

A library of notorious spammers could be made public to enable spam hunters to further choose what e-mails to block, Hallam-Baker says.

The economics of spam are compelling. A direct mailing via the post office costs $1.39 per letter, while spam costs half a cent, Judge says. That means spammers need responses from one in 1,000 recipients, while direct mailers need 2 percent.

Regardless of how successful the spam fight is, it won't be any more successful than any other battle against an evolving foe. "We've been fighting viruses for 15 to 20 years, and that cat-and-mouse game is still going on," CipherTrust CTO Paul Judge says.