Next: Under the hood Up: Technology behind SpamCheetah spam Previous: Technical Summary Contents

Unique Value Proposition

Internet e-mail has existed for quite some time now. In fact e-mail continues to be one of the most commonly used applications on the Internet. It has become an important business tool. The criticality of e-mail infrastructure for any business is obvious in today's world.

Spam however continues to dog us though spamming as a method to advertise and sell online was discovered much later in the evolution of Internet mail.

A variety of spam control techniques have been attempted by the smartest minds on the planet. And spammers have also been fairly good at circumventing even the best spam protection tools available today.

But along with containing spam with various levels of success we also face another important issue.

We lose legitimate mails occasionally if the spam control mechanism's internal clockwork goes haywire. In fact no one can guarantee you that you will never lose a legitimate mail.

Not all mails are lost due to spam control techniques however. But content scanning is a highly risky business. It is in the very nature of the math involved. Nothing much can be done over there.

We need something else.

We need a better tool or at least combine this technique with still better defences.

It is always nice to have more tools in our armoury especially since the adversary is well aware of the latest tools we possess.

In this arms race like in the case of cryptography and computer security, we have to attack at the level of human psychology.

In crypto, the keylength of encryption is made only as long as necessary to thwart a dedicated cracker. The cost incurred in cracking an encryption algorithm/security mechanism should be more than the value that one hopes to derive from the cost incurred.

Remember, it is a question of evolution and change. We can never know when the adversary will get the better of us. As our tools evolve so does his.

Spam control as a problem can be compared with computer security since we only need to make it unattractive for a spammer to offload his junk on us. Granted, he can always 'break our system'. But ultimately it is about the cost involved, it is about the motivation of the individual.

If we make it unattractive for the spammer to deliver his junk and at the same time ensure that the legitimate senders do not get affected in a big way then we are set.

Greylisting is a method to delay the delivery of e-mail since we force the mail sender to retry at a later point in time. Legitimate mail senders survive this acid test. But spammers don't.

Why?

Their business model does not account for the cost incurred retrying every mail. Theirs is a volume business. So what if you don't accept his mail? He can send it to thousand other unsuspecting victims.

A spammer does not have any particular interest in delivering his message to you alone.

Whereas a legitimate mail sender is very different. Moreover spammers run automated mail sender programs called as 'botnets' that do the sending.

Botnets are not written by overbright people but still they are written to ensure that they can thwart the defences of most of the spam control techniques.

But their motivation is limited to making money. If something does not make business sense they will not do it. This is the fact that we bank on for our defence.

Next: Under the hood Up: Technology behind SpamCheetah spam Previous: Technical Summary Contents

Girish Venkatachalam 2009-08-09