Paying for Email with Computational Time

What if it cost a spammer $.10 to send each spam? They'd stop. The sell-through rate of spam is very low. At a dime per email it would simply become unprofitable.

Unfortunately, it would be difficult to charge spammers without hitting ordinary users.

But what if all computers were somehow magically limited to only sending 10 emails per minute? That would be plenty for most users, but it would make sending junk emails extremely difficult. Major companies that needed to send mailings could rent a farm of servers to send the email.

It would be very similar to charging postage for email, except it would be effectively free for end-users.

But how would this system to block spam e-mail really work?

As fast as computers are today, there are some things they can't do that quickly.

It's pretty easy to pick two large primes out of a list and multiply them. The result is called their "product." However, it's extremely difficult to "factor" the product back into its prime roots.

Time for a math test - feel free to use a calculator.

What two primes does 837 factor into?

It's not easy.

Now, what's 27 * 31?

Much easier.

So I picked out two prime numbers, multiplied them and then asked you to come up with the original numbers. Hard for you to do, but it's easy for me to find out if you came up with the right answer. This is referred to as a "one-way" equation.

The bigger the numbers, the longer it takes. For a given size number, it's possible to come up with an estimate for how long it will take to factor. Just to throw out an example it might take a 3GHZ Pentium 10 seconds to factor a 20 digit product of two primes (this is just for illustration, I have no idea what the actual time would be).

So, by requiring the computer sending email to factor a specified product of two primes, you're effectively forcing it to pay a "computational tax" for each email it sends. The receiving computer can easily verify that the answer is correct. If the answer checks out, all other spam blocking software could be ignored. False-positives and spam would be things of the past.

In the simplest implementation, the source of the product would be from a "challenge" by the receiving computer. If the sending computer fails to factor the prime, the email would be deleted by the spam solution.

More pragmatically, this technology would be used along side existing email systems. The exact details will be included in another article.

Back