The Great Email “Calculation Debate”
Some readers of my proposal for Spam Guarantees believe that my solution is too complex or costly. I concede that simpler approaches, such as authentication will be tried first, and will provide temporary relief, but ultimately a market-based solution will be necessary.
When such a market is created, it will appear gradually, and it will need to use real money, or it will be unable to police fraud.
It will not eliminate all spam, but offer a way for legitimate senders to avoid becoming spam filter false positives.
Here’s an overview of my views on whether market-based solutions should use “Computer-Time” or “Real Money”. The issue reminds me of Hayek and the “Calculation Debate“.
Anti-Spam Currencies: Computational vs Monetary
Two of the biggest obstacles to implementing a market-based solution to spam are:
- Expense: the cost of administering and billing for each email transaction; and
- Fraud: the inevitable attempts to capture the newly created currency
Expense
The “computational” school of thought argues that a postage or guarantee system that uses real money will cost too much to administer.
Accounts will need to be created, tracked, and billed. With so many transactions, the overhead will be enormous.
A system that requires computers to solve a math problem will require no accounts, tracking or billing.
Fraud
Both approaches are vulnerable to fraud.
If the system uses real money, hackers will surely attempt to steal a sender’s key and use it like a stolen credit card.
If the system treats computer-time as currency, hackers will surreptitiously commandeer computers to poach the postage.
Scope Modesty: Certified Mail for the Web
With my proposal for email guarantees, I don’t pretend to the immediately eradicate all spam forever. Imagine instead a system that starts as a premium service, much like current-day certified and registered mail.
People and businesses who really value the email they send will set up accounts to guarantee their mail. This will ensure that their messages make it through the spam filters unscathed.
Not all transactions will need to be tracked for billing purposes, only those messages unwanted by their recipients.
Monetary Incentives
Because senders have to pay for “cashed” email, they will limit the unwanted messages they send, limiting the transaction load that their Credit Company needs to process. The Credit Companies can charge a fee to cover the cost of processing the “cashed” email.
To limit the amount of fraud, Credit Companies will track the amount of guarantees that have been collected. Accounts that reach their limit will be frozen.
Credit Companies may offer to absorb the cost of stolen keys, provided that account holders follow certain security practices. If not, the account holder would be liable, up to the limit of their account.
Computational Incentives
In the computational model, senders and computer owners take no responsibility for securing their systems.
A hacker could install a program on my computer that allows him to offload email computations to my computer.
If my credit card is not being charged for the postage he counterfeits, and the hacker is smart enough not to make my system unusable, I am unlikely to devote the resources necessary to:
- avoid getting the program in the first place
- removing it, once I have it
Holding someone liable for poor security creates an incentive to limit the exposure of currency and tighten up security.
Some people may decide that their security is so poor that they are unable to offer a guarantee.
Their messages may still make it through the spam filter, but they run a greater risk of being overlooked.