They've been talking on cypherpunks about various ideas for authenticating the sender of mail, and it brings up something I've been increasing my agit-prop efforts on lately.
Unfortunately, the ultimate way for any e-mail authentication scheme to work is for the whole message to be signed, headers and all, and then for the sending SMTP server to sign the message passes it on, headers and all, and so forth. The consequences of this approach, in terms of machine resources, not to mention privacy (plus or minus how close you can get to perfect pseudonymity), will probably be excessive.
The solution, here, I think, is economic, since this is, at root, an economic problem. It's also one we've talked about here at length on e$. That solution is postage.
If you use some kind of hash-collision microcoin protocol like Rivest/Shamir's MicroMint, or the "hash-cash" stuff people have been talking about on cypherpunks, you end up with a completely off-line, extremely small value digital cash system which could be used to pay for stuff like e-mail postage. The only thing you need to do is to encrypt the payment to the $MTP (:-)) machine, which then sends your mail. The only overhead is in handling the money, and, since it's handled offline anyway, the overhead is going to be much less than authenticating signatures either with an internal table, or, worse, in an on-line scheme of some kind.
However you do the protocol, you probably want to run it on a sender-pays basis, as that's where all the economic incentive to spam comes from. Actually, now that I think about it, a truly sender-pays system would mean that receiver-collects, and so the receiving $MTP machine would be the one which should actually get the money. This is not too far from the idea of getting paid to read mail, which has also been discussed on cypherpunks before. The receiving $MTP machine could even raise or lower its postage price depending on the load at the time, thus auctioning its processing resources to the highest bidder. People like e-mail spammers, needing to send messages to you very cheaply would have to wait until the price comes down. Some of them could wait a very long time. :-).
Frankly, I see no other solution to this problem in the long run except for postage, which means it's probably time to start figuring out, in earnest, how to make it all work. Whoever becomes the lowest cost producer of this kind of software stands to make a whole bunch of money.
Cheers,
Bob Hettinga
[ Back to the previous Rant ] | [ e$ home page ] | [ On to the next Rant ]