I’m sure getting a lot of junk mail lately. In the war between spammers and spam filters, the spammers are winning. I remember Paul Graham speaking five or six years ago at the AI lab about his ideas for Bayesian spam filters. I don’t think there was a single person in the room who didn’t think, “But why don’t the spammers just send their message in an image?” Well, pretty much all mail clients and many institutional filter’s have implemented Paul’s ideas anyway. It worked for a good while, but now of course the bad guys are sending pictures. I feel that I’m missing something important in not understanding why it has taken them so long.
I also don’t understand why mail works the way it does. Why doesn’t my email client generate a new un-guessable unique return address for me each time I send out an email? Something like 7b22795f7aadfa4602e7c0ce189eb594@myAddress.net. When you receive the message from me, your email client should enter the address as a new address for me in your address book. You could send me mail using any of the stored addresses, or you could pass that address to someone else.
You should never have to see that gobbledygook. On OSX now, for example, mail and address book each display addresses as a lozenge with the person’s name. I can drag the lozenge around and even into the text of another mail, and the right thing happens.
But then suppose the unique address finds its way into the hands of a spammer. If I mark something as spam, my client should just revoke the address that it came in on. My client (local or ISP-based) would not display any mail received on a revoked address – regardless of whether it was sent by you, a third friend, or a bad guy. It should also bounce the message to the sender and (the first time) to the person I originally sent the address, so that their clients could remove the bad address from their address book. (It knows the person I originally sent the message to because it’s in my ‘Sent’ folder.)
Am I missing something? Is this really so hard?