ietf-822
[Top] [All Lists]

Re: making mail traceable (was Re: Received header Considered Pathetic)

2004-01-17 12:59:03

I think this is a good discussion, so I'll do my best to fan the flames. :-)

Keith correctly and clearly stated the core problem: automating abuse tracing without providing non-repudiation. It's an impossible goal if stated in terms of absolutes, though, so let me restate my own vision of what the goal should be: providing as much automated abuse tracing as possible without non-repudiation, or (in less technical terms) to provide the best possible spam tracing without eliminating anonymous email.

This is a problem that needs to balance privacy rights with law enforcement. Historically, the way that open societies have typically dealt with this kind of issue is via the checks and balances of distributed control. We can go a long way towards that by using cryptographic tokens to validate ISP's, but requiring legal procedures to access ISP's records in the course of a spam investigation.

Thus, getting a million complaints about spam messages that were cryptographically shown to come through an ISP should be enough to get a warrant to trace the senders, but one or a few complaints should be subjected to a much higher standard, to prevent police "fishing trips" in the name of spam control. Automated traceability is a key to distinguishing between the two cases.

Of course, the previous paragraph would be more precise if, instead of "ISP" I had said "privacy-sensitive administrative domain." Any domain-administering entity could choose to make tracing information from within its domain completely private, as long as it took responsibility for working with the authorities when it proved to be the terminal publicly-ascertainable node to which major spam could be traced. Such domains would likely include large corporations and other institutions, including those that choose to run anonymous email gateways and fight spammers in their own ways (for example with computationally intensive challenge-response systems).

And yes, I realize that the above discussion completely ignores the jurisdictional issues, but I doubt that it will take more than a decade or two to work them out, and we need to think in the long term if we really want to control spam. -- Nathaniel

PS -- Am I afraid that all of this will help Big Brother? You bet. That's why I want to design as many checks & balances into the system as we can come up with. But I think the CANSPAM act has made it very clear that there are likely to be ever more detailed regulations governing email systems, and I think it would be a losing battle (and therefore arguably irresponsible) to oppose any particular proposal without having a more moderate alternative proposal to endorse. The US congress passed this law, however ill-informedly, because they correctly perceived a public demand to fix the problem of spam. Enabling Big Brother is, quite simply, the easiest, laziest solution to the problem, and therefore it is precisely what will happen unless we go to the effort of designing a less objectionable one. -- Nathaniel


<Prev in Thread] Current Thread [Next in Thread>