ietf-mailsig
[Top] [All Lists]

Re: Anonymous signed mail

2004-08-29 08:22:02

On Aug 18, 2004, at 4:59 PM, <Atul(_dot_)Sharma(_at_)nokia(_dot_)com> wrote:

To dissuade spammers, we shall need a way to be able to identify
them direct or indirect.

Not necessarily. I agree that in practice most "anonymous" mail really is traceable, i.e. "indirectly traceable," but you can make tracing require rather more than a casual effort. More abstractly, though, sender identification is only one of three major general strategies that people are taking to fight spam:

        -- identifying spam senders
        -- filtering spam content
        -- disrupting spam economics

I think most of us would agree that the second strategy, filtering, is a running-in-place palliative rather than a solution, with Moore's Law working against it. Unfortunately in most circumstances today it's almost the only thing we have, which is why we're here trying to build something else even while most of us explore ever-cleverer filtering as well. Most of the non-filtering approaches, including MASS, are in the "identifying senders" category, but I believe that the third (economic) alternative has its value, particularly in facilitating anonymous mail.

There are still a wild variety of economically-oriented proposals, most but not all of them nearly hopelessly implausible. Schemes that seek to charge postage for every email message are a long way from reality, but there are other approaches. At the ISP level, a simple metering strategy can change the economics of spam originating with customers who are either spammers or zombies -- simply give each email account an outgoing mail quota, set generously enough to let any non-bulk mail users just slip on by, and with special rules/mechanisms for mailing lists. This requires no standards setting at all, and is being implemented in primitive form by several mass market ISPs already.

Most interesting to me, however, is the use of computational challenges for the purpose of "economic filtering" of spam. I'm willing to let anyone in the world send me email if they're willing to invest, say, 60 CPU seconds in the effort, because I know it isn't worth it to spammers. That's no big deal for legitimate users, although clearly they'll prefer to avoid the 15 second penalty and use authentication-based approaches when they aren't trying to be anonymous. But such computational challenges are, to my mind, the key technology that will permit us to preserve privacy in an era of stricter spam control.

None of this places any limits on what we're trying to do in the MASS group, but I'd prefer for us to try to view our authentication mechanisms as part of a larger picture that preserves computational or economic alternatives to sender identification. And I'd particularly like to try to build a consensus understanding that anonymity *can* be preserved in an era of spam control, rather than letting spam be used as an excuse for killing off anonymity. -- Nathaniel


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