METROCAST BLOCKS RESIDENTIAL E-MAIL

[Jason Stephenson]

aluminumsulfate at earthlink.net wrote:

> Just off the top of my head...
> 
>  * Mandating SMTP AUTH
>  * Universal use of GnuPG + message signing
>  * HashCash (or similar systems) http://www.hashcash.org/

They're all hacks. The only *real* solution is something completely 
different.

> 
> In general, any spam-proof messaging system will follow these rules:

There's no such thing. Never will be.

> 
>  (1) By default, do not accept any messages

You can do that now, with greylisting, which eliminates the majority of 
spam and viruses. Greylisting means returning a temporary failure the 
first time that a new sender tries to deliver an email to your server, 
or it could be configured on a user by user basis. Spam agents and 
viruses don't generally try again, so those messages are never 
delivered. Legitimate MTAs will try again, so legitimate mail will get 
through. However, this won't stop "spammers" that use real MTA software.

>  (2) Accept messages from authentic senders

Who determines authenticity? If it's just that there's a key pair on a 
server somewhere, then there's nothing to stop spammers and viruses from 
creating their own key pairs. There's nothing to stop them making new 
ones when the old ones are revoked, or have no trust. (And AFAIK, only 
the key owner can revoke their own ky. I can't revoke yours and you 
can't revoke mine.)

>  (3) Retract sender authority if/when it's used to send spam

You've got that now with black lists, and you'll still need black lists 
with PKI. If you only trust keys signed by people or organizations you 
know and trust, you'll never get mail from strangers, who may want to 
offer you a real job, etc.

The real problem with anything designed to work with SMTP as it is, is 
that the cost of delivery and the cost of determining what's ham and 
what's spam is squarely on the recipient. It costs a spammer with an 
army of bots nothing to send out 1,000,000 emails. It costs the 
recipients of those emails in bandwidth, server resources, and even man 
hours to deal with the influx of spam. All of that adds up to money.

If the spammer had to pay for the storage of their messages before 
delivery (or pickup, rather), then spam would disappear very quickly. 
This is, in fact, what the IM2000 proposals have been about, making the 
sender bear the cost without adding some ridiculous email tax or 
micropayment scheme.

It is an extremely tough nut crack. Numerous proposals have been 
discussed, and there are many critiques of them on the web. (If you 
search for IM2000 discussion or proposal, I'm sure you'll find many of 
them.) Nothing that's been proposed so far seems adequate to me. Every 
proposal so far can be shot through with holes.

I'm starting to think that it is the very open architecture of the 
Internet that is the real "problem." At its very base, the 'Net is 
designed to be open. The basic plumbing was designed at a time when 
there were only a few thousand nodes, and the admins all new each other, 
more or less. You could pretty much trust everyone else to behave more 
or less responsibly.

Today, that architecture really makes it like a frontier environment. 
Each individual is pretty much on their own in protecting themselves 
form the hazards and predators of the environment. If you have an email 
server, you must run anti-virus and anti-spam software. If you don't, 
that's like a colonist in 1640 coming to the New World without a 
firearm. It's more or less the same for firewalls and whatever the 
latest whiz-bang security device is. It has gotten so that even on 
corporate, government and ngo LANs, you need firewalls on each machine 
to protect them from each other.

It's also a human problem. Some people just are not ready for a frontier 
environment. If it were a real frontier, those people who keep opening 
the virus-laden attachments in their email would have been eaten by 
wolves by now. Ditto for those people who have fallen for phishing 
schemes, etc. That is the Internet equivalent of being eaten by wolves.

Things are only going to get worse when IPv6 becomes mainstream and 
there are trillions of throw-away addresses.

What are the alternatives? Something like AOL or Compuserve before they 
joined the rest of the 'Net? No. There was abuse there, too.

I can't say for sure. However, I'm convinced that without completely 
redoing the network architecture so that it resembles a virtual police 
state (read: "prison or public high school"), then all bets are off. 
We're just going to have to deal with things as they are, unless someone 
has the cajones to pony up a better solution, and can convince 
1,000,000,000+ people to switch to it all at the same time.

Cheers,
Jason