Economics of "Wanted" and "Unwanted" Messages

[Tim May]

I'm reading with only 10% of my brain engaged the many (and 
oft-repeated) articles about how Hash$ash or MailBucks or whatever will 
"fix" the spam problem.

First, what is the spam problem? It's the receipt of unwanted messages, 
like a celebrity receiving unwanted attention when he walks in public. 
(Issues of his choice to appear in public are not central to the 
argument. He receives attention because his name and face is known to 
many. A "many to few" problem, just as more letters are sent to the New 
York Times than to me.  Ordinary people don't get bothered on the 
street because they are not one of the relative handful of celebrities.

With e-mail, we're all celebrities. Rather, the costs of sending 
(contacting) are so low that we can all become celebrities to a subset 
of stalkers.

The analogy with celebrities is intentional. Their are three main 
approaches for dealing with the problem of unwanted attention towards 
celebrities:

1. Pass laws making it a crime to approach a person unless one has good 
reason to. Courts decide who is authorized or justified. This is the 
"unsolicited e-mail is a crime" approach.

2. Celebrities have agents who screen all potential contacts for the 
right amount of attention, deals, cocaine, or other forms of 
"consideration." This is the HashCash approach. "Pay me to be 
interested."

3. Reclusiveness. Celebrities live behind guarded walls, or on their 
own Bermuda or Bahamas island estates. When in Hollywood or London, 
they stay in 5-star hotels with rigorous screening of who gets to their 
floor. This is the "spam filter and personal responsibility" approach.

Now it may be sad that Michael Douglas and Catherine Zeta Jones choose 
to live on Bermuda, in a walled compound no less, but this is the price 
of notoriety.

As for complicated schemes involving using CPU resources to compute 
cryptographic numbers and thus decide whether a sender was sufficiently 
motivated or not, this is nonsense. If I recognize a sender's name, I 
am likely to read his mail, regardless of whether my 1 GHz G4 decides 
that he did not spend enough of his own CPU resources in contacting me. 
Etc.

And if I see a bunch of Chinese characters in the message, or I see it 
is from a name I don't recognize, I will likely click on "Think of this 
kind of message as Trash" and will never again see this kind of 
message. (I use OS X Mail, a descendant of NeXTMail, and it has this 
kind of Bayesian filtering built in and easy to use.)

If and when spam from harvested messages gets to be too great, I expect 
I'll change my e-mail address again. (My spam dropped from about 100 
per day to about 1 per day when I changed my 7-year-old address of 
"tcmay at got.net" to "timcmay at got.net."  It hasn't gone up in the past 
few months. This suggest to me, BTW, that maybe a lot of the spam is 
using a relatively static list harvested a while back. Which makes 
sense. Those selling "500 million e-mail addresses for $29.95" have no 
real incentive to try to keep current: they are selling to dummies 
trying to sell fake Viagara sugar pills to suckers. Maybe they're 
selling each other the same lists of harvested addresses during the dot 
com peak? Just a thought.)

The fundamental flaw which is leading to the spam problem is that 
sending a million messages costs very little, not that recipients are 
not "charging" for e-mail. If bandwidth actually cost money, if various 
links in sending either JPEGs or copies of the Koran or e-mail actually 
were charged, a lot of the "ontological" problems with free mail would 
be lessened.

Notice that I am NOT saying that all carriers should charge. Indeed, 
whether they carry traffic for free or charge for it, and how they 
charge, is their business.

Whether on the uplink (sending) or downlink (receiving) or in the 
middle side, that some traffic is charged for, and must be paid for 
somehow, should be enough to fix things.

The reason we don't get inundated with hundreds of pounds of physical 
junk mail each day is that there are ontological limits on sending of 
physical goods, on consuming of physical resources. (And learning takes 
place. A few years ago I was getting 10 times as much weight in glossy 
catalogs. I surmise that some of the glossy catalog makers gave up on 
sending expensive catalogs, for which they also had to pay postage, to 
tens of millions of potential customers.)

On the downlink side, I expect my ISP is not much interested in trying 
to charge me or the sender for the e-mail I receive. But on the uplink 
side, on the side which fires off a million e-mails, I wonder why they 
don't try to charge the sender?

And why traffic in between, bouncing amongst nodes, is not charge for? 
(Such charges would eventually force backwards a charging mechanism, of 
whatever form, a la the arguments by Hardy, Tribble, etc. for "digital 
silk road." Nobody expected the camel caravan from Tashkent to Baghdad 
to carry stuff for free. So why is it expected that 50 GB of JPEGs and 
e-mail should travel for free on the Net?

Long term, unwanted communication comes with the "celebrity" effect. 
Greg Egan describes it well in "Permutation City": a constant barrage 
of agents attempting to make contact, with movies, fake voices, and 
every other possible trick. And so the spam filters will spend more 
time filtering. (And getting rid of the socialized "all bits are 
carried for free" pricing policy at least solves the worst of the 
ontological problems.)

Should crypto people be worrying about how nodes on the Net charge for 
traffic? Maybe. But it won't involve "HashCash" tokens. For the 99.99% 
who never bother with the complications of crypto and signatures, 
convincing them to learn about how to use HashCash will be 100 times 
harder a sell. And if some mail system tries to ram it down the throats 
of the users, they'll find alternatives.

--Tim May