Financial Technology Evangelism
Boston, Massachusetts 


   Promoting Digital Commerce, Financial Cryptography, Digital Bearer Settlement


Rambo on Beacon Street: An Internet Message Clearing Corporation as an Economic Solution to Spam

The Geodesic Economy

Robert A. Hettinga

January 28, 2003

In January, 2003, MIT hosted what was supposed to be a small conference on spam. It got slashdotted and, after a room change to a large auditorium, a 500-plus standing-room-only crowd showed up, myself -- and most local television stations -- included. One of the speakers was Barry Shein, the founder of The World, the world's first internet service provider. I'd heard of him before, of course, everybody who's been on the net long enough has. I met him first at a DCSB meeting I had last June when John Quarterman came to speak. After the MIT meeting broke up, he and I got to talking about how the conference was -- with the notable exception of himself, and a lawyer from DC who sues spammers for a living -- about receiver-side filtering, and how filtering there did absolutely nothing to solve the actual problem of spam, which was, everyone agreed, a theft of services. What was needed was a way to price those services, and, more to the point, pay for them.

And, so, the following Friday, I spent the afternoon with Barry, and, that evening, had dinner with him and Mary Riendeau his unindicted co-conspirator in The World, and, along a good deal of hollering and stamping, a good time was had by all.

The next morning, at breakfast, I got with one of those genuine sprung-from-the-mind-of-Zeus ideas that I have had occasionally. Haven't had one in quite a few years, in fact. Makes me have faith in the old idea well, even. Unfortunately, it's not really my idea. It was, it dawned on me, Barry's all along -- and it is, if I may say so myself, a doozy...

Barry, who invented the commercial internet when he offered dial-up shell accounts to all comers in 1989, also has beautiful, gold-flecked, ancient and honorable Red Piranha (Serrasalmus nattereri) named "Rambo" swimming in a tank in his reception area -- along with the odd tail-nibbled and occasionally missing goldfish or two.

Like Rambo, who tends to prefer fish flakes to a live lunch these days, you get the feeling that Barry's mellowed out a bit, but only just. Like Rambo's namesake, Barry had to take a few body-shots in order to first invent the commercial internet in an environment of actual socialism, and then to survive the dot-com stock-scam bubble by actually earning revenue the whole time while *not* going public, and, finally, to thrive today in a world of still-merging internet service providers, even as economic common sense dictates otherwise, as Steve Case, if he still has a job next week, might attest.

Which, one supposes, is probably why Barry is still not only standing, but smiling, these days. That is, of course, when he's not tearing the lower-case-world a new one for not paying attention to the real electronic violence on the net, that self-same theft that goes by the brand name of a once-popular meat derivative.

Which is also how I found myself sitting next to Barry in front of a terminal session, with my eyes almost squinted shut, standing in the breeze while Barry bellowed at me that even "opt-in" senders of bulk e-mail should pay for it. And, when I noted this event to Mary some time later in Barry's presence, she started bellowing too, in her own fashion, and insofar as it's possible for someone as small as Mary is to actually bellow.

I figured I was in the presence of actual market information, here, and, as a good market "analyst", I'd better listen and figure out why they were, in fact, so agitated.

The reason I had actually sought Barry out that afternoon, getting a very nice free dinner at Brasserie Jo in the process, was because my solution to spam has, and, still is, postage. Not necessarily postage for the *delivery* of mail per se, requiring liability on the ISP as a common carrier, but actually postage payable to the recipient, to literally bribe the receiving SMTP machine, physically at the client in an eventual world of ubiquitous end-to-end always-on IP dialtone, to let the mail through to the email's intended recipient. Messages from someone not on a white-list with insufficient postage would be bounced as such. Voila. No Spam.

My sound-byte mantra until now on all of this has been "sender pays, friends mail me free, and all others pay cash", and I felt pretty smug about that one, the same way I felt about "financial crypto is the only crypto that matters", or "the tragedy of the commons is that nobody *owns* the commons", or "politics -- to paraphrase Heinlein on religion -- is dandruff", or any of my other self-concocted bon mots and neologisms.

And, of course, that postage would, of necessity, be in digital bearer form -- underwritten by IBUC, of course :-), or, more likely, by a huge market of underwriters just like IBUC so it would scale robustly -- using a financial cryptography protocol which transfers an asset instantaneously, executing, clearing, and settling a transaction all at once, and, for the most part, anonymously, or at least orthogonal to is-a-person biometric identity.

David Chaum invented these protocols first with blind signatures, useful for "large" value transactions from a dollar on up. For streaming internet events, we at IBUC were looking at Nicko Van Someren's statistically-tested Rabin signature coins, denominated in the millidollar range or less, sent down an encrypted tunnel of some kind.

Postage for SMTP email, $MTP, if you will, would probably be blind signatures of some kind, because, unlike streaming internet goods and services, music, video, or VOIP, each individual mail message is a statistically independent event, and so you can't use statistical process control to prevent double spending. It's possible that you could do something like Mark Manasse's Millicent, in which hash collision coins are issued by the SMTP machine, specifically counted there to prevent double spending and underwritten by a broker of some kind.

In the end, though, I expect that you would have to rely a lot on Moore's law to get caught up to do blind signatures that small, but I figured we were probably there, now. I'd even talked to friends at a very large ISP, a couple of hardware companies, and a mail-transaction maven about this stuff, and we were plinking spare cycles at a skunk-works project I was actually hoping to interest Barry in.

We'd already understood, for instance, that, frankly, if you had SMTP AUTH on your behind-the-firewall SMTP servers, you could stop doing stuff like port-25-blocking, and people on Cable and DSL accounts could run their own SMTP services locally if they needed to, and suffer their own slings and arrows without the ISP catching all the heat. We'd also known that if you signed mail, you could create a more-or-less bullet-proof way of white-listing your friends' mail. Finally, if encrypted all mail to the recipient's public key to begin with, you'd have to create a unique message for every recipient, including a reasonable proof of work.

All of which would certainly kill present spam in its tracks, without any need for postage, which required all of that just to get off the ground. Of course, getting people to do all of this was, to paraphrase Eric Raymond at the MIT meeting, not gonna happen, especially if it weren't deeply buried in the existing mail architecture somehow, which is why, at the moment, he liked filtering spam so much.

We weren't even looking at the "loading problem" yet, that is, how to withdraw cash, preferably from the Automatic Teller Machine network, or at least from book-entry internet fiat-currency payment systems like PayPal, or even commodity-payment systems like GoldMoney or E-Gold. Not as hard as it was three or four years ago, but certainly still a hard problem, as I said, to a tired audience of mail-filter people who didn't want to hear it, during the end-of-conference panel session.

In fact, this very morning, as I write this, for want of the "buck" part of Buck Rogers, :-), I was supposed to be in Guadeloupe, in the French West Indies, on a panel discussing exactly this kind of thing with the smartest guy my brother the mathematician ever knew of, Andrew Odlyzko, formerly of Bell Labs, now running his own institute at the University of Minnesota, and Ron Rivest, the 'R' in RSA, and the hardest working man in cryptography today. The moderator was, past tense now, moderated by Nicko Van Someren, CTO of nCipher, and probably the only guy with a real job in financial cryptography these days. While I got the impression I was the sacrificial lamb, or at least the Judas goat, I was honored to be invited.

Besides, since it was all of 3 degrees Fahrenheit at that exact time here in Boston, and I had actually *founded* the International Conference on Financial Cryptography to *be* in the Caribbean this precise time of year, I find the irony of the situation less than satisfying, to say the least.

So, there I was, listening to Barry talk about his idea of billing them all and letting God sort them out, which he and Mary have actually gone and done a couple of times, and how to actually sell high-volume senders of commercial email that that was actually a good idea, in a sort of inverse of Gresham's Law, "good mail washing out the bad", as it were. That, by actually *pricing* the delivery of commercial mail to his 10,000 customers at, everyone involved, especially would-be brand name buyers of that access, would seek to prevent illegitimate senders of commercial email from abusing what would then be an enclosed part of the internet "commons", through means legal and otherwise.

As someone who actually owns, in addition to a Red Piranha, the intellectual property rights to the 50th Anniversary of Rock and Roll (July 4th, 2004, according to an approximation of the right citation of Alan Freed), Barry was thinking about how ASCAP worked, where a statistical sample of songs was taken from every radio station and then billed out from a central clearinghouse, and how that seemed to be good model for email, aggregating everyone's bills in one place, to be sent, en masse, to senders of large volumes of commercial, promotional, or, other, email.

So, after noting that his search through his SMTP logs was coming up with people like the New York Times, and, as a recipient myself of that kind of mail, full of links to stuff on NYT's website, I blithely pop off about "friends send me mail free", and how his customers would be pissed off about opt-in lists not getting through whatever billing system was set up.

And, as we noted before, Barry bellowed, and I stood in the breeze for a bit.

After we both calmed down a bit, and after Mary came in and bellowed a little herself -- and even meeting the actual Kibo, in the flesh, checking off my list of net.legend.handshakes, including Captain Crunch, who I met at the MIT spam conference -- we went off to have dinner at Brasserie Jo, where, over steak, halibut, and chocolate mousse to die for, I blathered on about things like clearinghouses, and IBUC's work with CRESTCo, the central securities depository for the UK and Ireland, now in the process of merging with Euroclear, which is where all of the rest of Europe clears its trades.

And we talked about DTC, the Depository Trust Company, which does the same thing for the New York Stock Exchange, and how brokerages "cross" all their trades against each other on behalf of their clients, and then net out at the end of the day to a single number, positive or negative, per security per member brokerage. I talked about how, someday, IBUC could become a customer of DTC or the CREST system, and issue, to the net, in bearer form, unsponsored depository receipts to the net, just like it could against a PayPal, GoldMoney, or simple commercial bank account balance, denominated in dollars, or pounds, or yen, or whatever.

Remember, I'm still thinking about how to apply postage, a bearer transaction solution of some kind, to Barry's theft-of-SMTP-resources problem. Ultimately, to be paid in cash, but certainly with tokens on the mail to allow individual messages of the right kind through the spam-filters, and so on.

And, before I went to see Barry, Eric Johanssen and I figured that simply sending snail-mail bills to individual ad agencies' direct-mail people, cc'd in registered mail to the offending company's corporate counsel, would be a good way to start as a shot across the bow to a few publicly-held brand-name mass-emailers. The next stage would probably be an encrypted batch delivery of "flyers" that could be stuffed into TheWorld's email boxes, with guaranteed delivery of some kind by Barry and Company.

That last idea made Barry bellow just as much as he had later when I brought it up in that afternoon's discussion, and it wasn't because of whether his customers would bellow a bit themselves. You don't survive for 14 years as the original ISP without having a very sensitive nose for useless complication.

So around midnight or so, with a belly full of wine, steak au poivre and chocolate mousse, I was dropped off, and still surprisingly sensible :-), at Farquhar Street by Mary and Barry. I woke up the next morning, and, at of our standing Saturday breakfast date at Auntie B's in West Roxbury, I started explaining to my wife Carol what went on the previous evening. About halfway through, I said the word, "clearinghouse", and I was treated to the aforementioned fully-formed epiphany, a complete picture in my head about how all this could work. It was, for lack of a better word, "Kewl", and I said as much. I even immediately called Barry's number at the World, and left a message saying so.

Like all really great ideas, it's feindishly simple, and, if I say so myself, even beautiful. In fact, when I talked to Vinnie Moscaritolo, my alter-ego and Samoan Attorney, he himself said, "Kewl. Just use PGP-signed email and MySQL, hang it on a rack somewhere, and you're done". I called up Dan Geer, who I sat next to at the MIT spam conference. Dan, as a refugee of the academic networking security, Wall Street network security, and the dot-com certification authority wars, is currently CTO of @Stake, providing what could be called adult supervision for the white-hat hacker brain-trust formerly known as the L0pht. Dan's someone I could say "DTC for email" to, and, when I did, he said. "Yeah. That'll work. Force ISPs to do outbound filtering, too. Kewl."

I haven't talked to Michael Froomkin yet, but I bet *he'd* even go for it, if his paper in this month's Harvard Law Review on the IETF as Habermasian discourse is any guide.

So, what am I talking about?

Here it is, DTC for email.

Remember, it's simple. Gilding the lily would only make people who work for a living, like Barry Shein, bellow at you. :-).

You create a clearing corporation, just like DTC, since we're talking about institutional asset transfer, in this case liability the cost of email transmission. Not a non-profit association like VISA, but an actual business :-), organized like DTC, or the NYSE, or CRESTCo, or, even, the Federal Reserve Bank or the Bank of England.

The users of the system buy shares in the corporation, and then are charged fees for using the clearinghouse. It may be that, like the Boston Athenaeum, or even the old Puritan New England churches, there are eventually shareholders (or pewholders) and simple annual-fee membership, but, to capitalize the thing, you need money, and that means shareholders, so those are the first members.

The actual operation is, as Vinnie noted, just a box on a rack. Periodically, members mail a signed message to that box containing a summary, by domain name or IP address owner, or whatever works, of the top users of their inbound SMTP machine by number of messages sent. These mail messages to the server are actually invoices to those very counter parties, payable in cash, by some predetermined and standard settlement date. To keep *every* mail message from being accounted for, only the top-volume mailers will be summarized, because, in addition to some fixed yearly operating assessment, there will also be some reasonable fee per transaction.

Periodically, in what amounts to a two-dimensional matrix with domain owners on both axes, mail traffic is crossed and netted out. The World's mail to Earthlink is deducted from Earthlink's mail to The World, and the ISP with a the positive balance pays up.

If someone is not a member of the clearinghouse, they are sent a bill, by the clearinghouse, on behalf of member firms involved. When the money comes in, it is paid out to the appropriate member firms. If the bill is unpaid long enough, and it accretes a large enough balance, the clearing corporations hires a lawyer to collect it, or, better, just sells the receivable, which is frequently done today for all kinds of things.

That's it, and here's what you get when you do this.

First and foremost, the cost of email is allocated somewhere. Most important, it's allocated to the people who incur that cost, the senders of email. The more people send mail, in particular mail which games the current email "commons", the more they're liable for.

Second, it completely ignores the small, individual, "retail" sender of occasional email. It's absolutely certain that the market can only price large aggregates of mail, and not every single piece of mail. Furthermore, only spammers and commercial emailers will send so much mail to a single ISP that they will show up on that ISP's accumulated transaction radar.

Third, as you've probably noticed, it puts, for the first time, an economic value on the damage of spam both to specific recipients of spam, and across the *entire* internet; furthermore, it sends spammers, and more important, ISPs that enable spammers, a bill, which they must pay, or suffer economic, and eventually legal, consequences.

It charges up a giant legal-fee "capacitor", which any self-respecting lawyer would cheerfully pull the trigger on sooner or later, zapping spammers and their technological enablers, first with extremely documented lawsuits, and, if they persist, with criminal action, all without passing any additional legislation or promulgating any new regulations. The last time I looked, theft of service was illegal in all jurisdictions around the world.

Pretty nifty for a box on a rack, yes?

Of course, there's the mechanism for billing both inside and outside the clearinghouse, and legal research into, and probably even litigation about, the standing of invoices and the transactions on the database, not to mention the management of that capacitor full of pent-up legal liability. But, still, this is all well-traveled ground in the last couple of centuries of securities law, and, more important, in actual *millennia* of mercantile law going back to the fertile crescent and the very invention of writing itself.

This is The Net, to paraphrase Gibson, finding its own use for The Street's technology, if you will.

So, why is this interesting to me, of all people? What I'm proposing we do is a quintessential book-entry system, isn't it? The exact opposite of a digital bearer solution, the thing that I've said will revolutionize financial operations, and, by extension, creating an entire geodesic economy, a geodesic society, someday, to go with our emerging geodesic network infrastructure. I mean, debits are being swapped for credits in a database here. We're even using lawyers and, ultimately, the state's guns, to enforce non-repudiation of the transactions at hand, right?


However, think about this a little bit. We haven't said *anything* about how to *settle* those mail "transactions" currently being "executed" in SMTP, and now to be cleared in a clearinghouse. At some point, somehow, money has to change hands. In the first stages, checks in the mail will do fine. However, sooner or later, if my hunch about internet bearer transaction cryptography and transaction cost holds true, we'll get there from here sooner or later.

That's because we're not only creating, out of whole cloth, a clearing method for the execution of email "transactions", we're doing it on the net itself, which is, as IBUC has discovered in its research on creating digital bearer transaction markets, the absolute first step to the "dematerialization" if you will, the digital immobilization of book-entry assets, and eventual the conversion of those assets into digital bearer form.

Just like the first days of book entry settlement required immobilizing physical bearer certificates -- locking them up in a vault with armed guards around them so that, instead, you can transfer them on ledger books, then punch cards, and eventually databases -- you need to immobilize book-entry assets, by putting them into a custodial account somewhere, accessible to the internet, before you can use them to collateralize, to reserve, a given issue of digital bearer instruments. At DTC, to this day, even for so-called "electronic" issues of stock, there's a piece of paper in a vault, even if it represents the *entire* issue of stock being cleared on the clearinghouse's books.

This paradox was pointed out for the first time, rather jokingly, by Doug Barnes in the rump session of the *first* International Conference on Financial Cryptography, back in 1997. Doug he noted that the Mondex smart-card cash system, of all things, or something like it, would be required to collateralize and "load" it's exact antithesis, an issue of anonymous Chaumian blind-signature bearer cash.

We all had a laugh at Doug's 30 second presentation, featuring a single slide, done by hand, on the spot, with a green marker, as we were all sitting around drinking beer. Moving his hand down the slide, uncovering each point with a piece of paper, Doug noted that what he called "latency" -- float, in finance -- was the problem.

Since you're selling an asset predicated on zero settlement risk because it has zero settlement time, an underwriter must either expose himself to bouncing checks or charged-back credit cards, or he must not issue bearer cash until the transaction funding that issue of cash is cleared and settled. Doug noted takes 90 days to clear and settle a credit card transaction, before your money actually belongs to a given merchant; at least 5 working days to clear and settle a check (in Anguilla, where we were at the time, it took *30* days to clear and settle a check from outside the country); even in the capital markets it took at least three days to clear and settle a stock trade (though, at the time, in the UK, it was still 5 days); the answer was, Doug said, with a comedic pause... Mondex! Which got a big laugh.

Mondex smart-cards, he noted, were more or less instantaneous, even though, as a book-entry method, they still relied on, as Doug had noted for the press earlier that day, "and then you go to jail" as an error handling step -- something that doesn't work at all well for an internet transaction protocol, something a few of us have taken to calling Barnes' Law.

Yet, here I am, proposing exactly that, for the sending of email, and I think you can see where I'm going.

It's a marvellous paradox, straight out of a middling Star Trek movie plot. It takes Nixon to go to China: Just like you can't bootstrap cellular/wireless systems without wired telephony, or a packet-switched internet without a circuit-switched telephone system, we can't have internet bearer transactions until we have internet book-entry transactions, transactions that are not merely executed on the net (or, in the case of credit cards, *instructed* from the net to be executed, cleared and settled *off* the net), but actual transactions which are executed, cleared and *settled* on the net, even if they're only done so in book-entry form. As I noted before, PayPal is a good example. So are the various digital gold currencies out there, starting with E-Gold, and, more recently, GoldMoney and EBullion.

They are important, because when people start to use these book-entry methods on the net, the necessity to converge to instantaneity is inexorable, and sooner or later we converge, I would contend, on a pre-authenticated digital object with only one copy in one place at one time, secured by financial cryptography, that does not require -- actually forbids -- the identity of the buyer or the seller in order to execute, clear and settle transactions transferring the ownership of that asset. Because of those characteristics, including functional anonymity, it does all three steps at the same time, enormously reducing risk in transaction settlement and, dramatically collapsing transaction cost in the process.

That, in the long run, is where I think we're going with this idea. For large, high-volume email senders, a clearinghouse makes economic sense, and it makes perfect sense right now. As the size of transactions falls, the cost of those transactions has to fall accordingly. And, because I'm betting that internet bearer transaction costs will be three orders of magnitude cheaper than book-entry ones some day, *that* will be how recursively smaller, and more geodesic internet email, with always-on IP dialtone, and client-side SMTP, will work.

So, enough about me, or even what *you* think about me, :-), let's think about what Rambo, as good a pet name as any for our quick and dirty institutional SMTP-transaction clearinghouse, does once it's grown up to ancient, honorable, and, hopefully, long, mature adulthood, and work backwards from there to the present.

In its venerable old age, say 5 or 6 years out, like its namesake in the tank on Beacon Street, Rambo eats mostly fish-flakes, that is, it crosses traffic between its members. Occasionally, it bites a piece of tail off of one of the denizens in its fishtank, a brand-name commercial email user who's not a member of the clearinghouse, by sending them a bill. Once in a great while, if necessary, the odd goldfish disappears, and everybody, Rambo, his owners, and his owners' customers, are happy to see it go.

In middle age however, Rambo the clearinghouse is a natural born killer, travelling in packs, making various fish, rodents, and the occasional pig -- in the form of people who host free-but-not-really porn sites, Nigerian Spanish-prisoner scammers, penis enlargers, and herbal Viagra vendors -- disappear in a boil of financial, even criminal, bloody water because they can't get access to free, high-volume SMTP service anymore.

And, in the beginning, like Rambo in his callow youth, the clearinghouse performs opportunistic involuntary tail-surgery services on behalf of its shareholders on otherwise legitimate brand-name bulk commercial direct-email advertisers, like, say, free-as-in-beer opt-in email newsletters pointing to the advertising and subscription-supported web-sites of various newspapers of record.

The fun part about this is, you can start Rambo off as a small-fry, and he'll take care of himself. Even with one or two members, Rambo is pretty self-sufficient, and, as he grows, Metcalfe's law starts to kick in. And, as we noticed, Rambo can even play well with others, as Red Piranhas are wont to do in the wild, by sharing information and aggregating liabilities, particularly aggregious net-wide liabilities, with other competing SMTP clearinghouses, in the same way that Rambo and friends would go after a larger bit of dinner that happened to go swimming in the wrong place some pleasant Amazonian evening.

It shouldn't cost too much to get Rambo in the water. You need a box on a rack, like I said, running MySQL and able to answer PGP/GPG-signed email. You need to make sure the law is on your side. You need to sign up some shareholders, and you hang out a shingle.

Those are, to my mind, the next steps here. If you're interested in this, email me, and, when I pick you out of my spam filter, :-), we'll talk.