e$: The Book-Entry/Certificate Distinction

From: rah@shipwright.com (Robert Hettinga) 
To: cypherpunks@toad.com
Subject: e$: The Book-Entry/Certificate Distinction

At 2:30 AM 8/22/95, Timothy C. May wrote:
> (An important point is that in a cash economy, identity is 
>almost irrelevant. It's only in non-cash, or "account-based,"
>economy that True Names are demanded. Lots of interesting
>issues to discuss here, which I won't now.)


I'd like to vamp on this for a while.

I think Tim's talking about what I call the book-entry/certificate distinction. It goes right to the heart of finance, and it's why I think that the strong crypto/Moore's law combination is going to do more to free people than any political system ever could. (How's that for a wind-up to a rant?)

Accounting, well, double-entry bookeeping anyway, was invented in the late middle ages so people could measure their financial progress, but most importantly, so that people could trust other people doing the measurement for them: their bookeepers. This allowed the Renaissance entrepreneur (or more orignally an Italian nobleman, who was more of a gangster than anything else) to delegate financial measurement and consequently build a much larger enterprise than he could have otherwise.

Before double-entry bookeeping, the first book-entry system, everything was done on a cash, or certificate, basis, and, like the apocryphal Silicon Valley entrepreneur's bedroom shoebox, you knew you were making money when there was more in the till today than there was yesterday. The first certificates were actually the first form of writing: the original Cuneiform(sp?) of the Babylonians actually evolved from little pieces of clay formed into interesting three-dimensional polygons with a signature seal stamped on them, certificates, or abstractions of value, like "3 cows, so says I, J. Nebbuchenezzar", these were, in turn, used like money.

Book-entry allowed the creation of larger and larger enterprises, and it's no surprise that a hundred years or so after the invention of double entry bookeeping, joint stock companies, and eventually limited liability, came into being. The paradox here was that this was an enherently certificate-based system. People met in front of a buttonwood tree on Wall Street, and traded ornately engraved certificates representing some fractional amount of a company's shares or debt outstanding. To trade your shares, you had to physically send them to Wall Street to have them traded for you.

Eventually book-entry systems got into the trading of securities when stock exchanges were founded. Specialists or market-makers would take your order, and cross them on their books with their own inventory or with people on the other side of the trade, or, in the commodities markets where the trade volume is much higher, people stand around a pit, execute the trade with a contra party, and sign each other's orders.

This was compounded enormously by the advent of faster communications, particularly telecommunications. People could trade their stock from anywhere by wire, or by physically talking to a branch office of a securities firm who was itself linked by wire to an exchange through its main office.

In the early days of book-entry trading, delivery of physical certificates was done at various traders' operations departments, called "cages" for obvious reasons, to settle a trade. After a while clearing houses were instituted to do book-entry settlement of exchange-traded securities. In fact, when someone says "book-entry" these days, they usually mean book-entry securities settlement. A clearing house has all the physical certificates in a vault, and keeps books as pointers to all that paper to show who owns what. Those books match the books of the various traders, and through the miracle of double-entry bookeeping, everything balances so everyone agrees with each other. In fact, the need for all those certificates sort of goes away in this system, and when I first got on this list a year or so ago, Perry and I talked about the existence, at the Depository Trust Company -- the New York Stock Exchange's clearinghouse -- of a single certificate representing a company's entire common stock.

Okay, so what does this have to do with strong crypto? A lot. Every once in a while I've compared strong crypto and privacy to flight. Flying is an inherent good for lots of people, but what makes it really economically useful is how fast you go when you fly. People buy airline tickets because they can get anywhere much quicker if they fly than if they do anything else. The same with strong cryptography. Privacy, and the strong cryptography which makes it possible, is an enherent good for most of us, especially here on *this* list. But cryptographically strong protocols become economically useful when used for the creation of certificates which abstract value: digital bearer certificates like digital cash, and eventually digital certificates representing stocks, bonds, and various derivative securities.

To understand why, we need to make one further digression. Remember all those telecommunications lines going into the exchanges? The inherent geometry of a system where lines are cheaper than nodes is a hierarchy: you pick up a phone to call someone, and your central office routes you up to its central office, and further on up the hierarchy until the call can be switched back down somewhere to complete the call. The interesting correlation between this "switching" and most of our current social structure can be left for another list, but the reason for this methodology is all bound up in the economics of wires and switches: at first, wires were cheaper than switches, who, in the beginning, were people. We all know the story from there, people were first replaced with electomechanical switches, then with solid-state switches, then with microprocessing switches. When we got microprocessors, the economics of the nodes/lines problem got turned on its head. We run into Moore's Law.

Moore's Law, for Moore of Intel, is an observation that the price of a given semiconductor falls by half every 12 months, though it was 18 months when Moore figured this out. From a cryptological prospective, we understand what this means to the future cost of breaking a keyspace of a certain size. Moore's Law also means just as much to the structure of the network: in collapsing the cost of a network's nodes, it converts it from a hierarchy to a geodesic. Like Bucky Fuller's geodesic domes, a geodesic network consists of nodes connected to an arbitrary number of lines, with no real directionality to the motion of traffic: there is no up or down like in a hierarchy. A message could originate at the node, or could be just passing through in any direction to any other line the node is connected to. A good example of this is a company's PBX, which allowed intra-company switching of phone calls for large organizations over long distances directly, without going up and down the telephonic heirarchy. The abolition of monopoly in the telephone system was primarily caused by the advent of the geodesic network, and the phrase "Geodesic Network" was the title of Peter Huber's analysis of the phone system for Judge Green, the judge who broke up the Bell System in the early 80's. Huber's reccommendation to Green in 1986 as was to allow competition in the Regional Bell Operating Companies' core markets: the local loop. They're just getting around to doing it almost 10 years later. Another consequence of hierarchy, right?

Once again, the social consequences of geodesic information networks are interesting; ubiquitious computing, telecommuting, flattening organization structures, and lots of other phenomena can be attributed here, and even Mr. Archer's transaction-based tax proposal can. However, let's look at what it does to book-entry, particularly in the presence of strongly cryptographic certificate systems.

First, like Tim says, a book-entry system is bad for privacy. The IRS knows how much money you make, the bank knows how much you spend on your credit card, another book-entry system, even your grocery store knows what kind of food your cat likes. In a certificate system, like digital cash, no one knows anything, and it's cheaper that way anyway. Who did what to whom and for how much has no economic value whatever. No one can track the origin of all the bits of cash going through the till, much less audit it's ownership trail, and cryptographic protocols allow the secure handling of the money anyway, so the existence of the money speaks for itself, without needing to know who gave it to you. Book-entry stays where it belongs, on the store's accounting system, and there's reason to believe that as processors get cheaper and cheaper, there might be a micro economy (ecology?) in the making in an enterprise's various economic units, as a result of ubiquitous computing. Transfer payments have always been a vexing problem in large businesses, because there's no way to arrive at a fair price for intra-company trades of goods and services without looking to some outside market price. In fact, it's safe to say that the current paradigm for financial analysis, the efficient market hypothesis and it's offspring, CAPM, Black-Scholes, etc., is a solution to the transfer pricing problem which evolved into something infinitely more useful.

But let's take this idea of strong cryptography in a geodesic network to another financial abstraction, the securities markets. What you get then is a reversion to certificate based trading. You've gone back to the Buttonwood tree, only you're now trading digital certificates on the net. People announce their intention to trade, what they're willing to pay or accept, and, when the trade is made, people trade certificates under a strong cryptographic protocol, announce the price to the market, and walk away. People can "make markets" in securities by holding inventories and simply posting their prices bid or asked, to buy or sell their inventory of securities. Notice I said people here. What we're talking about is behavior which was originally done by *people* with *certificates*, and then by ever-larger *institutions* with *book-entry*. That is, until the advent of computers, when the pendelum swung back toward smaller entities. After all, in 1972, anyone with the money could get a Quotron or a NASDAQ Level 3 machine, and soon, at least the sell side of the market was competitive again. The stodgy old firms like Morgan Stanley and Salomon had to turn into very efficient carnivores to survive. Paradoxically, the buy side of the market, composed of banks, pensions, and mutual funds, had its development arrested by regulation and is just approaching the centralizing phase. However, we now know that the operation of both the buy and sell sides of the capital markets are going to get very interesting very soon, right?

With a geodesic, certificate-based financial system, the economies of scale caused by hierarchical communications systems go out the window. There is no need to concentrating processing power to be used to used on the market's information, because the information is everywhere without having to concentrate it. What's more, the specialization of processing enabled by Moore's Law creates such a flood of information that no centralized entity can process it all, and so it collapses under its own weight.

Everyone trusts their transactions because of the difficulty of forging certificates. That means that once again, a certificate has it's own inherent worth. It speaks for itself, and when it changes hands, the trade is, as Eric Hughes says, "immediately and finally" cleared and settled. The overhead of keeping books is gone, at least for the trading parties, and especially for the clearing houses, who, like exchanges, just kind of disappear, along with any way to regulate them. Somewhere, Joe Kennedy, the first Chairman of the Securities and Exchange Comission, is probably either crying his eyes out or laughing his head off, depending on your interpretation of his role in regulatory history.

This brings us to the reason securities exist, anyway. A stock certificate is really a legal document, right? How can you enforce laws in a geodesic environment protected by strong cryptography. Well, in a geodesic market, cryptographic protocols, instead of laws, enforce agreements. If the cryptographic protocol isn't satisfied, the software won't work. It's easy to see how a combination of certification technology and cryptographically anonymous voting protocols allow the direct election of a company's board members without proxies -- another hierarchical device -- for instance. It's easy to see how the presentation of digital interest coupons to a company's bond trustee could result in a payment in digital cash. This seems to work for just about any kind of financial instrument you could imagine. The presence of the digital certificate is prima facie claim upon the abstracted item, votes, interest/dividends, hog bellies, you name it, and the protocol could be put into software everyone could check themselves.

Obviously, we aren't talking about the end of bookeeping as we know it (BAWKI? ;-), but it does mean that days of book-entry as a means of social control are numbered.

Cheers, Bob Hettinga


[ Back to the previous Rant ] | [  home page ] | [ On to the next Rant ]