e$: Interbank Digital Cash Clearing, Better Living through Walletware, Microintermediation, Net.Currencies and ECM


Robert Hettinga
44 Farquhar Street
Boston, MA 02131

Comments: rah@shipwright.com


June 3, 1996


People have asked me what I think about interbank digital cash clearing, and I say I'm all for it. Flippant comments aside, I'm not sure I can add anything to the discussion, but I'll take a shot at it.

It's easy to argue that there's no real need right now for on-net interbank clearing, especially with online cash. If you handle the transactions right, the merchant just verifies and deposits cash as it comes in. This works especially well with the underwriter-as-internet-ATM-machine model I like to use in discussion. (I include that here only by reference, as I've beat it to death. The e$ home-page is http://thumper.vmeng.com/rah/ ) If the trustee behind the underwriter is, well, trusted, to do deposits as well as withdrawls, then the "problem" of interbank clearing on the net itself goes away. Actually, we just translate it into "normal" world of book-entry banking transactions, and the trustee does the interbank clearing there. Same as it ever was.

In this scenario, if the customer presents a certificate requiring change, the change can be issued in the same transaction as the deposit of the rest of the transaction's proceeds. Frankly, in a world of ubiquitous networks, this is clearly the way to go for merchants of any means whatsoever, because it's completely safe for them, and it doesn't matter who the underwriter of the cash is. With the proper level of software abstraction, it may not even matter what the cash protocol is.

However, networks are a long way from ubiquitous, and that means offline transactions for a lot of cases. Fortunately, if we want to think about offline transactions, we can pilfer some more ideas from, as usual, the market for traveller's checks. When a merchant deposits a traveller's check, her bank clears it like any other check. That's possible because not only is the check secured, but it is hard to replicate. Most importantly, American Express offers a virtual guarantee of payment even if the checks are proven to be lost or forged, which speaks volumes about the susceptability of the system to any large-scale fraud. However, in the unlikely event that AmEx goes out of business, the merchant, not the bank, would be left holding the bag, just like she would be with a bounced check. Obviously, the same thing happens with money orders, a market in which there are many more issuers. All banks issue money orders, but traveller's checks require large consumer marketing campaigns, so there are fewer underwriters of them.

So, if it is possible to build linkages between digital cash and the checking system through the merchant's bank, we might have another solution. Fortunately, this is no problem. A merchant takes in various forms of off-line cash all day and deposits it electronically at the time she deposits her other money. The bank validates, clears, and deposits the cash right there at the teller/ATM, through it's connections to the net. This is (barely!) analogous to what happens to a traveller's check or money order, which is deposited and sometimes physically flown to the issuer's bank to clear, and paid back to the depositor by fed wire to settle the transaction. Making change for the transactions themselves in this scenario is problematic, but the merchant can just hand back physical cash, just like she does with traveller's checks.

An all-electronic variant on this "mixed-money" change method is for the merchant to hand back whatever digital-note change she has, regardless of who issued, and this implies a lot of preconditions, but it's not as scary as it looks, which we'll see in a little bit.

Right now, when we talk about the need for interbank clearing, a lot of people seem to be talking about offline cash. As we saw, online clearing is pretty much a geodesic process, because you're connected straight to the underwriter of the cash you're using for the transaction under way. It's hard for me to see any need for on-net intermediaries for online cash, at least for the time being, because the trustee, the actual financial intermediary in the transaction, is hooked into the "physical" financial system of banks, ATM networks, and central-bank fund-wires. I'll talk about on-the-net financial intermediaries a bit later, though, and there might be one wild card, but I'm not sure.

In the same way that I claim that certificate-based clearing and settlement is always going to be cheaper than book-entry methods, I think that offline cash is always going to be cheaper at the margin than online cash. The risk of offine transactions is always going to be double-spending, and that has to be traded against convenience, lower cost of not needing net access and the absolute anonymity you can get by not having to ever reveal yourself to even the underwriter of the digital cash in question. My claim of "marginally cheaper" for offline cash is strictly because of the lack of a net connection, but this cost-component will continue to fall into the forseeable future, so it will be less of an issue. Today, however, with only tens of millions of people on the net out of a world population in the billions, off-line digital cash has a significant cost advantage, and as long there's a bank with in a few transactions' proximity to the offline transaction in question, I think it can still be pretty safe to do offline digital cash transactions. Precicely because we're only a transaction-hop or two from an on-line transaction, double spending is reduced to a physical phenomena at the smartcard-to-smartcard level, which is much easier to deal with. The hoary old bugbear people like trot out at times like this, that of a bajillion salami-slice transactions done simultaneously all over the net all at once, goes back under the bed -- or back into the monster closet -- where it belongs. (Speaking of bugbears, remember, the problem of someone stealing a bank's key, and literally printing free money, is more a problem of issuing digital cash batches with expiry dates than anything else. More to the point, the problem belongs to a single underwriter, and not to a robust market with many competing underwriters.)

Anyway, the upshot of the double-spending problem is that it assumes a world of strictly offline transactions, which is almost as ludicrous as a world of strictly online transactions. ;-). Any robust and marketable system of digital cash will need to be able to do both.

Like I said above, the transaction handling mechanisms are everything in a multi-underwriter, multi-trustee, and even multi-certificate-protocol regime. We need to have a set of standards for digital cash clearing and settlement which makes the actual issuer as transparent as possible to the transaction's participants. After all, that's what we have with checks. The bank the check's drawn on doesn't really matter. To a lesser extent, neither does the issuer of a traveller's check or money order. Who printed the check certainly doesn't matter.

That's about where we want to be with digital cash. If I want to buy something from you with ecash, the last thing we should care about is the mechanics of the transaction, and that includes who the issuer of the cash is, who the trustee is who's going to do the interbank settlement out in the book-entry world, who the protocol designer is who invented the type of cash we're using, who the software developers were who developed my wallet and your register. None of that. We just want to settle the trade. Better living through walletware.

That means that we need think about multi-underwriter clearing from the outset, preferrably at the merchant level, and I'm sure that's what Digicash is moving towards. Digicash's walletware and registerware, or anyone else's for that matter, should be able to transact business with ecash of any form, without discrimination. If I spend some ecash with you, your cash register should take any and all combinations of my Deutche-ecash, or my Twain-ecash, or my Finn-ecash, validate them online with their underwriters (whose responsibility for transaction clearing turnaround should make speed a major selling point to their customers), and hand me back change however I want it, choice of underwriter (or not) and all. In fact, if everything works out, and markets for digital cash underwriting become efficient and competitive, then who underwrites my ecash becomes less of an issue to me over time. One form of ecash is as good as another, because it all interoperates. So, if we assume that walletware takes different protocols, it seems that the dominant digital cash protocol would be that which operates best from the underwriter's standpoint, which is where it should be, since they're the ones whose reputations are being risked, "rented", as it were.

Okay. Let's look quite a few years ahead, to a time when most money that comes onto the net stays here and just gets moved around, to digital bearer bonds, or to digital mutual fund certificates, or whatever. Let's say that Tatsuo Tanaka's scenario has come to pass. That is, because the money's not leaving the net as soon as the transaction takes place, like we've been talking about, the digital cash underwriters and trustees, most likely with the knowing collusion of users who want to pay lower purchase discounts, start "creating" money by issuing cash against fractional, instead of 100% reserves. More to the point, to follow Tanaka some more, a panic or two brought about by these shenanigans has caused the underwriters to police themselves by creating some kind of independent currency control for various associations of fractional-reserve underwriters, much in the same way that fractional reserve bank panics were handled prior to the advent of central banking. We may even have several full-blown internet currencies, controlled by currency boards of some kind. I'm not talking about nation-states, here, either. All of this could be done on a private basis. Real live private currencies, offered by an association of digital cash underwriters.

This is a long way off, but do you see what I'm getting at here? What we end up with is an on-the-net interbank clearing system, doing just what the book entry system does now, only without governments or central banks in the middle. We have intermediaries in the form of a board which manages the quantity of a given private currency based on "foriegn" reserves, that is, the holdings in its member banks in the other currencies, which prevents monitary inflation, and that's one place where the "interbank" clearing takes place. In this environment, we also have currency hedgers and speculators, who make (or save) money by trying to figure out where currencies of various kinds are going to go relative to one another. These days we have institutions moving large amounts of currencies around, doing insecure trades on secure networks. However, the technology of digital bearer certificates and ubiquitous public networks could enable a legion of very small autonomous entities to do the same kinds of activity that the big boys do now. So we don't get "disintermediation", which lots of people see right now with the merger of your local savings bank into a big conglomerate, or your local stockbroker getting bought or put out of business by a discount broker or mutual fund. We end up with "microintermediation".

Actually, I've written about the same idea elsewhere, of underwriting "bots" providing ubiquitous auction markets for things like personal digital bearer bonds, etc., but in this case we have a bunch of trading 'bots making secondary markets in currency, hedging and speculating on price fluctuations for money, all like a bunch of microscopic George Soroses. Like George Soros and his famous takedown of the EU's exchange mechanism, these bots would be who actually "determine" the price of a given net.currency versus another, and not the net.currency boards at all.

Okay. I've wandered way out here on a limb, and I'm going to climb off of it soon, before I either fall or someone cuts it off. :-). However, before I go, let's look at something which actually happened, and which may be a pointer to what could happen again soon, without too much trouble. I'm talking about ecm, or the electronic cash market trading list.

Last summer, when I got back from my New Orleans trip, I was up in Montana hanging out while my wife was at an educator's conference, and Lucky Green sends me e-mail about having just sold, for cash, some of the demo "cyberbuck" certificates that Digicash was issuing at the time. I commented about this to cypherpunks, one thing led to another, and the next thing I knew, Rich Lethin had started up a mailing list and set up a protocol for trading these beta-certificates for cash over that list. He named the list ecm. (Send "info ecm" in the body of a message to majordomo@ai.mit.edu, if you want to see what the fuss was all about.) It was used sporadically up to the time when, you guessed it, Mark Twain came on line with actual digital cash, and people stopped trading beta-certificates altogether. I can't even remember what the last settlement price was, but it was pennies on the dollar. Actually, now that I remember it, there was a period where the beta-certificates were traded for real Mark Twain ecash on someone's web-page and then announced on ecm, but things have pretty much gone moribund on ecm lately. I haven't seen a trade go across in many months.

ECM is the "wild-card" exception to using the current banking system for interbank digital cash clearing. The one that I was talking about above, after I said we didn't really need an interbank clearing mechanism on the net itself. If someone wanted "clear" these different versions of ecash on the net some day, they could just take positions in both the Finnish, Mark Twain, and eventually Deutchebank certificates, and run a little currency exchange operation for fun, and maybe profit, between the three by announcing their bid/ask prices on ecm in both certificates. Hint: buy low, sell high. :-). They could even do currency-speculation-by-proxy by taking different positions in these certificates if they wanted to. It may even be that someone could do this and make a living at it, someday, if enough digital cash was used on the net, particularly if they could do it cheaper than it would cost someone to "deposit" their cash, through an underwriter, into their own bank account.

Would income or capital gains taxes be considered part of that deposit "cost"? As Francis Urquart (RIP) used to say, "You might say that, but I couldn't possibly comment."


Bob Hettinga


Updated: June 13, 1996


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