dbts: Geodesic Payments

[Robert Hettinga <rah@shipwright.com>]

-----BEGIN PGP SIGNED MESSAGE-----

At 5:49 PM -0500 on 2/19/99, Ben Laurie wrote:


> Robert Hettinga wrote:
> > 1. Geodesic payment on geodesic networks.
>
> OK, so what exactly does this mean?

That's kind of a FAQ, but I feel a much more recent answer coming on. :-). If
you want to see a bunch of *old* answers, see <http://www.shipwright.com/>.

"Geodesic" means pretty much what it does in a dictionary: "the straightest
line across a sphere." Bucky Fuller used the word, more or less correctly, to
describe his domes, the design of which which are now considered to be in a
class of structures called, oddly enough, "network structures." :-). He called
his domes geodesic because, when you push on the dome at some point, the lines
of force are transmitted through the various pipes and connectors in a radial
direction, away from the point of force across the sphere to the ground on the
opposite side. Geodesically, in other words. Of course, all spheres and domes
work this way, but Bucky, being moderately hyperjargonate, needed a new name
for his new toys, so he went for it. :-). With Bucky's dome structures, the
more connectors, nodes, in the structure, the "straighter" this line of force
goes.

(The fun bit is, Bucky Fuller was trying to emulate bubbles, which he
considered the most perfect structure in nature, but what he got, instead, was
a "perfect" form of soot, :-), which has an incomplete geodesic structure.
Geodesic carbon has been found, or, more to the point, built, and they call it
"Buckminsterfullerine" in his honor. Some people call it the first
nanostructure. I think its formula is C60 or something. People are now
apparently constructing Buckytubes, and they call them "nanotubes" these days.
They are even threatening to use them in computers someday, if I remember
correctly. )

Anyway, I mean "geodesic network" in the same way that Peter Huber does when
when he coined the expression writing for Judge Green on the progress of the
AT&T breakup in 1986 in a report called, oddly enough, "The Geodesic Network".
:-). A good read, and in it resides the core of my claim that our social
structures map to our communications structures, though Huber would probably
wince, like a lot of people do when they hear me say it. :-).

This bastardation of poor Huber results, of course, in my use of such
atrocious neologia as "Geodesic Society", or "Geodesic Economy", or
"-Politics" (crypto-anarchocapitalism? ;-)), or "-Warfare" (terrorism, maybe,
but more like the the stuff in the Army's "The Mesh and the Net", which can be
found somewhere at the bottom of <http://www.shipwright.com>, though the site
itself is moribund these days), and so forth.

Okay. So, what's a geodesic payment, then?

Well, if you look at one of Bucky Fuller's domes, it's made, like most stable
structures, more or less out of triangles, with each triangle consisting of
three connectors and three sticks, or tubes, or bars or whatever. In
telephony, Huber likened switches to the connectors, and telephone lines to
the tubes between the connectors in a geodesic dome, claiming that Moore's
law, by changing the cost ratio between telephone lines and switches, changed
the telephone network from a hierarchical one, with big expensive switches
wired into a giant hierarchical pyramid, to a geodesic one, with lots of cheap
switches wired into a geodesic structure. Direct lines between the main office
and the branch PBXes, for instance, instead of everybody calling each other on
Centrex lines over the hierarchically arranged telco network. Now you can see
why I say that human organizational networks of more than, say 14 (an
admittedly savanna-derived magic number :-)), become hierarchical. Human heads
are expensive, and, as always, talk is cheap. ;-).

And obviously, when you look at a map of the internet, it looks like the
mother of geodesic networks, especially at its lower levels and earlier
stages. With the advent of backbones, we started to see hierarchy, but, I
think, and heretically so, :-), that that's more a symptom of a book-entry
economy than anything else. That is, backbones need things like peering
arrangements, which require lawyers, and finance guys, and network engineers
to figure out and implement, and, ultimately, cops to enforce, if people lied
about book-entries, or packets switched. In the old days of the net, people
just bought a direct line between any two routers based on load and the cost
of the line, and that's the way the packets went -- oddly enough, that meant
more or less geodesically, if you were ignore geography and "flatten" the
network out into equal "length" lines. These days, economics plays more a
part, but the switching of that economic information is hierarchical, and it
maps to the net. It's easier, from a Misean "market-calculation" standpoint,
to buy a big switch and a big line, only talk to people of a certain size, and
transfer-price the asset instead of auctioning it in realtime.

It's this problem of transfer-pricing which gave us most of modern financial
theory (only the market has the "right" price of an asset), and, of course,
killed command economies like the Soviet Union.

I claim that if we could have router/switch/whatever based cash settled
auctions for bandwidth, we'd end up with a much more cost-effective, and, you
guessed it, geodesic network, because routers would have actual
profit-and-loss responsibility, if you will, and would be selling their
processing capability into more efficient markets than they would under
transfer pricing. You'd get more smaller routers hooked into smaller lines,
with many more lines per router. The "straightness" of the line across the
geodesic network would increase, and there would be more lines radiating away
from a given load, just like what happens to force in a geodesic dome. Thus,
net would be more robust, stronger, just like the American economy, with,
arguably these days ;-), more auction processes than that of the old Soviet
Union, was more robust than the Soviet one.

That's where I get all my "digital silk road"-decended jokes about routers
which buy packets low and sell them high across the network, or underused
routers which buy a line to a busy routers to make more money, or busy routers
which save enough out of their net income to buy a copy of themselves, and so
forth. :-).

Okay, so, again, what's a geodesic payment? Remember those triangles. A
geodesic payment is the financial equivalent of a single-intermediary peer to
peer transaction. In meatspace, for instance, could be a cash transaction,
because I give you a dollar, you know by inspection what a dollar is, so you
give me what I bought and put the dollar in your pocket. These days, the
Federal Reserve Bank, the underwriter of the dollar bill, is the intermediary,
and the reputation of the underwriter for maintaning an unforged, stable
supply of those bills is part of your implicit inspection "calculus", what you
went through in microseconds, before you pocketed my money, now yours.

In fact, the ultimate, (probably mythical, I don't remember enough specific
financial history, here, but you'll get the idea) geodesic meatspace bearer
transaction would be the purchase of a paper bearer financial instrument with
paper cash, especially if the cash was a privately issued bank-note. Say 10 US
Steel bearer bonds, due what? 1934?, bought at issue, (we're in 1904) for par
(such a deal) in exchange for a $10,000 privately-issued bank note.

You have intermediary on the bond side, the bond's underwriter (J.P. Morgan
and Co.), whose reputation is at risk (they don't call 'em underwriters for
nothing) and is being "rented" to the bond's issuer, U.S. Steel. You also have
the intermediary on the other side, the $10,000 banknote's issuer/underwriter,
say, the First City Bank of New York (just to make it interesting), whose
reputation is less at risk. We don't call it "cash", for nothing, after all.
:-).

So, imagine what the reputation/risk/transaction triangles look like:

                           You
                         /  |  \
(bond underwriter) Morgan   |   City Bank (banknote underwriter)
                         \  |  /
                            Me

You have a straight line between buyer and seller down which the bond and bank
note goes in opposite directions, and then a risk triangle on either "side" of
the trade, with J.P. Morgan on the apex of one, and the City Bank at the
other. With a paper bearer transaction, this is done with a Brinks guard, or
just Great Grandmama, :-), bringing the $10,000 note to a bond dealer's
operations department, appropriately named "the cage", forking over the
banknote and walking out with a rather large envelope with 10 rather large
(lots of legal mouse-type, not to mention the coupons) U.S. Steel bearer bond
certificates in it.

In meatspace, between two consenting business partners, :-), I might even go
so far as to include a *check* as a geodesic transaction, because you trust me
enough to take my check, which is what usually happens, and you trust my bank
to be there honoring my check, and I trust your bank to clear the check
correctly. (Yeah, I know, these days, I have to deposit it in *my* bank, and
from there the transaction goes up and down this huge check-clearing hierarchy
with the Fed, or a Fed-equivalent, or both, at the top. But you get the idea.
In the old days, if we both lived in the same town as our banks, there would
be no Fed, just a messenger, making it look much more geodesic, like the
transaction triangle I'm talking about.)

But, a credit card sale, I think, can't ever really be called geodesic
transaction. First of all, to accept the credit card, I have to be a special
class of person called a credit card merchant, and that means, frankly, I'm
closer to a server than a peer. :-). And, of course, there are two separate
hierarchies in a credit card transaction, one in the credit card association,
and the other in the wiring of money from one bank to another.

Right. Now we get to the net. First of all, as I've said, a credit card number
through an SSL session is just an instruction to the merchant's bank to
execute a transaction with my bank. It is, for the sake of our discussion
here, an execution. But it's not really cleared or settled. That's done
completely out of band, off the net.

Now, an electronic check, like the FSTC electronic check
<http://www.fstc.org/> is *more* geodesic than a credit card, to the extent
that if you trust me, you'll take my check, even in email, endorse it, and
mail it to your bank, who'll, again, clear and settle it out of band with a
Fed Funds or other wire. Look ma, peer-to-peer. Back to the two triangles of a
crosstown check, sort of.

FSTC checks, require, however, the same albatross that DigiCash liked to wear
around its neck on occasion, a smartcard. :-). They wanted to replicate the
check process down to a bank controlling the issuing of it's own checks, just
like they do now. The smart card "checkbook" has signing capability, but it
also has a number-counter which decrements as each check is "written", and
that number can be only "recharded" at a bank. Now, the first FSTC electronic
check, named, oddly enough, "echeck", :-) went across the internet in June of
last year. It was written by the US treasury for $32,000, from the Department
of Defense to GTE (probably to pay a BBN bill :-)), who emailed it to their
bank, BankBoston, who then ran it through ACH, and, wonder of wonders, it
cleared. Heavy players indeed, and it reflects, rather clearly, FSTC's
top-down marketing approach to the problem.


Now, there is a simpler, bottom-up way to think about this, and that is to put
a lobotomized SMTP-only Linux box at the bank, with a IETF-PGP keyring of all
its depositors. The server gets email, checks the depositor's key against its
keyring (say on that new spiffy NCipher SCSI keyring box), and checks that the
bank itself has signed the key stored there. If that is the case, out pops,
onto the bank's network, an ACH transaction and DDA record. Easily three
orders of magnitude cheaper than a paper check, on probably all fronts.
Certainly much cheaper the FSTC echeck, even. :-). Geodesic, three orders of
magnitude, nothing but net.

Since the depositor's the person on the hook financially if the check bounces,
as long as the depositor's bank, its deposit server, can recognize the
depositor's signature as one valid for deposit, that's all that's really
required, risk wise, for the check to clear. If the check bounces, it comes
back out of the depositor's account. If, later, the supposed check writer sees
unauthorized check transactions coming out of his account, he can rescind
them, just like he can now with an improper debit-card charge. Clearly, if a
bank has an account which does this too much, they part company. :-).

At one point, Paul Harrison, Rodney Thayer, Vinnie Moscaritolo, Bob Antia,
Fearghas McKay and I were kicking around this idea as an open-source,
IETF-able, internet check protocol, called, if I remember it right, Internet
Deposit Box, or IDBx for short. We sort of ran out of gas. Everyone had a day
job. F=MA and all that. :-). Talk to Rodney about it. I think he, or Paul, has
the none-dare-call-it-preliminary spec somewhere.

The idea was that any cypherpunk with a bank for a customer could have a
working "deposit window on the internet" in a few hours of hacking.


Even more geodesic than a check would be cash or cash equivalents,  Mondex
smartcards and all :-), if you could plug a Mondex card into your PC or PDA.
Funnily enough, Dave Birch says that they were testing Mondex over Newtons
almost as soon as the Newtons came out, way back when. Maybe they even
prototyped Mondex wallets that way, I can't remember.

So, Mondex is geodesic because, once the money-counter in the card is set at a
bank somewhere, it can be sent anywhere, over the phone, or the net, or
wherever, peer-to-peer. The amount you can put on the card is a, um,
"cryptographic risk-management" problem, one with slightly political
overtones, depending on your local money-laundering rules, but, at low enough
amounts, the risk-management problem is obviously non-existant. Again, the
problem is the card, which must be plugged into a Mondex-card reader, on every
machine intending to use Mondex Money. Violates, just barely :-), my "nothing
but net" dictum.

Finally, there are digital bearer transactions. If I give you blind-signature
digital bearer certificate and keep the resulting new certificate, even though
we only do it online, we still have that triangle between you, me, and the
underwriter. When you take the money immediately off the net, the transaction
becomes much less geodesic, because you're destroying the bearer certificate
you have and turning it into an accounting change in the underwriter's
custodian/trustee bank, and that change then percolates through the
wire-transaction hierarchy back to your bank. The same thing happens in
reverse, of course, when, using your ATM card at the underwriter's web page,
you withdraw money from your bank account in the form of digital bearer cash.
So, when you take money out of your bank, put it onto the net, then buy
something from me, and I immediately deposit it, that destroys the entire
cost-basis equation.

Anyway, I claim further, and again heretically, that from the standpoint of
ease of use, the present value of human time, and the cost of bandwidth,
processing power, etc., that the redemption/reissue step, taken by itself, is
about as cheap as my physical inspection of a piece of paper cash prior to
putting it into my cash-register, and should cost me the same thing. Nothing.
:-). Even if, as the underwriter, you have to archive all the spent
certificates, and delete them later on a risk-adjusted basis. Even if you have
to periodically change keys, also on a risk adjusted basis, and create new
epochs, issues, whatever, of cash, to prevent the Borenstein(sp?) attack (I
have the underwriter's private key, so I can print money) on a digital bearer
underwriter.

Finally, with MicroMint-type hash-collision coins, you can do transactions
completely offline (or at least with minimal stochastic testing) without the
redemption/reissue step you'd need for blind signatures. This, coupled with
the scale-economies of their generation, makes MicroMint coins real cheap, but
only for very small value *and* low risk transactions.

So, now, let's look at completely geodesic transactions on a geodesic
internetwork.

The first, of course, would be the exchange of two digital financial assets,
in digital bearer form, of course. We'll use blind signatures, and leave names
out to protect the innocent:

                                   You
                                 /  |  \
(Digital bearer bond underwriter)   |   (Digital Bearer Cash Underwriter)
                                 \  |  /
                                    Me

Of course, the reputation/risk/transaction triangles all need physical
internet connections, not to mention the underwriters at least needing
persistant digital reputations (though not necessarily biometric identity,
oddly enough :-)), in order for this to work. Fortunately, this doesn't
violate the "nothing but net" rule, in fact, it's implied.


Now, let's look at something else, a good or service which is digitally
transferrable, like an MP3 file. (We'll leave legal opinions, or teleoperated
brain-surgery, alone for the time being. :-))

This one's a lot simpler:

                        You
                         |  \
                         |   (Digital Bearer Cash Underwriter)
                         |  /
                         Me


That's a geodesic payment, and, a whole bunch of payments like this, with lots
of underwriters offering graded fungible financial instruments to buy and sell
things with, so that lots of people can sell graded and fungible software (and
wetware) over the net in so-called "perfect" market competition, is a geodesic
economy.

Hope that helps?



Cheers,
Robert Hettinga








-----BEGIN PGP SIGNATURE-----
Version: PGP for Personal Privacy 5.5.5

iQEVAwUBNs5Nj8UCGwxmWcHhAQEINwf+K/h/gGhGqQYFBaowaL9qzR9HLaEt5bDH
fL5BB+kC6gOYBM3lO2fVRwJ6LsW3q4uvmYLr7pCV2L+qo7RBpHfP9ejn+lBTTpoo
iLvQgfon3CobfLBHTzUZjY534GoP1JX6F63oauG1F8gmqXL5vSgQmbKxilFq2CQl
Adn6f6X4ioCRbuLFXyPnQPgUNEGhJnMNYPF2jHlur0SbNX74RI5Pv5KIeyE/5ht0
eJzEN+60bpoE0HkOJ+1NDalXmiuppkYAyzfusjVih/v1JlG3csWW5b8JSnljTjdf
Z7RpFWqWzv6pa5KOX/cEufNT4xoEbYHEFZqE+g6qhWR4UWVyQGSfgA==
=EeTF
-----END PGP SIGNATURE-----
-----------------
Robert A. Hettinga <mailto: rah@philodox.com>
Philodox Financial Technology Evangelism <http://www.philodox.com/>
44 Farquhar Street, Boston, MA 02131 USA
"... however it may deserve respect for its usefulness and antiquity,
[predicting the end of the world] has not been found agreeable to
experience." -- Edward Gibbon, 'Decline and Fall of the Roman Empire'

For help on using this list (especially unsubscribing), send a message to
"dcsb-request@ai.mit.edu" with one line of text: "help".