Re: IMP (was Re: ecash-info (fwd))

hughes@ah.com (Eric Hughes) replies,

>   1.) Chaum's e-cash coupled with WWW/Mosaic is a de facto internet
>   mercantile protocol.
>
>Hardly.  The announcement just says it's available, not that anybody's
>using it.  Since the information came from a press release, we can
>assume that lack of mention of an important customer, like a bank,
>means that there are no such customers right now.  What that says to
>me is that DigiCash has looked for customers, and not found any.
>They've certainly had the time.

Chaum's chumming the hook.  I get it. But the claim has still been made
that the techonolgy works, right?

>
>Furthermore, it's not clear that this software can be both legally and
>usefully deployed in the USA.  The Foreign Bank Secrecy Act of 1974
>requires the microfilming of all checks of value over $100, with
>administrative provisions for extending the required recording
>keeping.  Other check-like transaction accounts have since been added.
>
>So can a bank avoid this?  First, they can limit transaction amounts
>to less than $100.  That violates my criterion of usefulness; it would
>have some utility, to be sure, but just as surely would be a severely
>crippled utility.  Second, they might be able to record the
>transaction as a "cash purchase".

I believe that that is the case, and thus whether or not it's considered a
check is immaterial.

>The problem here is that this
>accounting technique may be ruled non-compliant by the regulators,
>which would make the transaction _illegal_ (since there's not way to
>comply by recording both parties).  The regulators have been
>authorized to move activities across the boundary of legality by
>legislative action.  Now, one cost of deploying any such system would
>be the expected (negative) value of the risk taken in losing the whole
>development investment to an adverse regulatory decision, let alone
>possible actual penalties.

True. That is a risk of deploying the protocol from the financial entity's
standpoint.  Like most things in the banking system, a consensus (inside
the beltway and out) would have to be reached. But this is a political, not
a technical, challenge.

>
>Even beyond this, there's the IRS $10K cash reporting limit, and the
>attendant restrictions on structuring.  Detection of structuring
>becomes much more difficult, and banks are held responsible for at
>least some of the enforcement. Here's another set of risks, like
>above.

I'm hard pressed to see the difference between $10K of paper money and $10K
of e-cash.  That's the point of the technology. If you treat it the same
way, you can regulate it the same way.

>
>Just how big is the potential Internet market (in, say, two years),
>compared to other banking segments?  Precious small right now, really.
>Just plain profitability is also an issue.

If it is possible to sell, maintain and support software on the internet,
there will be an incentive for sellers to use it to reduce costs.  In my
own experience it is easier and much cheaper to identify customers who want
my somewhat specialized software on the internet than it is in a lot of
other places. It is possible to distribute my software to those people much
cheaper. It is cheaper to maintain that software through net.mail and ftp.
If I had an e-cash-register coupled with a transaction-ftp capability, I
could sell my software without knowing who bought it, and put the money in
the bank more efficiently than if I had to deal with checks, credit cards,
etc., I would jump at the chance. (If wishes were horses, beggars would
ride :-)... )

This forms demand for vendors to take e-cash. The demand from e-cash users
comes from the same advantage cash confers.  It's easier to use. Since
anyone (including a machine) can tell that it's e-cash, that it can't be
anything else, it's acceptable for all transactions.  A vendor doesn't have
to check the purchaser's bonafides. It doesn't matter. Privacy is a
by-product of this.

>   2.) It seems to me that that e-cash, contrary to the status quo's thinking,
>   is *critical* to internet commerce.
>
>No, it's not critical.  Some form of transaction mechanism is
>critical.  Privacy is not critical to the bulk of the economy, though.

I think you're right.  Personal privacy is a by-product of an iron-clad
interpersonal funds transfer system, like e-cash. The cladding comes from
encryption technology. A form of privacy, but I would be equivocating in
the way I used the word above if I used it that way. E-cash is critical
because of it's efficiency.  With it, I can sell software or
computer-related services from any net-connected machine to customers
Singapore, or Japan, or down the street without having to worry about who
they are, whether or not their checks clear, or without having to pay their
credit card company's customer accounting costs, or without having to
convert their e-cash from one form to another unless I wanted to.

>Face up to it.  If it were, it would be so obvious that we wouldn't be
>discussing it on a mailing list.  In fact, _we_ wouldn't be discussing
>it, but rather a whole bunch of bank vice presidents.

This is the "if we lived here, we'd be home now" argument. Goddard, Korolev
and Von Braun were making liquid-fueled rockets long before governments saw
the efficacy of missles as weapons of war, and "all" Von Braun wanted to do
was to go to the moon.  Admittedly, a war was required before people built
large rockets.  I expect that some powerful economic forces will have to
emerge in order to create demand for e-cash. I also that think that the
Internet is creating those forces. I also don't believe that strong crypto
is as engineering intensive as rocket science is (all my protestations
about its comprehensibility to the contrary....).

>Is anonymous cash really the most efficient?  No, not in all cases.
>When no one is looking, the anonymity is irrelevant, and
>identifier-based schemes work fine.  Is, for example, anonymity the
>most efficient for the Federal funds transfer network?  No, because
>the values of money are so large that default on a transaction would
>case serious systemic problems.
>
>Cash does have some advantages, in particular its immediate and final
>clearing.  These can reduce transaction costs in certain markets.
>Anonymity, however, is not a panacea.
>
Conceded. Anonymity is the result of the most efficient kind of personal
transaction. A cash transaction. I think the efficiency can be examined by
affirming the negative here.  Suppose that all cash transactions had to be
recorded and each party of the transaction had to be identified and
reported to some other third party (the government, say). Besides the
specter of big brother watching you, the economy would choke in
administrivia (I *like* that word, Eric).

>   3.) Since a big pile of the discussion on this group lately has been about
>   our collective concerns about an RSA-approved version of PGP, I think there
>   is a real parallel here in e-cash.
>
>PGP only requires the cooperation of your email correspondent in order
>to function.  The risk of a patent infringement suit is small, since
>the parties involved are small.  Digital cash requires the
>participation of many more parties, some of whom have, almost of
>necessity, deep pockets.
>
>The parallel does not extend very far.  Without the creation of an
>entirely black market which can remain completely unexposed (and this
>is more difficult that it appear even on second or third thought), it
>is unlikely that digital cash technology will be usefully deployed
>independently.

I don't think I was clear enough.  The parallels had to do with the
distribution of the technology. With PGP being everywhere, the value of PGP
grew.  If the people who control patents to the "wallets" and
"cash-register" technology would let that be available for all, then the
value of their patents on the means of producing the medium of exchange
(providing it can be readily converted into *real* money somewhere) goes up
enourmously.

I liked doing this. I hope all my yammering above holds water, and isn't a
waste of bandwidth.  These are precisely the kinds of arguments people in
the financial markets will use when they throw darts at participation in an
e-cash node. If the questions can be answered here, then whoever makes the
pitch will be that much ahead of the game.

Thanks,
Bob Hettinga


-----------------
Robert Hettinga  (rah@shipwright.com) "There is no difference between someone
Shipwright Development Corporation     who eats too little and sees Heaven and
44 Farquhar Street                       someone who drinks too much and sees
Boston, MA 02331 USA                       snakes." -- Bertrand Russell
(617) 323-7923

>Now, one cost of deploying any such system would
   >be the expected (negative) value of the risk taken in losing the whole
   >development investment to an adverse regulatory decision, let alone
   >possible actual penalties.

   True. That is a risk of deploying the protocol from the financial entity's
   standpoint.  

It's a risk, that risk has costs both direct and indirect, and
therefore Chaum's systems are _more_expensive_ than they appear.  These
risk costs _will_ affect what gets deployed.

   Like most things in the banking system, a consensus (inside
   the beltway and out) would have to be reached. But this is a political, not
   a technical, challenge.

Almost all the problems in deploying a digital cash system at this
point are financial and political.

   re: IRS reporting

   I'm hard pressed to see the difference between $10K of paper money and $10K
   of e-cash.  That's the point of the technology. If you treat it the same
   way, you can regulate it the same way.

Smurfing is easier in the electronic domain by a long shot.  Smurfing,
for those, not in with the jargon, is sending out flunkies with a few
thousand in cash each to fetch cashier's checks (i.e. non-cash
instruments).  Since the transfer of e-cash and the creation of
nominal accounts is much easier, it's that much better for moving
anonymous money.

The Treasury Department will see this as a Bad Thing.  It will most
definitely be a regulatory hurdle.

   re: getting profitability

   If it is possible to sell, maintain and support software on the internet,
   there will be an incentive for sellers to use it to reduce costs.  [etc.]

I elided an important point.  It seems clear to us that there's a
large market available on the Internet.  Will it be clear to the
financiers?  Not without a lot of education.

   If I had an e-cash-register coupled with a transaction-ftp capability, I
   could sell my software without knowing who bought it, and put the money in
   the bank more efficiently than if I had to deal with checks, credit cards,
   etc., I would jump at the chance.

This is a feature of any all-electronic payments system, not only of
electronic cash systems.  There are alternatives which can work
economically.  Deployment of anonymous digital money is not an assured
event.

   E-cash is critical
   because of it's efficiency.  

Almost all the efficiency comes from the fact that it's electronic,
not that it's cash.  It is true that cash systems more quickly
consolidate receivables, but the advantage over paper is _relatively_
small.

   With it, I can sell software or
   computer-related services from any net-connected machine to customers
   Singapore, or Japan, or down the street [...]

As soon as foreign exchange transactions come into play, life gets
more complicated real quickly.  I think there really is a large market
available in low level foreign exchange, but it's much more likely
that single currency money systems will be the first to be deployed.

   Suppose that all cash transactions had to be
   recorded and each party of the transaction had to be identified and
   reported to some other third party (the government, say). Besides the
   specter of big brother watching you, the economy would choke in
   administrivia (I *like* that word, Eric).

Choke?  I think not.  Costs would go up a little, certainly, but all
the reporting could be put into software.  Ever heard of the term
"compliance officer" in banking?  It's someone who goes around and
makes sure the firm doesn't inadvertently break any laws.  Well,
compliance for cash reporting would be in software from day one of the
requirement.  It might add a bit to computer system costs, but not
appreciably to labor costs.  After all, filing would be done
electronically, for real-time monitoring.

   If the people who control patents to the "wallets" and
   "cash-register" technology would let that be available for all, 

The 'purchaser' package of DigiCash will be freely distributed.  I
don't think the 'merchant' package will be.  I infer this from looking
at the questionnaire for self-qualification of DigiCash's that got
posted here.  There was a one category for banks, certainly to be
licensees, and one for merchants, therefore also to be licensees.

In summary.  Anonymous cash systems are not clearly better than
identity money systems.  It's not clear at all that one will win out
over the other.  In the USA, there are strong governmental forces
against anonymity.  The best we can hope for is that both get
deployed.  The market will then be able to choose.

Eric

I agree with virtually everything Eric says with one small exception...

Eric Hughes says:
>    E-cash is critical
>    because of it's efficiency.  
> 
> Almost all the efficiency comes from the fact that it's electronic,
> not that it's cash.  It is true that cash systems more quickly
> consolidate receivables, but the advantage over paper is _relatively_
> small.

The advantage is that its electronic AND that its secure. Since its
secure, the intermediation costs drop dramatically as the possibility
of fraud goes down. One could do electronic payments with credit cards
and email right now -- but the costs would be pretty bad.

There is another advantage you've glossed over, which is the fact that
since fraud is difficult, anyone, not just vendors, could receive
payment. (Vendor fraud is a huge cost in credit cards.)

I agree, however, that any truly secure electronic payments system has
these advantages -- anonymity isn't needed to gain most of the cost
benefits. I'll also note, by the way, that the stupid smartcard
systems that simply rely on "tamperproof" (ha!) cards that "know"
balances aren't going to have especially lower costs than credit cards
-- increased fraud might even raise costs! A truly secure system has
an enormous advantage over such systems.

Perry

The advantage is that its electronic AND that its secure. Since its
   secure, the intermediation costs drop dramatically as the possibility
   of fraud goes down. 

But it is also possible to make systems that are secure and
non-anonymous.  Admittedly, I spoke of "identity-based systems", which
is not quite right.  Rather I should have said "identifying systems",
which include the identity but do not rely upon it alone to verify
payment, as do credit cards, say.  These kinds of systems can be just
as secure and completely lack anonymity.  

To pick just one, consider certified digital checks.  The drawer
writes a check, the bank certifies it (and puts a hold on the
account), the check is transmitted and deposited.  Secure, low level,
and totally identifying.

   One could do electronic payments with credit cards
   and email right now -- but the costs would be pretty bad.

I agree.  There's an interesting parallel.  As it turns out, credit
card fraud is _dropping_, because of various educational programs and
anti-fraud measures.  The one segment that credit card fraud is
increasing is in technical card forgery, which is way up.
Transmitting card numbers electronically over the Internet can only
exacerbate that problem.

Eric

on Thu, 02 Jun 1994 09:33:23 -0700  hughes@ah.com wrote:
> Transmitting card numbers electronically over the Internet can only
> exacerbate that problem.
> 

Yes, if transmitted in the clear, PGP is legal now :-).  Vendors on the
net need to be pushed to use encryption.

        Dave Otto -- dave@marvin.jta.edd.ca.gov -- daveotto@acm.org
    "Pay no attention to the man behind the curtain!"     [the Great Oz]
    {I *DO* have a life, it's just that my kids are using it right now!}

> Transmitting card numbers electronically over the Internet can only
   > exacerbate that problem.

   Yes, if transmitted in the clear, PGP is legal now :-).  Vendors on the
   net need to be pushed to use encryption.

I'm not referring to the problem of sniffing credit card numbers off
the net.  I'm referring to the problem of credit card fraud by the
operation on the receiving end.  Even if the transmission is
encrypted, there's still risk.

Eric

Eric Hughes says:
>    > Transmitting card numbers electronically over the Internet can only
>    > exacerbate that problem.
> 
>    Yes, if transmitted in the clear, PGP is legal now :-).  Vendors on the
>    net need to be pushed to use encryption.
> 
> I'm not referring to the problem of sniffing credit card numbers off
> the net.  I'm referring to the problem of credit card fraud by the
> operation on the receiving end.  Even if the transmission is
> encrypted, there's still risk.

Eric is, of course, pointing out the fact that credit cards qua credit
cards are inefficient. (By the way, the transmitting end is also a
source of risk -- fraudulent possession of the card number is
possible.)

In general, you can't make credit cards secure by encrypting the
transmission of the numbers because the credit card mechanism has
inherent flaws irrespective of interception.

The only information needed to use the card is the card number.  Given
the card number, there is no restriction on how much of an account I
may draw. Stealing the (short) number, which must be communicated to
use the card, is the equivalent of stealing the account. The merchant
has no restrictions on how much he can draw other than the fact that
he'll be caught if he draws more than he says he will.

Fraud is naturally rampant, since it is childsplay to commit fraud. It
is a major cost of the system.

In even a primitive public key based system, there is no need to take
anyones word for anything, and no need to reveal the "key" to the
account in order to use it.

Perry