Why digital cash has not taken off (yet)

Ciamac Cyrus Moallemi (ciamac@alum.mit.edu)
Thu, 09 Sep 1999 10:44:50 -0400

Short answer: companies are shifting their pricing models to sophisticated
price-discrimination for expensive goods, and flat-rate pricing for cheap
goods, neither of which is well-suited to digital cash.

Date: Wed, 08 Sep 1999 01:48:17 -0400
From: Howie Goodell <goodell@mediaone.net>
To: cypherpunks <cypherpunks@cyberpass.net>
Subject: Tuesday 9/8 Digital Commerce Society of Boston meeting
Sender: owner-cypherpunks@cyberpass.net
Reply-To: Howie Goodell <goodell@mediaone.net>

Hi --

I took reasonable notes of Dr. Odlyzko's talk:

"Why digital cash has not taken off (yet)"

Tuesday 9/8/1999 at the Digital Commerce Society of Boston meeting.

Howie Goodell
- --
Howie Goodell Senior Software Engineer HCI Research Group
28 Lucille Avenue FEI Company - Micrion Computer Science
Salem, NH 03079-2054 1 Corp Wy Centennial Park Univ. Massachusetts
(603) 898-8407 Peabody, MA 01960-7990 1 University Avenue
(810) 222-2042 fax (978) 538-6680 -6699 fax Lowell, MA 01854
goodell@mediaone.com hgoodell@eng.micrion.com hgoodell@cs.uml.edu
End-User Programming: http://www.cs.uml.edu/~hgoodell/EndUser
"You have zero privacy anyway. Get over it." (Sun CEO) Scott McNealy

=================== BEGIN INCLUDED MESSAGE ============================

Digital Commerce Society of Boston meeting 9/7/1999

Dr. Andrew Odlyzko
Head of the Mathematics and Cryptography Research
AT&T Laboratories
(see his home page at <http://www.research.att.com/~amo>)

Why digital cash has not taken off (yet)

(Please note that these are based on my informal notes and
recollection: I apologize in advance for inevitable errors; caveat
lector. All
trademarks are property of their respective owners.)

Dr. Odlyzko gave a fascinating talk, even worth the annoyances of Boston
traffic and parking. Instead of talking about digital cash, he spent
his time on the economics of the new age it must find a place in.

Part of the problem is that payment is a deeply traditional matter.
Only Americans use a lot of checks. Japanese carry wads of cash.
Europeans use something else. Romanians used Kent cigarettes.
Radically new habits will take time to form.

A much more significant hurdle is the economics and politics of the
coming age. A fashionable picture of Information Age economics
is a wide-open superhighway, where costs will be driven down close to
the physical limits of the production process, and the consumer will be
king. Nonsense! The Information Superhighway will be full of bumps and
barriers. Yes, many costs will be driven down, but in many other areas
producers will have more control than ever. How is this possible?


1. Political Barriers are still around

The most obvious political barrier is tariffs. These are not new: he
quoted Frederick Bastiat's 1848 masterpiece, "The Petition of the
Candle makers Against the Competition of the Sun." There are more
subtle barriers. For example Portland Maine built the original track to
Montreal with a nonstandard 5-foot gauge; so shipping companies would
have to pay Portland labor to unload all the freight from the standard
cars and load it onto the other kind. [I thought of the US standard for
HDTV, which was very consciously crafted to keep out Japanese sets;
Europeans would not have to think.]

Other barriers can be erected just by a company's market power. For
example some local phone companies in the US took the letters off pay
telephone keypads; so customers would have trouble dialing long distance
codes like, "1-800-CALL-ATT."

2. Price Discrimination has economic value

Example: Alice is willing to pay $1000 for a study of digital cash, but
Bob will only pay $700. If it costs $1500 to produce, there is
no uniform price that recovers the cost:


Yet if the producer is able to discriminate, both customers get what
they want at a price they are willing to pay. Uniform pricing,
whether enforced by law or custom or anonymous digital money, would
result in fewer total goods being produced.

3. Marginal costs are plunging

Most of the cost of a new drug or microprocessor or software program is
upfront development and marketing; producing more units
is a small fraction of the market cost. For instance a Pentium
microprocessor sells for $100-$500, but only costs $30 to produce.
Communications satellites are almost all upfront cost [witness Iridium's
predicament.] Even automobiles cost $2-3 billion for design
and tooling. This fact creates increasing incentives and opportunities
for price discrimination in the Information Age.

4. Can price discrimination persist in competitive markets?

Books can't usually be sold at widely different costs. So publishers
come out with a hardcover edition first, for about $25; later a
trade paperback for $15; later a regular paperback for $6. These cost
similar amounts to produce, but the "artificial versioning" lets
them discriminate between customers. There is also a segregation of
markets: e.g. libraries get you the book for free, but you pay in
time and inconvenience.

5. "Damaged Goods"

An extension of this is what economists call "damaged goods", which are
made inferior deliberately so as not to compete with higher
rates for other customers. For example in 1990 IBM produced a regular
laser printer that printed 10 pages/minute, and also a "Laser
Printer E" that cost much less, but only printed 5 pages/minute. It
turns out they were identical except one chip that was inserted into
"Laser Printer E" to slow it down! So it actually cost more to produce,
but sold for much less. Similarly Fed Ex "afternoon delivery"
packages often arrive with the morning-delivery ones. However they are
held until afternoon, even if it means scheduling an extra run.

6. First, Second, and Third Class

Classic 1849 quote about differentiation in train transportation by
another French economist, Emile Dupuit (e.g. at
http://ksgwww.harvard.edu/iip/econ/varian.html , which has a full
discussion of several of these issues.)

"It is not because of the few thousand francs which would have to be
spent to put a roof over the third-class carriage or to upholster
the third-class seats that some company or other has open carriages with
wooden benches ... What the company is trying to do is prevent the
passengers who can pay the second-class fare from traveling third class;
it hits the poor, not because it wants to hurt them, but to frighten the
rich ... And it is again for the same reason that the companies, having
proved almost cruel to the third-class passengers and mean to the
second-class ones, become lavish in dealing with first-class customers.
Having refused the poor what is necessary, they give the rich what is

In today's airline market, average fares continue to fall, while the
differential between business class and economy grows. About 1/3
of the audience today agreed this was a good thing. This is apparently
the highest percentage he's seen yet; among those that were probably
more representative of the general public, 1% or less did.

Note that anonymous digital cash tends to create uniform pricing to
faceless customers. This is *the opposite of the relationship*
desired by vendors.

7. Bundling

Microsoft Office a few years ago (Windows 3.1 version) cost $350,
although most of the components cost $250 apiece (total $850).
This makes sense, as the following example demonstrates:


Selling Price______Sales________Return

Bundling enables the content producer to *average uneven demand* for
components. This enables products to be produced that otherwise could
not be. Note that *this means micropayments won't be used very much*,
because producers want to bundle, e.g. the whole Wall Street Journal
into an electronic subscription, instead of selling individual articles.

Speaking of the WSJ, an article on (Page 1?) of the January 7, 1999
issue described the ultimate in customer segregation. Some
banks' computers use AI programs that color code customer records to
indicate to customer service reps how good a customer they
are talking to. "Green" customers make money; they get lots of
consideration and tolerance; "Yellow" are borderline and get less;
"Red" customers lose money, and the reps are instructed to treat them
harshly in hopes they'll go to another bank.


The two main countervailing tendencies to the above are:

* People's sense of fairness (often embodied in laws)
* The tendency for simplification

1. The Ultimatum Game

Researchers bring strangers together for a one-time game. First Alice
gets to propose a distribution of a real $10.00 amount, and
Barbara decides whether to accept or reject. If she accepts, they both
get the proposed amounts; if she rejects, neither gets anything.
Rationally Barbara should accept anything, even $0.01. Yet most
distributions are 50:50. Apparently driven by people's perception of
what's fair, and by simplicity. (Note if it were a real $10,000,000
amount, results might be different.)

2. Telephone Pricing

In 1984, AT&T's business long-distance rates varied considerably by time
of day and distance. While distance has become less important to actual
costs, time has not: the network must be sized for peak daytime demand;
so discounts to shift times are justified. Yet today prices are
independent of both distance and time.

Cellular service shows the same effect. Sprint now has unlimited long
distance all weekend long. AT&T offers "Digital One Rate", currently
$90 for 600 minutes regardless of long distance or roaming. Roaming has
a very real cost difference: AT&T pays substantially more to other
companies that provide roaming service; yet this simple plan has
enormous appeal.

Another demonstration is local telephone rates. The US is almost unique
in offering unlimited local calling: in most countries it's all
per minute. In the 1970's the FCC pushed the Bell system to offer other

- a very cheap "Lifeline" rate that charged a lot per minute after the
first few calls;
- an intermediate rate with a fixed number of free calls or minutes and
a lower per minute rate afterwards, and
- the traditional flat rate.

Interestingly 60% of callers who made almost no calls a month still paid
extra for the unlimited flat rate. Why?

1. Insurance: if my son comes back from college and calls his
girlfriend all day ....
2. Overestimating call volume: estimated calls were log normal to
actual calls
3. "Hassle factor"

For evidence of (3) he pointed out that non-flat-rate callers had
shorter calls, even where they were billed by the number of calls rather
than minutes. People have a strong preference for avoiding hassle, and
are willing to pay a lot for it. [I suggested this was the market
value of one of the human brain's 7+/-2 short-term memory locations.]

3. Internet Local Access Charges

He presented quotes on the evolution of AOL's fixed-rate plan, from
_Aol.Com : How Steve Case Beat Bill Gates, Nailed the Netheads,
and Made Millions in the War for the Web_, by Kara Swisher. [$11.20 on
Amazon right now -- at least for me.] The biggest gripe
people had (from a rich cornucopia of possibilities) was the clock. One
customer continued to insist to an incredulous Case that AOL
was cheating her by charging based on time; even after he demonstrated
she was paying less that way.

AT&T's WorldNet showed a similar effect. Even though their $19.95 plan
offered 150 hours/month (which less than 5% of
customers exceeded), there was demand for a $21.95 plan with unlimited
time (the small price differential proves most of the people
opting for it were not in the 5% of extreme users.)


1. Expensive goods like airline transportation and automobiles will
have sophisticated pricing schemes
2. Low-priced goods will tend toward simple, flat-rate pricing.

The crucial point: **neither of these major market categories fits
digital cash very well.**
Example: airlines love the photo ID security check, because it prevents
a secondary market in tickets and lets them price-discriminate.


Q&A (I only caught the first 5-10 minutes of this)

Q: (Bob Hettinga:) Pre-industrial prices were always negotiated.
Today's financial markets -- e.g. commodities -- always use
negotiated prices. Automating the process will allow this to be used
generally [auction sites already sell everything from heirlooms to

A: This will be true primarily for larger business entities, where the
large sums and repeated transactions justify hiring specialists.
- ----
Q: What's your basis for saying digital cash hasn't succeeded?

A: Just the failure of many such schemes, such as Digicash.
- -------

Q: What field *do* you see for digital cash?
A: If you want to buy a Coke, you don't want a soft drink subscription;
you want cash.
- --------

Q: Was the phone company example influenced by falling local phone
company tariffs?
A: Yes; the access charges now dominate. Long distance averages
$0.12/minute, of which only $0.01/minute is the cost of the long
distance network itself. [I don't know if he made any other point
- ---------

[I add one I didn't get to ask:]

Q: Dr. Odlyzko's examples mostly involve some form of monopoly
competition: copyrighted or patented goods; natural monopolies
like local phone service, or high barriers to entering e.g. auto or
microprocessor production. Is it a part of his argument that the prices
of commoditized goods *will* fall close to their production costs,
leaving these exceptions as a proportionally larger part of what we
spend money on? If so, then a major goal of social design should be to
minimize unproductive "rents" (as well as other negative
externalities like loss of privacy, the "tragedy of the digital
commons") while still allowing economically useful price
For instance, while remaining fully anonymous, I might present a
trustworthy digital credential that my income was below a certain
level to purchase a product at a reduced price.

- --- end forwarded text