[NYT] How Much Am I Bid for This Imperfect Marketplace?

Date view Thread view Subject view Author view

From: Adam Rifkin (adam@KnowNow.com)
Date: Sat Dec 16 2000 - 20:51:21 PST

[Rohit, *everyone* should just move rapidly to the Nash equilibrium and
be done with it already... I wonder if http://www.nashequilibrium.com/
is taken...]

How Much Am I Bid for This Imperfect Marketplace?

December 13, 2000

E-BUSINESS auctions are going, going . . . gone? Not exactly, but they
are far less successful than expected.

Half to two-thirds of all electronic auctions fail to result in a deal,
researchers say. And instead of the predicted 10,000 or so
business-to-business exchanges, with Web sites conducting countless
"reverse" auctions on behalf of buyers in which the lowest price wins,
at most several dozen such operations will survive.

What went wrong? It wasn't just that the pool of venture capital dried
up. More important, analysts say, is that what was once seen as the
electronic pathway to the perfect market -- a cyberspace buzzing with
flexible prices and competitive auctions -- has turned out to have only
limited applicability to most real-world business transactions.

"Initially, Internet technology was used to try to build marketplaces
that conformed to the simple model of a public auction," said Hal
R. Varian, dean of the School of Information Management and Systems at
the University of California at Berkeley. "But real markets are much

Conducting an auction is an efficient way to determine the price that
best balances supply and demand, but it requires assembling all
potential bidders in one setting. "The challenge is to make the Net
serve the variety of ways people conduct transactions," Professor Varian
said, "rather than force people to conform to one model that happens to
be easy to do on the Net."

Moreover, price-driven auctions are most useful for either one-of-a-kind
items or standardized commodities traded in large volume. Most business
deals, however, involve differentiated goods and services, turning on
factors like quality, convenience and reliability as well as price.

"Price isn't everything," said Paul Milgrom, an economics professor at
Stanford University and co-founder of Perfect.com, a software company
devoted to more fully automating the business purchasing process. "In
fact, it's rarely the most important thing."

Perfect.com is one of several software start-ups trying to move the
Internet beyond the basic supply-demand curves of Economics 101, with
its idealized market of uniform products and well-informed participants.
Many of them are being started by economists leaping from the ivory
tower and into the fray of putting theories regarding so-called
imperfect markets into practice.

"All this economic theory that once was dismissed as too esoteric,"
Professor Varian said, "is suddenly becoming the basis for a variety of
promising new business models."

Many business transactions, for example, come down to individual
negotiations. At the initial stages, few buyers want to reveal how much
they are willing to pay for a specific product. Most suppliers do not
want to say how low they will go to close the sale.

In a public auction, maintaining that kind of secrecy is impossible. In
reaching a deal, not only do the participants learn each other's
positions, so do their competitors. Even when negotiations are conducted
in private, two-sided deal making takes time as each side tries to eke
out a slight advantage.

A new company called SplitTheDifference (www.splitthedifference.com),
based in New Haven, hopes to cut through these difficulties with an
online negotiating tool that it plans to operate on its own and license
to other companies. "Think about the typical business deal," said Barry
Nalebuff, a management professor at Yale and a game theorist who is
chairman of SplitTheDifference. "There's all this gamesmanship: 'You go
first,' 'No, you go first.' Do I tip my hand or do I bluff? Well, we've
acted strategically in designing the game so the players don't have to
act strategically when they play it."

SplitTheDifference operates like a combination neutral mediator and
escrow service while creating an environment that safeguards buyers' and
sellers' confidential information: reserve prices, identities and
willingness to sell.

Consider this simplified example. Suppose you, a fish wholesaler, are
willing to pay as much as $5,000 to acquire a particular load of fresh
salmon. The fisherman is prepared to sell his catch for as little as
$4,000. Both sides disclose their secret positions to the computer agent.

The software program recognizes whether a deal is possible. If not, it
tells the participants to give up, try cutting a deal with somebody else
or come back with different offers. If it is and both sides agree ahead
of time, one solution is to split the difference automatically and
conduct the transaction at $4,500. Or the gap can be closed by a random
amount, preventing either buyer or seller from knowing for sure how far
the other side was prepared to go. That helps to preserve freedom of
action in future deals.

"This approach promises to get people to resolution very quickly,"
Mr. Nalebuff said. "It also allows people to do transactions on the Web
without posting prices and to put different prices to different
players. You know, if you make markets too perfect, there's a good
chance most people won't want to participate in them."

The theory behind the company's model, which allows for much more
multifaceted deals than this basic example suggests, comes from research
by two economists, Geoffrey Miller of New York University and Robert H.
Gertner of the University of Chicago.

SplitTheDifference says it is in beta-testing, trying its software with
two clients: an established e-commerce player in the aviation parts
industry, and a business-to-business start-up that will rely exclusively
on the company's automated negotiation technology to execute

In contrast to SplitTheDifference, which concentrates on private
negotiations between businesses, other software makers are focusing on
the fact that most transactions, particularly consumer purchases,
continue to involve seller-set prices attached to specific items. But
there is room for innovation within this time-tested formula. Most of it
is geared to helping companies better use customer feedback to establish
prices and change them in response to varying conditions.

The Optivo Corporation, for example, has received money from Norwest
Venture Partners to start a software application service that would
allow online merchants to test an array of prices at different times.
The goal: calculating how best to maximize return.

Amazon.com experimented with posting different prices to individual
customers, depending on their past buying patterns. But it dropped the
trial after critics attacked it as discriminatory.

Andy Atherton, Optivo's co-founder and vice president for business
development, said the company preferred to let its clients "test prices
sequentially, so they can understand demand without annoying customers."

"Today, companies are used to applying rules to prices and depend on
manual systems to keep track of price changes," said Robert Drescher,
Optivo's chief executive. "But that's becoming humanly impossible. We
offer a method that allows companies to automatically put up the right
prices at the right time across the entire product line."

The application program, Mr. Drescher said, can be tweaked to maximize
either profits or revenue, or to generate the highest level of sales
consistent with a minimum gross profit margin set by the company.

Perhaps the most ambitious of the companies trying to bring greater
economic efficiency to the Internet marketplace is Perfect.com, whose
software is already being used by several online business-to-business
exchanges. These include Bandwidth.com, which serves as a broker between
companies shopping for long-term or temporary telecommunications
services and carriers offering high-speed data capacity. It is a market
where a number of variables, not just price, are important.

"With at least 50 carriers hawking their wares," said David Morken,
president of Bandwidth, "what you had was cacophony, with everybody
yelling at customers all at the same time. Now we've got it down to a

Perfect.com's technology was inspired by the government's
"request-for-purchase" system, which allows agencies to specify on paper
dozens of requirements for the things they buy. But the process is
normally so expensive, time-consuming and complicated that it is useful
in limited circumstances.

To overcome these limitations, Perfect.com has created a kind of
automated request for purchase, in which buyers can spell out quickly
what they want -- from speed of delivery and length of warranty to
quality and reputation. As many as 100 variables are possible. Potential
sellers also fill out Web forms that describe their wares and indicate
ways in which they are willing to adjust their bids to satisfy
requests. The technology ranks the best offers, based on the priorities
established by the buyer.

"Almost all suppliers try to differentiate themselves on things other
than price," said Kevin Surace, Perfect.com's chief executive. "We
modeled this on the way business is really done."

Will these academically inspired markets work?

"The true test," Professor Varian of Berkeley said, "is the marketplace."


# of people on the internet by 1990 / # by 2000 is a small fraction. Installed base will kill you in linear markets and liberate you in exponential ones :-) -- Rohit Khare

Date view Thread view Subject view Author view

This archive was generated by hypermail 2b29 : Sat Dec 16 2000 - 20:56:49 PST