FOR IMMEDIATE RELEASE

Summary:
One of the first long-term research projects chartered by the IBM Institute for Advanced Commerce is exploring the economic and technological impact of a massively networked, highly decentralized electronic world comprising billions of potential buyers and sellers.

"We're asking, 'What's going to happen when there really are a billion people connected to the network via tens of billions of devices, each commanding hundreds of intelligent software agents?'" explains Dr. Steve R. White, an IBM Research senior manager heading the project. "We want to anticipate the rules and behaviors that will evolve in such a complex and dynamic economic environment."

White and a team of IBM researchers working on this project, called "Information Economies," are already running complex computer simulations that are revealing interesting trends and potential models of behavior, including the emergence of specialized markets and roller-coaster pricing.

Project Background: Agents in Cyberspace
The Internet is already showing great potential as a medium for commercial transactions. Individuals and businesses can gather information, discuss, negotiate, and form associations more rapidly and easily than they can within their own physical communities. As the scale of interaction increases, software agents will be used more extensively to manage the flood of information and opportunities.

Agents are used today to simply seek, sort, and select information on the Internet, but in the future they will engage in complex transactions on behalf of their human owners -- or on behalf of the devices on which they reside.

IBM researchers anticipate a scenario in which billions of intelligent agents will roam the virtual world, handling all levels of simple-to-complex negotiations and transactions. Profit-driven agents that can make money by offering services such as translation or data mining are expected to be the most prevalent form of this technology. The researchers have dubbed this scenario an "information economy."

In an information economy, certain characteristics of software agents could radically alter the nature of commercial interaction.

"Agents are much faster at making decisions than humans, but also much less flexible. For example, they can't cope with unanticipated situations," explains White. "They excel at sifting through dramatically greater quantities of information and options than their human clients, which means that market and system efficiencies -- and flaws -- will operate on an accelerated level."

Initial Observations: Gas Station Wars and Roller Coaster Rides
The research team at the Institute is using advanced simulations and modeling to investigate how information economies will resemble and differ from human economic systems. Two immediate observations are the emergence of specialized markets and roller coaster pricing.

For example, price wars result when agents are programmed to maximize profits by buying or selling at the best price. If provided access to bidding information about incoming bids, each selling agent sets its price just below the lowest offer.

Since agents negotiate and make decisions rapidly, without the psychological and other factors that affect human decision-making, the bidding process is accelerated, and prices drop precipitously in a matter of seconds. When profits from selling a smaller volume at a higher price exceed those from selling larger volumes at the prevailing low price, an agent raises its price and the other follow suit. In simulations, this cycle repeats itself endlessly.

"We're observing the cyber-equivalent of gas station price wars, where four dealers on the four quarters of an intersection reactively drop their prices to ridiculously unprofitable levels. One finally gives up, raises prices again, and the rest follow suit," says White. "The challenge to programmers will be developing more sophisticated agents that can react to scenarios such as this and make more educated decisions about pricing and tactics.

"We've no doubt more complex agents will evolve," continues White. "Such agents will have the ability to search vast amounts of data on past transactions, using data-mining and other techniques to model markets and make predictions. These agents will learn from past market cycles and mistakes, in order to avoid those situations as well as developing strategies to profit from them. With this learning ability, the range, variety, and complexity of market factors will increase exponentially."

That's why Institute researchers predict that agents will then develop unique approaches to differentiate their offerings and protect profits -- in effect, the spontaneous emergence of specialized online markets where agents can go to trade in certain goods, information, and services.

As information economies become more sophisticated, factors such as market friction -- for example, slow consumer response to a price change, or overhead costs associated with a price change -- also will influence agent behavior.

"We believe that hierarchies of agents will emerge to maximize efficiencies," says White. "For example, manager-agents will oversee the work of simpler agents, and agents will form associations to cooperate and compete in the market."

The IBM Institute for Advanced Commerce began operations on January 1, 1998, and is dedicated to exploring the impact of emerging technologies on the future of business and commerce. Its inaugural roster of advanced commerce research projects has initial funding commitment of more than $10 million and is supported by more than 50 scientists from IBM Research. These projects include a diverse mix of near-, mid- and long-term research efforts designed to overcome current barriers to robust and widespread electronic commerce, and are primarily focused on complex business-to-business applications.

# # #