Jeffrey K. MacKie-Mason

Improving Learning Performance by Applying Economic Knowledge (Download full paper)

Christopher H. Brooks, Robert S. Gazzale, Jeffrey K. MacKie-Mason, and Edmund H. Durfee

Abstract: Digital information economies require information goods producers to learn how to position themselves within a potentially vast product space. Further, the topography of this space is often nonstationary, due to the interactive dynamics of multiple producers changing their positions as they try to learn the distribution of consumer preferences and other features of the problem's economic structure. This presents a producer or its agent with a difficult learning problem: how to locate profitable niches in a very large space. In this paper, we present a model of an information goods duopoly and show that, under complete information, producers would prefer not to compete, instead acting as local monopolists and targeting separate niches in the consumer population. However, when producers have no information about the problem they are solving, it can be quite difficult for them to converge on this solution. We show how a modest amount of economic knowledge about the problem can make it much easier, either by reducing the search space, starting in a useful area of the space, or by introducing a gradient. These experiments support the hypothesis that a producer using some knowledge of a problem's (economic) structure can outperform a producer that is performing a naive, knowledge-free form of learning.

Model Selection in an Information Economy: Choosing what to Learn (Download full paper)

Christopher H. Brooks, Robert S. Gazzale, Rajarshi Das, Jeffrey O. Kephart, Jeffrey K. MacKie-Mason, and Edmund H. Durfee

Abstract: In an economy in which a producer must learn the preferences of a consumer population, it is faced with a classic decision problem: when to explore and when to exploit. If the producer has a limited number of chances to experiment, it must explicitly consider the cost of learning (in terms of foregone profit) against the value of the information acquired. Information goods add an additional dimension to this problem; due to their flexibility, they can be bundled and priced according to a number of different price schedules. An optimizing producer should consider the profit each price schedule can extract, as well as the difficulty of learning of this schedule. In this paper, we demonstrate the tradeoff between complexity and profitability for a number of common price schedules. We begin with a one-shot decision as to which schedule to learn. Schedules with moderate complexity are preferred in the short and medium term, as they are learned quickly, yet extract a significant fraction of the available profit. We then turn to the repeated version of this one-shot decision and show that moderate complexity schedules, in particular two-part tariff, perform well when the producer must adapt to nonstationarity in the consumer population. When a producer can dynamically change schedules as it learns, it can use an explicit decision-theoretic formulation to greedily select the schedule which appears to yield the greatest profit in the next period.

Pricing Information Bundles in a Dynamic Environment (Download full paper)

Jeffrey O. Kephart, Rajarshi Das, Christopher H. Brooks, Edmund H. Durfee, Robert S. Gazzale and Jeffrey K. MacKie-Mason

Abstract: We explore a scenario in which a monopolist producer of information goods seeks to maximize its profits in a market where consumer demand shifts frequently and unpredictably. The producer is free to set an arbitrarily complex price schedule-a function that maps the set of purchased items to a price-but without direct knowledge of consumer demand it cannot compute the optimal schedule. Instead, it must employ a form of optimization based on trial and error. By means of a simple model of consumer demand and a modified version of a simple nonlinear optimization routine, we study a variety of parameterizations of the price schedule and quantity some of the relationships among learnability, complexity, and profitability. In particular, we show that fixed pricing or simple two-parameter dynamic pricing schedules are preferred when consumer demand shifts frequently, but that dynamic pricing based on more complex schedules tends to be most profitable when consumer demand shifts very infrequently.

Information Bundling in a Dynamic Environment (Download full paper)

Christopher H. Brooks, Rajarshi Das, Jeffrey O. Kephart, Jeffrey K. MacKie-Mason, Robert S. Gazzale, and Edmund H. Durfee

Abstract: Markets for digital information goods provide the possibility of exploring new and more complex pricing schemes, due to information goods' flexibility and negligible marginal cost. In this paper we compare the dynamic performance of price schedules of varying complexity under two different specifications of consumer demand shifts.

Two-sided Learning in an Agent Economy for Information Bundles (Download full paper)

Kephart, Jeffrey O., Rajarshi Das and Jeffrey K. MacKie-Mason

Abstract: Commerce in information goods is one of the earliest emerging applications for intelligent agents in commerce. However, the fundamental characteristics of information goods mean that they can and likely will be offered in widely varying configurations. Participating agents will need to deal with uncertainty about both prices and location in multi-dimensional product space. Thus, studying the behavior of learning agents is central to understanding and designing for agent-based information economies. Since uncertainty will exist on both sides of transactions, and interactions between learning agents that are negotiating and transacting with other learning agents may lead to unexpected dynamics, it is important to study two-sided learning. We present a simple but powerful model of an information bundling economy with a single producer and multiple consumer agents. We explore the pricing and purchasing behavior of these agents when articles can be bundled. In this initial exploration, we study the dynamics of this economy when consumer agents are uninformed about the distribution of article values. We discover that a reasonable albeit naive consumer learning strategy can lead to disastrous market behavior. We find a simple explanation for this market failure, and develop a simple improvement to the producer agent's strategy that largely ameliorates the problem. But in the process we learn an important lesson: dynamic market interactions when there is substantial uncertainty can lead to pathological outcomes if agents are designed with "reasonable" but not sufficiently adaptive strategies. Thus, in programmed agent environments it may be essential to dramatically increase our understanding of adaptivity and learning if we want to obtain good aggregate outcomes.

Automated Strategy Searches in an Electronic Goods Market: Learning and Complex Price Schedules (Download full paper)

Brooks, Christopher H., Scott Fay, Rajarshi Das, Jeffrey K. MacKie-Mason, Jeffrey O. Kephart and Edmund Durfee

Abstract: In an automated market for electronic goods new problems arise that have not been well studied previously. For example, information goods are very flexible. Marginal costs are negligible and nearly limitless bundling and unbundling of these items are possible, in contrast to physical goods. Consequently, producers can offer complex pricing schemes. However, the profit-maximizing design of a complex pricing schedule depends on a producer's knowledge of the distribution of consumer preferences for the available information goods. Preferences are private and can only be gradually uncovered through market experience. In this paper we compare dynamic performance across price schedules of varying complexity. We provide the producer with two machine learning method producer that is performing a naive, knowledge-free form of leanings (function approximation and hill-climbing) which implement a strategy that balances exploitation to maximize current profits against exploration of the profit landscape to improve future profits. We find that the tradeoff between exploitation and exploration is different depending on the learning algorithms employed, and in particular depending on the complexity of the price schedule that if offered. In general, simpler price schedules are more robust and give up less profit during the learning periods even though in our stationary environment learning eventually is complete and the more complex schedules have high long-run profits. These results hold for both learning methods, even though the relative performance of the methods is quite sensitive to choice of initial conditions and differences in the smoothness of the profit landscape for different price schedules. Our results have implications for automated learning and strategic pricing in non-stationary environments, which arise when the consumer population changes, individuals change their preferences, or competing firms change their strategies.