🔬 condensed matter

Data as Commodity: a Game-Theoretic Principle for Information Pricing

This paper proposes a game-theoretic framework for pricing data as a non-rival commodity by modeling strategic competition among players with asymmetric information, revealing that Nash equilibrium analysis yields unique market dynamics—such as zero-cost sharing and counterintuitive rivalry effects—that establish a theoretical foundation for valuing intangible goods in digital markets.

Original authors: Pasquale Casaburi, Giovanni Piccioli, Pierpaolo Vivo

Published 2026-09-16

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Pasquale Casaburi, Giovanni Piccioli, Pierpaolo Vivo

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the digital age, data has become the most valuable resource on Earth, fueling everything from artificial intelligence to financial markets. Yet, unlike physical goods such as wheat or oil, data behaves in ways that break the standard rules of economics. A bushel of wheat is consumed when eaten; if you sell it to one person, it is gone for everyone else. Data, however, is non-rivalrous. A single dataset can be sold to a thousand buyers simultaneously without the original owner losing a single bit of information. It can be copied at almost no cost, and its value is often hidden behind complex legal licenses. This creates a paradox: how do you put a fair price tag on something that can be infinitely replicated and shared without being used up? Traditional supply-and-demand models, which rely on scarcity to drive prices, struggle to explain this market. Without a clear way to value information, massive markets have formed on shaky foundations, leading to situations where tech giants pay billions for datasets while individuals give away their personal information for free services.

To solve this puzzle, a team of researchers from King's College London has proposed a new way to think about the price of data. Instead of looking at how much it cost to collect the data or how much privacy a person loses, they built a mathematical model based on strategic competition. Imagine a group of people betting on the outcome of a coin toss. In this scenario, some players have seen more past flips than others, giving them a better idea of whether the coin is weighted toward heads or tails. The researchers treated the data owners and buyers as players in this game. One player, the seller, holds a long history of past results that no one else has. Another player, the buyer, has only a short, public history. The seller can choose to keep this advantage to herself or sell a piece of her private history to the buyer. The core question the team asked was simple: under what conditions would both the seller and the buyer agree to a deal, and what would a fair price look like?

The researchers found that the answer is far more complex and surprising than common sense suggests. They discovered that the value of data is not a fixed number but depends entirely on the strategic landscape of the market. In their model, they identified several distinct scenarios. Sometimes, the seller and buyer can cooperate to exploit the less-informed players in the game, making a deal that benefits both at a positive price. In other cases, the competition between the two informed players becomes so fierce that they actually hurt each other, making a trade impossible. Perhaps the most counterintuitive finding is what the authors call a "symbiotic" regime. In this specific situation, the seller can give her data to the buyer for free, and both end up better off than before. This happens because the exchange changes their betting strategies in a way that allows them to avoid competing against each other directly, effectively turning a zero-sum game into a win-win for the two of them, even though they are giving away valuable information.

The study also revealed a phenomenon the researchers describe as a "blessing of ignorance." When the two best-informed players compete intensely after a trade, their conflict can sometimes generate positive outcomes for the least-informed players in the room. While the two experts fight over the edge their data provides, they may inadvertently create conditions where the average player, who knows nothing, ends up winning more often than expected. This suggests that having more information does not always guarantee a competitive advantage; the structure of the market and the interactions between players can sometimes favor those with less knowledge. Furthermore, the size of the market matters significantly. The researchers showed that a trade that is impossible in a small group of players can become mutually beneficial when the number of participants grows. As the market expands, the direct competition between the informed seller and buyer weakens, allowing deals to happen that would be too risky in a smaller setting.

The team also explored what happens when players are not just interested in winning money but are also afraid of uncertainty. In the real world, people often dislike volatility; they prefer a guaranteed small gain over a risky large one. When the researchers added this fear of uncertainty to their model, the price of data changed again. Now, the value of a dataset depended not just on what the data said, but on how much of it there was. A longer string of data reduced the uncertainty about the coin's bias, making the information more valuable to a risk-averse buyer. This means that in markets where players are cautious, the sheer volume of data becomes a key part of its price, a factor that disappears if players only care about the average outcome.

Ultimately, this work provides a theoretical foundation for pricing intangible goods in a digital economy. It moves beyond simple cost-based calculations or subjective privacy loss to show how value emerges from the complex interplay of competition, information, and strategy. The findings suggest that the market for data is not a static place where prices are set by supply and demand, but a dynamic system where the act of sharing information can fundamentally alter the rules of the game. By treating data trading as a strategic interaction, the researchers have uncovered a landscape of possibilities where sharing information can sometimes be more profitable than hoarding it, and where ignorance can occasionally be a strategic advantage. These insights offer a new lens through which to view the trillion-dollar data economy, suggesting that fair pricing requires understanding not just the data itself, but the intricate web of relationships between those who hold it and those who want to buy it.

Technical Summary: A Game-Theoretic Principle for Pricing Data as a Commodity

Problem Statement
Data functions as the central commodity of the digital economy, yet it defies classical supply-and-demand pricing theories. Unlike physical goods, data is non-rival (sellable to multiple buyers without degradation), replicable at near-zero marginal cost, and subject to heterogeneous licensing. This creates a pricing paradox: traditional models relying on scarcity fail to capture data's value, leading to a trillion-dollar market built on shaky foundations, systematic inefficiencies, and a lack of rigorous valuation frameworks. Existing approaches (cost-based, privacy-based, utility-based, or seller-centric) often isolate specific dimensions of the problem, abstracting away the interconnected, multi-agent strategic complexity through which data value endogenously emerges. This paper addresses the need for a robust valuation principle for intangible goods in interacting digital markets.

Methodology
The authors propose a game-theoretic framework where the value of a data string emerges from strategic competition among $N \ge 3$ players betting on a binary outcome of an underlying stochastic process. The model is deliberately stylized to ensure analytical tractability:

The Game: Players pay a unit fee to bet on the next outcome ( $w \in \{0, 1\}$ ) of a hidden stochastic process. They interpret observed data as independent and identically distributed (i.i.d.) tosses of a possibly biased coin.
Information Asymmetry:
- Public Data: All players share a public string $\vec{s}_0$ of $R$ past outcomes.
- Private Data: Player 1 ( $P_1$ ) holds a longer private string $\vec{s}_1$ ( $R_1 > R$ ).
- Transaction: $P_1$ may sell a substring of her private data to Player 2 ( $P_2$ ), creating a new string $\vec{s}_2$ for $P_2$ ( $R_1 \ge R_2 > R$ ).
- Ignorance: Players $i \ge 3$ are "least-informed," relying only on $\vec{s}_0$ . Crucially, under Assumption (A5), they are unaware of the existence of better-informed players or the transaction itself.
Equilibrium Concept: Players do not share a common view of the game. Instead, each player $i$ computes a subjective Nash equilibrium based on their own information and their beliefs about the information held by others. $P_1$ can infer the strategies of less-informed players, but less-informed players cannot compute $P_1$ 's strategy.
Utility Functions: The study analyzes two player types:
1. Volatility-neutral: Utility is the expected monetary gain.
2. Volatility-averse: Utility includes a penalty term for the variance of the estimated coin bias, making the value of data dependent on both its content (bias) and its size (sample length).
Pricing Principle: The fair price range $[\Psi_{min}, \Psi_{max}]$ $[Ψ_{min}, Ψ_{ma x}]$ is derived by comparing utilities before and after the transaction:
- $\Psi_{min}$ : The minimum price $P_1$ demands to compensate for the loss of exclusivity (reduced utility due to increased competition).
- $\Psi_{max}$ : The maximum price $P_2$ is willing to pay, defined as the increase in her expected utility.
- A transaction is mutually beneficial if $\Psi_{max} > 0$ and $\Psi_{min} \le \Psi_{max}$ .

Key Contributions and Results
The analytical derivation of the Nash equilibrium reveals a rich landscape of counterintuitive market dynamics that depart from textbook economic intuition:

Strategic Regimes: The feasibility of trade depends on the beliefs ( $\hat{\rho}_1, \hat{\rho}_2$ ) held by the seller and buyer. The authors identify six distinct regimes:
- Cooperative: Trade occurs at a strictly positive price.
- Competitive: Trade is unviable because the seller's expected loss exceeds the buyer's potential gain.
- Symbiotic: A paradoxical regime where data sharing benefits both parties even at zero price. This occurs when the strategic shift in betting (e.g., $P_1$ and $P_2$ adopting opposite pure strategies) prevents them from splitting the pot, thereby increasing their individual utilities at the expense of others.
- Exploitative Cooperation: Trade is feasible only if the seller pays the buyer to offset the misleading nature of the data.
- No-deal: No mutually beneficial agreement exists.
The "Blessing of Ignorance": The model uncovers a phenomenon where fierce competition between highly informed players ( $P_1$ and $P_2$ ) generates positive utilities for the least-informed players ( $i \ge 3$ ). Conversely, the most informed player may suffer negative returns despite superior knowledge. This demonstrates that information abundance does not guarantee competitive advantage; complex interaction effects can favor less-informed parties.
Market Size Effects: As the number of players $N$ increases, the competition effects between the informed buyer and seller weaken. Transactions that are unviable in small markets (due to high competition) can become mutually beneficial in larger markets. In large markets, the privileged player becomes indifferent to the buyer's presence because there are many uninformed players to exploit.
Volatility Aversion and Data Size: For volatility-averse players, the equilibrium strategy and price bounds depend explicitly on the length of the data string ( $R_i$ ), not just the estimated bias. Longer strings reduce uncertainty, increasing the value of the data. Consequently, the price bounds become sensitive to the amount of data exchanged.

Significance and Claims
The paper claims to establish a theoretical foundation for pricing intangible goods in interacting digital markets. By treating data owners, buyers, and the market as a complex, adaptive system, the framework moves beyond ad-hoc valuation methods.

It provides an analytically tractable proof of concept where value emerges from strategic competition rather than static attributes.
It challenges the standard intuition that sharing exclusive information necessarily harms the seller, revealing "symbiotic" regimes where sharing increases utility for both parties without monetary compensation.
It highlights that market structure and the number of participants are critical determinants of data value, suggesting that small-market intuitions may fail in larger, more complex digital economies.
The framework is presented as a versatile tool for exploring foundational questions about data valuation, capable of being extended to multi-round settings, different network structures, and more sophisticated subjective models of data.

More like this