Information Finance: New Horizons Beyond Prediction Markets

Prediction markets, as an emerging blockchain application, are attracting more and more attention. As the founder of Ethereum, I have always had a strong interest in prediction markets. As early as 2014, I wrote about governance models based on predictions. In 2015, I actively participated in and supported the development of the prediction market Augur. In the 2020 U.S. elections, I made a profit of $58,000 through prediction markets. Recently, I have been closely following and supporting the development of Polymarket.

For many people, prediction markets are just election betting, albeit a different form of gambling. From this perspective, my enthusiasm for prediction markets may seem a bit unbelievable. But in reality, what I value is the deeper concept behind prediction markets. I believe that:

1. The existing prediction markets are already a very useful tool;

2. More importantly, prediction markets are just the pioneers of the broader field of "information finance." Information finance is expected to be applied in various areas such as social media, science, news, governance, and more.

Polymarket: The Dual Identity of a Betting Site and News Platform

Recently, in the U.S. elections, Polymarket demonstrated its advantages as an information source. In addition to providing a 60/40 win probability prediction, Polymarket directly revealed the truth when the results were announced: a certain candidate had a winning probability of over 95%, and the probability of taking control of all government departments was over 90%. This stands in stark contrast to the misleading reports from many experts and news sources.

The value of Polymarket is not only reflected in major events. During the Venezuelan presidential election in July this year, I saw on Polymarket that someone was willing to bet over $100,000 on the possibility of Maduro being overthrown at 23%. This caught my attention and made me realize that the opposition had adopted an unusual strategy that was worth noting. Without the initial signal from Polymarket, I might not have even noticed the significance of this event.

Of course, we should not blindly trust chart data. If everyone only looks at charts, those with capital could manipulate the market. Similarly, relying entirely on traditional news is also not advisable, as news often exaggerates situations for clickbait. A wise approach is to combine reading news with checking market data. If you see sensational reports but the market prediction probabilities have not changed, you should remain skeptical. Conversely, if the market shows unexpected high or low probabilities or sudden fluctuations, it is worth investigating the reasons behind them.

In summary, by combining news and market data, we can obtain more comprehensive information. For gamblers, Polymarket is a betting platform. For others, it is a news data website. Although one should not rely solely on chart data, I have made reviewing market data a part of my information gathering process, alongside traditional media and social media. This helps me acquire information more efficiently.

The Broad Prospects of Information Finance

Predicting election outcomes is just one application of information finance. The broader concept is to use finance as a coordinating incentive mechanism to provide valuable information to the audience. Some might argue that all finance is essentially related to information. Different participants make buy and sell decisions based on differing views of the future, and we can infer a lot of knowledge about the world from market prices.

However, information finance is a more precise discipline that requires us to:

1. Start from the facts you want to understand;

2. Carefully design market mechanisms to optimally obtain this information from participants.

A prediction market is a typical example: we want to know what will happen in the future, so we establish a market for people to bet on it. Another example is a decision market: we want to know whether decision A or B will yield better results, so we create a conditional market for people to bet on the metric values under different decisions.

Artificial intelligence is likely to have a huge impact on the financial industry in the next decade. This is because many applications of information finance involve "micro" issues: millions of small-scale decision markets. Traditionally, low-volume markets struggle to operate effectively because professional participants do not have enough profit incentive. However, artificial intelligence changes this situation, and even in markets with only $10 in trading volume, we may gain high-quality information. Even if subsidies are needed, the scale of subsidies for each issue is also manageable.

Information Finance: Refining Human Judgment

Assuming we have a trustworthy but costly human judgment mechanism, we want to access its "cheap version" in real time at a low cost. Robin Hanson proposed a solution: to establish a prediction market each time a decision needs to be made, predicting what results would be obtained if the expensive mechanism were called upon. The market operates with a small amount of funds subsidizing the market makers.

In the vast majority of cases, we will not actually invoke expensive mechanisms: decisions may be made directly based on the market average price. Only in very rare cases, perhaps through random selection or the market with the highest trading volume, will we actually run expensive mechanisms and compensate participants accordingly.

This provides a reliable, neutral, fast, and inexpensive "refined version" that roughly reflects the behavior of the original mechanism. Only the participants who help achieve this result can profit, while others will incur losses.

This method applies not only to social media but also to decentralized autonomous organizations ( DAO ). One major problem with DAOs is that there are too many decisions, leading to most people being unwilling to participate. If actual voting rarely occurs in a DAO and most matters are decided by the prediction market, with humans and AI jointly predicting the voting outcomes, then such a DAO may operate more effectively.

Other Applications of Information Finance

Personal tokens: Create tradable tokens for individuals that reflect expectations of their future status. Although current projects like friend.tech are not mature enough, they hold the promise of addressing important issues such as talent discovery through more prudent economic design.

Advertisement: Based on the "expensive but reliable signal" of actual purchasing behavior, information finance can help people make better purchasing decisions.

Scientific peer review: determining which research results need to be re-examined through prediction markets, helping readers assess the credibility of specific outcomes. Preliminary experiments have been successful.

Public goods funding: By tracking the entire dependency graph and determining each project's contribution to positive outcomes, it is expected to improve the "popularity contest" issue in the current funding mechanisms.

Conclusion

Although these ideas have been discussed for many years, I believe that information finance has great potential in the current decade, mainly for three reasons:

1. Information finance solves the trust issues that people commonly face in the fields of politics, science, and business.

2. Scalable blockchain technology provides the infrastructure to realize these ideas.

3. The involvement of artificial intelligence makes it possible to establish effective markets for small-scale issues.

To fully seize this opportunity, we need to go beyond election predictions and explore the broader application prospects of information finance.