Why & How Decentralized Prediction Markets Will Change Just About Everything.

Excuse the superlatives. I’m usually a lot more guarded with these. However, the more I think about it, the more I feel censorship-resistant & decentralized prediction markets will change almost everything about our society, and it has to do with one core difference: the automation of it. An open, permission-less market for rewarding those (humans and programs alike) who predict the future.

It will hold implications for sharing wealth, redefining organisations in the digital age & even help stop climate change. It sounds grand, but here’s how it will happen.


Prediction markets (or the original coined term by Robin Hanson: “idea futures”) aren’t new. What are they?

Prediction markets (also known as predictive markets, information markets, decision markets, idea futures, event derivatives, or virtual markets) are exchange-traded markets created for the purpose of trading the outcome of events.

You essentially bet against each other (the market) how an outcome will turn out. For example, Bernie Sanders will be the Democratic candidate vs Hillary Clinton. You buy shares of each outcome that basically correlate to a percentage chance that the event will occur. Once the event has occurred, the prediction market will allow your shares to be redeemed for $1, while the other shares become worthless. For example, if you buy a Bernie Sanders outcome at $0.6 and Hillary Clinton outcome at $0.4, and Bernie becomes the candidate, you believed with a 60% chance that it will happen. Your Bernie token becomes worth $1. And your Hillary Clinton token becomes worth $0. Getting tokens is as simple as paying $1, upon which you get both. You can then precede to buy or sell them with others at various prices. If you don’t trade either of the outcomes that you bought, you will just get your money back (since one will go to zero and the other to 1).

Prediction markets have existed and still do exist. Intrade was popular but had to exclude US traders. It’s not an easy space to be in, since in some jurisdictions it is seen gambling, while in others it is seen options trading. The other worry is that it could create controversial incentives such as predicting the death of a global leader.

A global prediction market has thus not flourished as well as it could have. Even if it worked properly, building on top of it as a platform (APIs and such) is also not an easy job, and there’s little guarantee that there won’t be another clamp down.

Enter Decentralization

Decentralized systems where innovation can happen without permission have allowed new (& old ideas) to flourish in wondrous new ways. We wouldn’t have Facebook, Wikipedia & Twitter if the web wasn’t open. Free permission to innovate with information has led to where we are today. An open prediction market platform will come. It will most likely come to live on a blockchain. It not only helps with maintaining an open infrastructure for it, but it also allows separation of concerns (who is doing what in this market).

Currently there are 3 known decentralized prediction market efforts underway: Truthcoin/Hivemind (Bitcoin-based sidechain), Augur (recently raised $5.3m from a crowdsale, built on Ethereum) & Gnosis (working prediction market platform on Ethereum). I will focus on what this will look like on Ethereum.

Ultimately, a prediction market will exist that will allow anyone to create markets, bet on outcomes and resolve/report the outcomes. The difference comes in when you add the following to this:

An Arbitration Market for Reporting:

At the end of the limit the outcome of an event must be reported. In the past, this was usually reported by the people who ran the prediction market itself (and you had to trust them to report correctly). With a decentralized system you can swap this out for various systems. A market for an event can have one person decide.

If this person is trusted, then liquidity will come. If they are not, then multiple persons can report an outcome (where 2 of 3 need to agree, for example). Market participants can vote for who they want to report as well. Systems such as Augur has a token system where those who hold the token, vote on outcomes as a crowd. All these styles are swappable. In some circumstances you wouldn’t even need a report to happen. If the information is already on Ethereum, the resolution will happen without requiring a trusted source. ie, if someone is selling their song on Ethereum, and that is publicly accessible by other contracts, you can use that as is.

Finally, and perhaps the most interesting, is that all you need is a threat of an outcome for the market converge to the right outcome. The closer to the time an event comes, the more it starts to converge to the actual outcome as clarity increases. Thus, in a way, the tokens become worth zero on the one side and 1 on the other, automatically resolving itself. In a scenario where this actually ended up wrong, users can put up a deposit to dispute it: which results in arbitrator that has to come in and decide.

Automatic Betting:

This is arguably the most important. If it is an open layer any program can start predicting. Prediction markets only played by slow humans who have do the thinking won’t ever have enough liquidity to be useful. Programs can absorb much more and make much better predictions about the future.

Automatically gathering knowledge about the world & combining it correctly will result in financial gain. Thus, companies like Google & Facebook will be at a considerable advantage, betting on these prediction markets. Dumb sensors in every avenue could either predict themselves or sell the information (more on this later). And finally, you can build bots that predict based on some model: it could be based off a person, a group, or any combination (also more on this later).

The difference here vs a traditional market maker is subtle. The purpose is not facilitate trading, but to automate predictions.

The result?

Once you add these parts, you get wonderful potential emergent behaviour. Here are some examples.

Information Markets.

You sell your information to be used in markets. If the information is in the same ecosystem as the prediction markets (say, Ethereum), then you can sell this information to be used in a trust-less manner inside the market itself. This is a holy grail for several reasons. The oft talked about “Internet of Things” will be extremely useful here. A sensor can produce information, & put in the blockchain (public). If it used for reporting, it can charge for that information. It could also result in markets where the information is not made public, but encrypted, upon which it can then be sold to others to gain knowledge to predict more effectively. So, information markets can develop around selling information to help predict & report. This won’t just exist for sensor data. It would eventually become the norm to report any kind of data into the blockchain, especially for reporting purposes. It not only means that you can have audited & transparent information in there, but it also means you immediately provide a trust source for information to be bet on, at very little cost to the producer. For example, you decide to log your monthly revenue of your company to the blockchain.

New kinds of organisations.

With the rise of social media, we saw its capability to move people together in new ways: driven by a common goal, bereft of normal bureaucratic process. The Arab Spring is a good example, where these movements remain relatively head-less. Recently, in South Africa a movement to reduce tertiary tuition fees rallied around a hashtag (#feesmustfall), that formed the locus of the liquid organisation.

These organisations are unlike what we’ve seen before, because social media allows near instant communication and allows important news & events to immediately filter up (based on retweets) and then subsequently affect and inform the rest of the organisation. Affiliation is as easy as using the hashtag. It’s the network’s version of an organisation. There is no permission to be a part of it.

How does one build ways to incentivise these new organisations (financial gain) & how do you help it make decisions? You use prediction markets. Just as these hashtag organisations move like crowds do, so should its decision making. As the organisation goes about its goals, various outcomes are constantly generated, upon which the people in the organisation and those outside of it, bet on the outcomes, leading it automatically towards outcomes which serve the goals of the organisation.

Since these network organisations move at the speed of social media, you might need some help from our bot friends to bet on your behalf. And this leads to the next part:

Wealth sharing.

If prediction markets will be so useful in creating wealth for those are in the know, then you might want to developed automated personalised prediction bots. These bots automatically bet on events based on who you are & what you do. If you join a hashtag organisation (the tweet appears in your feed), then the bot automatically detects this and assumes this as indicator that marginally this movement might succeed and thus bets on those outcomes (resulting in automatic financial gain). Another example, is where if you frequent a coffee shop, your bot will automatically start betting on the revenue that will be posted by this shop. You are directly influencing its success by being a patron and thus you can partake in the financial gain of it.

This presents a potentially whole new way of looking at organisations. Perhaps into the future we won’t even need things like shares/equity. Money simply flows towards some locus, upon which it is used to improve a metric that can be predicted upon. It paints a new model of “investment”. If you donate $10 000 to an organisation or group, you know that its metric of success will improve and thus you can benefit from that financial gain. There are countless metric which could be influenced in this manner, just by “being alive”, you affect them. It doesn’t have to be financial even (I’m developing an interesting application that is non-financial that will soon launch).

This opportunity will become available to all, at any scale.


Futarchies is a governance system where a metric is chosen (say GDP in 10 years). Outcomes are bet based on certain decisions (GDP in 10 years if investment in education). If yes -> what will GDP be? If no -> what will GDP be? Then you implement the market with the higher forecast. After 10 years, you reward those who predicted correctly.

Millions of programs & countless humans predicting the future will in anyway help current governments choose what to do (without even resorting to a futarchy).

Protecting Natural Systems.

Could we use prediction markets to protect our climate & environment?

Karl Schroeder describes such a potential system in this forum post. In essence you have AI automatically interacting with a natural system, in order to protect it. At the basis of this could be prediction markets. A metric could be something like: growth of new trees in an area.

If I am interested in seeing the ecosystem flourish, I can get financial gain from it, by simply protecting & fostering it (you first predict, then enact the change). You make sure that a certain amount of trees are planted. Sensors can provide much more nuanced feedback and reporting vs more concrete outcomes (such as trees planted). For example, measuring pollution. This information will be sold to interested parties. To bring these sensors into existence they can be crowdfunded by a group of environmentalists, who will earn these fees, which subsequently can result in further creating a sustainable ecosystem.

Decentralized AI Systems?

What happens if you have automated sensors flying around, collecting information for predictions? I’ll you let you do the rest of that thinking. ;)


A prediction market is a powerful idea. A decentralized prediction market is an even more powerful idea. Once we combine the capability to automate predictions combined with AI, Machine Learning and the Internet of Things, it becomes something that indeed will change just about everything. It will result in basically being able to model externalities in a much better fashion.

Patrick Collison of Stripe asked this question himself this year.

And, yes, it would be a huge breakthrough, and decentralized prediction markets can do this. So excuse the superlatives, but from this vantage point, I’d bet on it.

by Simon de la Rouviere, Engineer of Societies

At ConsenSys with Gnosis, we are aiming to build this generalised platform. We will even soon be able to support predictions that scale to millions of sensors.

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.