An Introduction to Futarchy

https://gyazo.com/86a0d067f8abb3cac498bdf8aad233e5

---.icon

Our interest in futarchy, as explained above, is in a slightly different form and use case of futarchy, governing decentralized autonomous organizations and cryptographic protocols; however, I am presenting the use of futarchy in a national government first because it is a more familiar context. So to see how futarchy works, let's go through an example.

Suppose that the success metric chosen is GDP in trillions of dollars, with a time delay of ten years, and there exists a proposed policy: "bail out the banks". Two assets are released, each of which promises to pay $1 per token per trillion dollars of GDP after ten years. The markets might be allowed to run for two weeks, during which the "yes" token fetches an average price of $24.94 (meaning that the market thinks that the GDP after ten years will be $24.94 trillion) and the "no" token fetches an average price of $26.20. The banks are not bailed out. All trades on the "yes" market are reverted, and after ten years everyone holding the asset on the "no" market gets $26.20 apiece.

---.icon

The Argument For

https://gyazo.com/f8311d738bf7c0b63d66345f766fb9ad

---.icon

The Argument Against

1. A single powerful entity or coalition wishing to see a particular result can continue buying "yes" tokens on the market and short-selling "no" tokens in order to push the token prices in its favor.

2. Markets in general are known to be volatile, and this happens to a large extent because markets are "self-referential" - ie. they consist largely of people buying because they see others buying, and so they are not good aggregators of actual information. This effect is particularly dangerous because it can be exploited by market manipulation.

3. The estimated effect of a single policy on a global metric is much smaller than the "noise" of uncertainty in what the value of the metric is going to be regardless of the policy being implemented, especially in the long term. This means that the prediction market's results may prove to be wildly uncorrellated to the actual delta that the individual policies will end up having.

5. A prediction market is zero-sum; hence, because participation has guaranteed nonzero communication costs, it is irrational to participate. Thus, participation will end up quite low, so there will not be enough market depth to allow experts and analysis firms to sufficiently profit from the process of gathering information.

As long as the market power of people willing to earn a profit by counteracting manipulation exceeds the market power of the manipulator, the honest participants will win and extract a large quantity of funds from the manipulator in the process. Essentially, for Hanson and Friedman, sabotaging a futarchy requires a 51% attack.

The most common rebuttal to this argument, made more eloquently by Hewitt, is the "self-referential" property of markets mentioned above.

---.icon

Self-reference

Bitcoin has this effect in an especially pronounced way; when the price goes up, all Bitcoin users get richer, allowing them to build more businesses, justifying the higher price level. The lack of intrinsic value for Bitcoin means that the self-referential effect is the only effect having influence on the price.

Prediction markets do not have this property at all. Aside from the prediction market itself, there is no plausible mechanism by which the price of the "yes" token on a prediction market will have any impact on the GDP of the US in ten years. Hence, the only effect by which self-reference can happen is the "everyone follows everyone else's judgement" effect.

---.icon

Absolutes and differentials

From this, what we can surmise is that futarchy is likely to work well for large-scale decisions, but much less well for finer-grained tasks. Hence, a hybrid system may work better, where a futarchy decides on a political party every few months and that political party makes decisions. This sounds like giving total control to one party, but it's not; note that if the market is afraid of one-party control then parties could voluntarily structure themselves to be composed of multiple groups with competing ideologies and the market would prefer such combinations; in fact, we could have a system where politicians sign up as individuals and anyone from the public can submit a combination of politicians to elect into parliament and the market would make a decision over all combinations (although this would have the weakness that it is once again more personality-driven).

---.icon

Futarchy and Protocols and DAOs

Now, however, there is an entirely new class of entities for which futarchy might be much better suited, and where it may finally shine: DAOs.

https://gyazo.com/43357ff1fde286262208f06818ee0239

Essentially, what this protocol is doing is implementing a futarchy which is trying to optimize for the token's price. Now, let's look at some of the differences between this kind of futarchy and futarchy-for-government.

Second, note the different mechanism for how the markets work. In traditional futarchy, we have a zero-total-supply asset that is traded into existence on a derivatives market, and trades on the losing market are reverted. Here, we issue positive-supply assets, and the way that trades are reverted is that the entire issuance process is essentially reverted; both assets on all losing markets become worth zero.

The biggest difference here is the question of whether or not people will participate. Let us go back to the earlier criticism of futarchy, that it is irrational to participate because it is a zero-sum game. This is somewhat of a paradox. If you have some inside information, then you might think that it is rational to participate, because you know something that other people don't and thus your expectation of the eventual settlement price of the assets is different from the market's; hence, you should be able to profit from the difference. On the other hand, if everyone thinks this way, then even some people with inside information will lose out; hence, the correct criterion for participating is something like "you should participate if you think you have better inside information than everyone else participating". But if everyone thinks this way then the equilibrium will be that no one participates.

Also note that the optimization objective is simpler; the futarchy is not trying to mediate the rules of an entire government, it is simply trying to maximize the value of its own token by allocating a spending budget. Figuring out more interesting optimization objectives, perhaps ones that penalize common harmful acts done by existing corporate entities, is an unsolved challenge but a very important one; at that point, the measurement and metric manipulation issues might once again become more important. Finally, the actual day-to-day governance of the futarchy actually does follow a hybrid model; the disbursements are made once per epoch, but the management of the funds within that time can be left to individuals, centralized organizations, blockchain-based organizations or potentially other DAOs. Thus, we can expect the differences in expected token value between the proposals to be large, so the futarchy actually will be fairly effective - or at least more effective than the current preferred approach of "five developers decide".

---.icon

Why?

So what are the practical benefits of adopting such a scheme? What is wrong with simply having blockchain-based organizations that follow more traditional models of governance, or even more democratic ones? Since most readers of this blog are already cryptocurrency advocates, we can simply say that the reason why this is the case is the same reason why we are interested in using cryptographic protocols instead of centrally managed systems - cryptographic protocols have a much lower need for trusting central authorities (if you are not inclined to distrust central authorities, the argument can be more accurately rephrased as "cryptographic protocols can more easily generalize to gain the efficiency, equity and informational benefits of being more participatory and inclusive without leading to the consequence that you end up trusting unknown individuals"). As far as social consequences go, this simple version of futarchy is far from utopia, as it is still fairly similar to a profit-maximizing corporation; however, the two important improvements that it does make are (1) making it harder for executives managing the funds to cheat both the organization and society for their short-term interest, and (2) making governance radically open and transparent.

---.icon

まず futarchy について。

Vitalik が提案した futarchy を応用した DAO ガバナンスモデルについて。

あと、ガバナンスレイヤーの重要性を再認識できたように思う。

最後。

But experiments are what cryptocurrency is all about.