These are great questions. It seems that your questions point out theoretical flaws in futarchy as a governance model, rather than our implementation specifically. We’re certainly open to ideas on how we can improve the implementation for this initial launch.
First I’ll say a few things about the signaling markets launch and what we hope to learn, and then I’ll address the more general questions about futarchy.
This will be the first application launched on Ethereum related to futarchy that we know of. The market topics will only be understood only by a small community, and participation rates will probably be low (maybe 10 - 20 individuals). Our primary goal is to test the UX of futarchy-app by presenting conditional markets to users on a topic that they are familiar with. We’re hoping to learn, directly from participants, their motivations for trading long or short on the various proposal markets. This along with trade data analysis will help us improve for future launches. It will not prove or disprove the viability of futarchy governance, but having data on a real implementation of conditional prediction markets might tell us something interesting. We’re thinking about this less like a scientific experiment and more like an early stage application launch - in general, we want to see what our users and data tell us, and try to improve for future iterations.
We’ve been asked about this market isolation problem before. If there are 2 conditional markets with the same success metric (price in this case), how does market A account for the conditional outcome of market B? If these were perfectly efficient markets, then prices would reflect all available information in the world. Part of the perfectly aggregated information for market A would be the likelihood of market B’s condition passing or failing, and the impact that market B passing or failing would have on the price given the condition where market A passes and where market A fails. This is complex to think about for 2 markets, let alone 10, 100, or 1,000 markets. Since in the real world there is no such thing as a perfectly efficient market, there is always an opportunity to compete. So it’s possible that as market participation scales up, new participants start accounting for cross-conditional info that isn’t already accounted for as a way to compete.
Let’s say traders bet on average that all of the proposals will increase the price of ANT. But then the price of ANT is lower than expected. Which proposal is to blame for the lower price?
We won’t know which proposal is to blame for the lower price. It’s possible that neither proposal was to blame - maybe there was a downturn in the crypto markets caused by a high profile hack. All we know is that the traders on the ANV signaling markets, at one time in the past, believed that the ANT price would increase if a set of proposals passed, and that they were willing to bet X amount of collateral on this.
What if the bettors on some proposals would have been correct if not for one of the other proposals being approved (a proposal they might have even bet against)?
Each individual trader should have factored in the likelihood of the other proposals being approved. If they risked collateral on a bet and lost value, another trader with better information gained that value.
Now let’s say that traders bet on average that half of the proposals will increase the price of ANT, and half of them will not (signaling that some AGPs should be approved and others rejected). But then the price of ANT is lower than expected. What if the price is lower because of the proposals that were rejected or because of the proposals that were approved? Meaning, the futarchy market results in the opposite of what you’d hope for: good proposals were rejected, or bad proposals were approved. And because all proposals are being judged by the same metrics, it’s impossible to tell which is which. All you can say is “well, the price of ANT is down, so all the people who bet it would be up if these proposals were approved lose”.
This is a tough one to answer. With low participation, imperfectly efficient markets like the ones we’re about to launch, this scenario is totally possible. If we were to use these markets to actually pass/fail AGP proposals, “good” proposals could be rejected and “bad” proposals could be accepted. Of course, this could also happen with token weighted voting. An individual proposal’s impact on ANT price will always be complex to determine, regardless of whether the proposal was approved with token weighted voting, futarchy, or any governance mechanism. Running signaling markets will not tell us what the ANT price impact of a proposal pass/fail decision was or is. But it might tell us what market participants thought the proposal’s impact on ANT price would be at various times throughout the history of a given market.