Summary

There are many things that we all would like to see realized, but no individual has an incentive to contribute to. Assurance Contracts have a novel payoff structure. We pre-commit to take an action —such as giving x amount of $/BTC/other cryptocurrency to the longevity/journalism/open source project of our choice— if and only if enough others commit to do the same to reach a critical threshold. Dominant assurance contract have the added condition that if the funding benchmark isn’t reached, the provider pays a prize to the pledgers. Pledging becomes a dominant strategy, or in Tabarrok’s words, “a no-lose proposition—if enough people pledge you get the public good and if not enough pledge you get the prize.” This meeting is part of the Intelligent Cooperation Group and accompanying book draft.

Presenters

Presentation: Dominant Assurance Contracts

This talk will span two papers on the private provision of public goods. First, we’ll hear about Public Goods, then about potentially better models for crowdfunding.

 

The Public Good Problem

  • Public Goods are a market challenge (not failure.) When it comes to a good that is truly in the commons, the rewards are shared equally, but the costs are not necessarily.
  • There are two main problems that prevent individuals from pitching in towards public goods:
  • The Free Rider problem: if others are contributing, and you’ll reap the public reward even if you don’t, and doing so would cost you something (money, time) then maybe you won’t bother…
  • The Assurance Problem: You’re worried that if you contribute when other people do not, then your contribution is wasted.
  • A case study: Averting a flood. The best outcome for a threatened town would be for everyone to pitch in hauling sandbags, but individual incentives may lead to some not contributing, which would prove fatal.
    • The Free Riders are thinking “eh, everyone else will help to save the town, I can skip it.” The Under Assured are thinking “man, if people don’t show up I’ll be out there stacking and working and STILL die in the flood!”
    • How can we change the rules to get more people to play?
    • The town’s mayor announces that helpers will meet at town hall and work will begin only if enough people show up to avert the flood (We’ll call this an Assurance Contract). This Assures all the helpers that their contributions won’t be in vain, but we still have a Free Rider problem. Altruism and social pressure help but may not be sufficient to get the people out.
    • So, the mayor announces that work will only happen if EVERYONE helps. This effectively eliminates free riders: if you don’t help, nothing will happen and your place is toast along with the town. It also provides assurance your efforts won’t be wasted. It does leave open an outcome where nobody helps and everyone loses though.
    • What to do?

 

Better Crowdfunding

  • Crowdfunding sites like Kickstarter use the Assurance Contract illustrated above quite successfully, but there’s a looming question: most projects fail. Why?
  • Certainly, there are some bad projects that simply don’t warrant funding, but there are plenty of perfectly good projects failing, too. This seems to point at too many of the “nobody pitches in and we all lose” scenarios playing out.
  • What can be done?

 

The Dominant Assurance Contract

  • The Dominance Assurance Contract is a stronger form of the Assurance Contract with an added feature: refund bonuses. Those who chip in to projects that don’t get enough support get a “refund bonus” for their trouble.
  • This makes contributing always a good move: you either win big with everyone, or you win small with a bonus.
  • This boosts participation, which thereby limits the cases where the bonuses are actually needed to be paid out.
  • The offer of the bonus does the lifting, not the bonus itself, much like bank insurance prevents bank runs with peace of mind.
  • The questions then:
    • How should refund bonuses be paid?
      • Fixed amount?
      • Proportional to contribution?
    • When should refund bonuses be paid? Should they be paid to everyone or only to early contributors?
    • It sounds great in theory. Does it work in practice?

 

To The Laboratory Experiments!

  • Alex has been running different crowdfunding experiments in the lab: simulated crowdfunding with groups of subjects playing with real money so as to give them some skin in the game. Many variations of refund schemes are represented: no refunds, large refunds, small refunds, refunds for early funders, refunds based on funding amount, etc.
  • What they found:
  • ANY refund increased project success by +20 percentage points across the board. The different schemes have some variation, but consistently better results.
  • Early contributors are important! It seemingly inspires others to fund. You can predict the success of a Kickstarter campaign with 85% accuracy after only 15% of their duration.
  • Refund bonuses paid only to early contributors was the most successful scheme: more early successes and more total successes than any other method.
  • Why do refund bonuses work? They make each potential contributor pivotal earlier, meaning the contribution point where a particular contributor would find it profitable to push the project over the threshold.
  • “By making people aware of their pivotalness earlier, there is a bonus to the early refund bonus”
  • In theory, refund bonuses are never paid: in reality, refund bonuses get paid ~33% of the time. So entrepreneurs must balance their bonuses paid out against their profits!

 

Other Approaches

  • “Liberal Radicalism” (https://arxiv.org/pdf/1809.06421.pdf) from Vitalik Buterin, Zoe Hitzig, and Glen Weyl brings Quadratic Voting into the mix and takes a completely different approach to funding public goods, effective in potentially different contexts.
  • Dominant assurance contracts work when you know what the public good is and you just need people to contribute. Quadratic funding is good to discover the public goods that people want and get them funded, but it does require outside investment.
  • Generally, mechanisms like Refund Bonuses, Dominant Assurance Contracts and Quadratic Voting can provide tools to entrepreneurs to unlock the market’s ability to create public goods, instead of dealing strictly with charitable or governmental activities.

Q&A

How does a dominance assurance contract solve the free-rider problem? The mayor requiring participation seems to not reflect in other uses.

  • Alex: Everyone signs up for the bonus right up until someone pushes it over the threshold, so all equilibria in theory are success states.

 

Why did your experimental subjects ever not contribute?

  • Alex: Well… there’s real money on the line, so there is positive incentive, but not really opportunity costs or other costs: they’re already in the lab.

 

What about projects that should NOT be funded but get funded because folks want the bonus?

  • Alex: Yeah you’re right… There could be a concept of Force Riders: folks jump on the wagon for bonuses and others are forced to pay for goods they didn’t want to pay for. We should study this.
  • Other: You see this in practice on the Gitcoin platform: projects boasting kickbacks and promises of future airdrops for supporting garner outsized support. They’re having to play with these incentive structures in the field to shoot for outcomes on the platform.

 

Kickstarter isn’t really producing public goods, because the whole idea is that you’ll get a public good even if you don’t contribute

  • Unanimous contract was the older solution to the same problem: you have a list of the people that would benefit, you have a price for each at or below what you believe to be the value of the good to each person. In a perfect information environment everyone agrees and the good is created. But in practice to can’t find these prices and things go awry.
  • In the unanimous case, The Dominant Assurance contract slightly increases the likelihood of everyone signing but costs more if you have to pay out bonuses, so doesn’t seem like really an improvement.

 

What is the most exciting near and long-term use case of these contracts?

  • Alex: Local public goods make a lot of sense to me: improving your neighborhood. And of course, smart contract based funding systems for blockchain protocols changes could certainly be a use case.

 

Lightning Talk: Identity-Disclosing Assurance Contracts

Slide reel:

  • Discussion:
    • The whistleblowing case seems particularly problematic: how do you set up the requirements and protect identities?
      • Anthony: Project Callisto does this for sexual harassment situations and that sorta like whistle-blowing, but you’re right. It’s a hard case.
    • Is there a way to make these contracts dominant? Is there something to a “bounty” system or incentives for whistle-blowing first or early? refund bonuses?
      • Alex: well, there’s no end point, you want to collect information forever until you’ve got what you need. There’s no point at which to say “ok, it failed, pay out the bonuses”
      • Anthony: I suppose you could have a timer set that you pay out after… though you’d need to be able to pay out. Involving money leads to very different dynamics. But in principle there’s no reason you couldn’t reward on failure like the dominant contract.
    • Allison: What about using assurance contracts for reaching escape velocity in longevity funding? …. maybe a prediction market for how much it will cost to accomplish hallmarks of aging milestones that ensure folks currently alive hit escape velocity
    • Other: Something interesting with funding is that once you start talking about REALLY large amounts of funding, the commitment of those funds is almost more important than actually having the money in the bank. So there are potentially REALLY high funding levels where the anonymous method is better than the dominant.

Lightning Talk: GoodX

Motivation

  • Motivated by Xrisk: human coordination seems like the bottleneck!
  • Can we get a group of people moving to the same physical location? How do you verify people will actually do what they commit to?
  • A very early MVP at conditionalcommitment.com! Check it out.

 

What can people do to help with your projects?

  • Anthony: If you know some expert cryptography folks, please send them my way!
  • Dony: I need a programmer! I have funding! (also let’s talk about impact certificates)
  • Alex: We want to do a field experiment: if you know folks at crowdfunding companies let me know!

 

A related proposal for rooting out corruption: on every voting ballot, you also vote for who you think is the most corrupt. Preseves privacy, gets information. The key is making people feel really safe before they put their name on something.

  • Anthony: Yeah, agreed. My approach is to start with lower stakes contexts instead of government secrets and corruption. Then we can build trust in the system over time.

 

Have you figured out how to encrypt contracts on-chain and pull them down trustlessly?

  • Anthony: not really, we are putting the requirements and such on the blockchain to prove that they weren’t changed later.
  • Folks have built encrypted chat between users on the Bitcoin SV blockchain

 

Seminar summary by James Risberg.