Governance of Decentralized Autonomous Organizations




 

Instructors: Andreas Park and Zissis Poulos




 

Agenda for the presentation

  • What is a firm?
  • What is a DAO?
  • What are the challenges with governance?

What is a firm?

the firm begins where contracts end
(paraphrased from Ronald Coase 1937)

Ideas for Organizations/Firms

  • Firm emerges when entrepreneur starts hiring instead of contracting
     
  • Why move from contracting to hiring?
     
    • Must be cheaper to build an organization that contracting
       
    • What makes it cheaper?
      • transactions/contracting costs (contingencies hard to specify etc)
      • team work

What is a DAO?

DAO = Decentralized Autonomous Organization

no central leader

runs without external influence

suggests something that binds everything together

Decentralized Autonomous Organization

  • First: The DAO:
    • Vitalik Buterin on The DAO: "This program was to serve as a replacement for centralized companies."
    • decentralized venture capital fund which allowed voting on projects worth investing in
    • accumulated almost 13 million in ETH ~ $150 million
    • the largest crowdfunding campaign in history at the time
  • Philosophy:
    • break corporation boundaries
    • communal ownership of service provision
    • eliminate managers
    • organize communal decision making

Types of DAOs

  1.  Investment funds
    • collect money
    • decide on which projects to support
       
  2. Smart contract control
    • control parameters of a blockchain code (fees, rates)

Disclaimer: more structures and ideas will emerge

Problems with DAOs

  1. Functions/Security
    • once set up, code is hard to change
    • vulnerabilities in code can lead to exploits (main example: The DAO)
       
  2. Legal status
    • Is a DAO legal at all? What is it? Who or what is liable for damages if they arise?
    • Wyoming, US: DAO are legal entities since mid-2021
  3. Decision making
    • decision making is difficult
    • compared to a firm, is it actually a solution?

Sidebar:

  • Corporate Governance is largely concerned with lowering agency costs.
  • DAOs to date do not create agencies.
  • (Actually "Organization" receives little attention in DAO design.)
  • Is that the next step?

Currently: governance = decision making= voting

Challenges with governance

Organization of UNISWAP

UniSwap Lab supports development

a website app accesses the code

 governance (UNI) token holders control contact features

don't own the code

operation = decentral

control = decentral

anyone can use the baseline code

core code runs on the blockchain

Source: Sybill.org

background: 3m=0.3%, 30m=3% of UNI tokens

Bottom line: governance can be hard because of lack of engagement!

The Three Big Questions

  1. Accountability
    • governance voters are collectively accountable for their decisions (poor decision => everyone’s coins drop to zero), but no voter is individually accountable
       
  2. Public goods provision
    • how to you incentivizes contributions to protocol development, upgrade and maintenance?
       
  3. Voting:
    • who, on what, and how?

Big Questions for Voting

  1. Who should vote?
     
  2. How to vote?

Who should vote?

  • coin holders: one coin = one vote
    The Problem: governance power without economic interest
    • borrower has governance power without economic interest
      lender has economic interest without governance power
    • ​borrow coins on platform short term
  • humans (proof of personhood):
    • network participant = one vote
    • A fair reflection of economic interest?
       
  • contributors (proof of participation):
    • people who made meaningful contributions to a network
    • What's meaningful?
    • Does this translate into a meaningful stake?

Big idea: blockchain does allow new and different vote participation rules!

How to vote?

  • can be very subtle
     
  • status quo vs new proposal (often has a status quo bias)
     
  • voting among alternatives plus status quo
     
  • single votes, resource vote ("quadratic voting"), ranked ballots, etc

The Three Big Questions

  1. Who should vote? e.g.,
    • coin holders: one coin = one vote
    • humans: network participant = one vote
    • contributors: people who made meaningful contributions to a network
  2. What should you vote on?
    • parameters​
    • one-off choices
    • pairwise
  3. How should you vote?
    • one at a time
    • ranked ballot

Impossibility Results

  • Bad news: there a quite a number in social choice...
     
  • Example: Arrow's Impossibility Theorem
     
  • Result says a ranked-voting electoral system cannot reach a community-wide ranked preference by converting individuals’ preferences while meeting all the conditions of a fair voting system.

Sidebar:

Ken Arrow said this: "Most systems are not going to work badly all of the time. All I proved is that all can work badly at times."

Other issues

  • 2000 US Presidential Vote with
    • Bush vs. Gore vs Nader
    • Most liberals preferred Gore and Nader to Bush
    • Suppose 55% for Gore/Nader and 45% for Bush
    • If liberals split 50/50, Bush wins.
  • Vote splitting and strategic voting is very real issue, especially in Canada (both on the left and, most recently, on the right)
     
  • Some voting rules: Majority, Super-majority, two-tier voting with delegates, plurality (among more than 2 candidates, the one with the most votes), ranked voting (e.g. with Borda count (top gets n points, next n-1, etc. )

Learning from Voting on Candidates

Desirable features (binary elections)

  1. Anonymity: If vote voters switch votes, the outcome should be the same.
    • doesn't hold for many delegated voting scenarios
  2. Neutrality: If A wins and we flip everyone's vote, then B wins.
    • Violated for the supermajority with status quo
  3. Positive responsiveness: take two voting profiles which differ by a single vote (the second has one more vote for A). If the first has a tie, then  A wins the second. If A wins the first it also wins the second.

May's Theorem: In binary voting, only the majority rule satisfies the above three requirements

Desirable features (multi-candidate elections)

  1. Plurality is problematic:
    • adding a candidate often changes the outcome
    • the winning candidate in a many-horse race may lose by a wide margin against each loser
  2. Ranked choices don't work either (see the impossibility theorem)
  • Condorcet Criterion: a candidate who wins every binary comparison
  • Ranked pairs: order binary comparisons by margin of victory and discard "conflicting" binary pairs with smallest margin.
  • Instant run-off:
    • People submit ranked choice
    • Round 1: eliminate worst candidate, re-allocate votes of those who voted the lower #1 to their second choice
    • Round 2+: repeat

The big take-home:

  • voting is tricky
  • there is no universally optimal rule

Quadratic voting (hot topic at the moment)

  • basic idea:
    • solve the "how much you care" problem
    • you pay for your vote
    • your marginal cost is increasing (to reflect how much you care)
  • how done
    • people are endowed with "voice credits"
    • to vote on a topic a voter chooses voice credits \(v_i\) at a cost function \(c(v_i)\)
    • votes to purchase: weighs marginal cost of additional vote against chance to be pivotal
    • voters are "price takers" and agree that pivotal means \(p\) votes
    • voters have utility \(u_i\) for winning \[\max_{v_i} 2p v_i u_i-c(v_i)\]
    • Result: quadratic cost is the only rule that ensures that commonly best utility is implemented (and that people vote as such)

Lalley, Steven and Weyl, Eric Glen, Quadratic Voting: How Mechanism Design Can Radicalize Democracy (December 24, 2017). American Economic Association Papers and Proceedings, Vol. 1, No. 1, 2018, Available at SSRN: https://ssrn.com/abstract=2003531 or http://dx.doi.org/10.2139/ssrn.2003531

Summary

Summary: DAO Governance

  1. Blockchain tech would allow a number of governance and voting models, but more work is needed
     
  2. DAO governance requires careful planning: a century of research taught us that there is no "optimal" off-the-shelf rule yet
     
  3. DAO governance has not even touched upon the hard problems of agency conflict when being a genuine organization

@financeUTM

andreas.park@rotman.utoronto.ca

slides.com/ap248

sites.google.com/site/parkandreas/

youtube.com/user/andreaspark2812/

Governance of Blockchains and DAOs

By Andreas Park

Governance of Blockchains and DAOs

  • 580