I discovered the idea of Radical Markets a few months ago through the brilliant talks of Microsoft Researcher Glen Weyl which are now captured in the book of the same title. Some of the concepts behind Radical Markets have permeated into the blockchain world as Vitalik Buterin embraced the concept in a blog post. Last week, Butterin and Weyl together with Harvard economist Zoë Hitzig published a paper that explores many of the ideas of the Radical Markets theory a bit deeper. Titled, “Liberal Radicalism: Formal Rules for a Society Neutral among Communities,” the method enables groups to allocate funds for the maintenance of public goods and services without becoming vulnerable to traditional challenges such as the “free-rider” problem. Central to Buterin-Weyl thesis is the idea of Quadratic Voting(QV) as a mechanism to make decisions in societies. While the principles behind QV are mathematically appealing, I found them really hard to implement in the current generation of utility token protocols. However, I found QV incredibly relevant to the nascent space of security tokens.
A good part of my Saturday was spent going thru the math in Buterin-Weyl’s paper and exchanging emails with a couple of thought leaders in the security token space about the relevance of QV to crypto-securities. It turns out that the principles behind QV can be used to address some of the imminent, and often ignored, governance challenges that we desperately need to tackle in the security token space. I’ve written about the topic of governance and security tokens before and I believe is one of the main barriers preventing security tokens to be embraced by large institutional investors.
Voting, Governance and Security Tokens
Activist investors are one of the most extreme examples of voting dynamics when comes to public equities. Some of the biggest names in Wall Street such as Icahn Enterprises(Carl Icahn), Elliott Management(Paul Singer) or Greenlight Capital(David Einhorn) have made their names and fortunes by running activist campaigns. In the activist model, an investor silently acquires a small but influential position in a company and then goes public proposing a series of changes that should benefit the company’s shareholders. Ultimately, the goal of the activist is to force a shareholder vote about the proposals in order to implement drastic changes in the target company. While I am not a big fan of activist investors, they are an example of a segment of the market that has taken full advantage of the voting mechanisms in public equities. Similarly, security tokens need to enable voting as a first class citizen of the platforms in order to attract mainstream investors.
Voting dynamics play a key role in the lifecycle of a security token. Let’s take the canonical example of a crypto-security that represents a pool of real estate leases and pays a dividend to token holders on a quarterly basics based on the performance of the underlying asset. If designed properly, many decisions about the behavior of that security token will be subjected to the vote of the token holders.
While tier-1 blockchain protocols are confronted with voting decisions in areas such as protocol changes or transaction verification, security token protocols should enable the implementation of voting and governance rules that reflect market, industry and regulatory dynamics.
Identity as the Great Equalizer
Part of the reason that makes voting so challenging in utility token protocols is the fact that they operate under pseudonymity or anonymity conditions. In the absence of true identity representations for the token holders, consensus protocols need to rely on complex mathematical puzzles to ensure the integrity of transactions. Security tokens don’t have that issue as identity is a key requirement of the issuance of securities in the form of regulations such as know-your customer(KYC).
The fact that the identities of security token holders are known at any given time, opens the door to the implementation of simpler voting protocols such as proof-of-authority(PoA). However, many of the ideas behind PoA rely on the parties in a voting decision acting honestly which we know is not always the case when comes to security tokens.
Quadratic Voting
The idea behind quadratic voting(QV) are as simple as elegant. Conceptually, QV looks to quantify the intensity or preference of each vote instead of just focusing on the qualitative aspect of it(in favor or against). In the QV scenarios, each voter is issued a group of “voice credits” that they can use to influence the outcome of a voting decision. When confronted with a voting decision, each person can buy votes for or against a proposal by paying into a fund the square of the number of votes that he or she buys. This model assigns a non-linear correlation between the cost of voting and the outcome of a decision. For instance, let’s take a QV scenario in which the voters are issued a 10 voting credits each. The cost of the first vote will be 1 credit, two votes will cost 4, three votes will cost 9 credits and the cost will continue to increase quadratically correlated to the number of votes. If a voter is interested on buying more than 3 votes, then they need to purchase additional credits. The voting is “quadratic” because the total amount you pay for n votes goes up proportionately to n².
The beauty of quadratic voting is that the cost of buying votes increases very rapidly which prevents all sorts of malicious behaviors. Additionally, quadratic voting operates on environments in which the identities of voters is well-established which imposes all sorts of social reputation constraints.
Quadratic Voting and Security Tokens
Quadratic voting(QV) provides many interesting ideas that are relevant to security token protocols. At a basic level, we can envision a model in which security token holders are issued an additional utility token that can be used for voting purposes. Issuers will submit voting proposals to the network and token holders with voting rights can use their utility tokens to express their intent. To buy a specific number of votes, the security token holder will have to spend the square of that number in utility tokens.
There are many different models that can be used to implement QV in security token architectures. The dual security/token structure is one that I like as it maintain the governance aspect of a security token network isolated from the security tokens themselves.
Avoiding Collusion in Quadratic Voting
Quadratic voting(QV) certainly prevents some of the traditional attacks on utility token models such as double voting or the same person using multiple anonymous addresses. However, QV is still susceptible to other forms of attacks such as collusion in which multiple voters conspire to achieve a specific outcome. For instance, holders of a real estate security token can collude to receive a bigger dividend payment any given quarter.
There are two ideas I think can be effective to deal with collusion in QV scenarios. Borrowing some ideas from consensus protocols such as DFINITY, we can implement a model in which the network nominates a random number of voters for each voting proposal. In that scenario, the colluding parties can’t be sure if they will be allowed to vote on a specific decision.
An even simpler solution to the topic of collusion in QV was proposed by Glen Weyl in a talk a couple of years ago during a talk at the Becker Friedman Institute at the University of Chicago. Weyl believes that there are two factors that will prevent collusion in a QV scenario:
a) The quadratic nature of the voting will make any collusive behavior immediately obvious.
b) Its likely that the number of voters colluding will be significantly smaller than the total number of the population.
With those two factors in mind, Weyl believes that voters can spot a collusive behavior and simply vote for the opposite outcome. This approach banks of behavioral economics instead of voting algorithms. Certainly clever.
QV is one of the simplest and most interesting algorithms I’ve seen to enable complex voting rules in blockchain scenarios. While QV might have some applications in utility tokens, I believe it has an enourmous potential in the security token space. Hopefully we will see some of the QV ideas incorporated into the next generation of security token platforms.