Against Inflation-Funded Voter Rewards

[Voter rewards] could, except by chance, only diminish the expected quality of the block producers running the network.

This article also posted on Decentium, the decentralized publishing and tipping platform.

Introduction

A common criticism of EOS is that it is centralized, as there are only 21 block-producing entities. This is true, but with a different perspective on decentralization, the critique is unconvincing. We decentralize not for the sake of decentralization, but for the benefits it affords us in censorship resistance, security, ownership, incentive alignment, and privacy. If we find a solution with these qualities, let us betray our ideals and embrace progress. 21 block producers is more decentralized in practice than most other blockchains, and those with greater decentralization (to my knowledge) sacrifice the network speed and incentive alignment needed for thriving innovation and adoption.

Similarly, let us consider the effects of voter rewards on the decentralization of our voter base. Prudence demands an extra measure of skepticism in the face of a proposal so in line with our natural desire to accumulate wealth. I’ve found not only that the case for voter rewards has not met its burden of proof, but in fact there is good reason to suspect voter rewards could threaten the long term success of EOS.

The Incentive to Vote

EOS New York is one of the most important entities in this ecosystem. They work non-stop on the most pressing issues faced by all EOSIO based blockchains, and they deserve a vote from every single account on every EOSIO network in which they participate. I have to disagree with their recent assessment, however, that “the EOS incentive model omits… work critical to EOS, voting” ([1][2]).

Presumably, if you’re reading this, you’re already voting. So what is your incentive? Why do we vote? We vote because this enterprise holds the promise to align incentives within and across all industries, globally, forever. Because the value of some portion of your wealth is tied directly to the health of the network. And because EOSIO technology could be a key component to unlocking equality, privacy, and liberty, for all mankind.

These are the incentives for people to participate in EOS governance. One could argue they are insufficient, but they already exist.

The Value of a Vote

Voting will cost you. By design, your loss of liquidity from staking is the cost you incur when you choose to vote. That cost is the signal to the rest of us that you care enough about the governance and well-being of the network to assume some risk, and many would argue that this assumption of risk is the source of the vote’s legitimacy. In fact, this is the premise of stake-time weighted voting, that those with the most to lose are also most incentivized to act in the best interest of the network, and therefore deserve the most influence ([3][4]). If you agree with this premise, it follows that unrewarded votes are perceived by the voters as more valuable than paid votes.

Without exception, every new voter solicited by a voter-rewards scheme will be a person for whom their influence in the long-term sustainability of EOS was not worth the temporary loss of their asset liquidity. If new voters arrive for whom this is not true, then they would have voted with or without the voter reward.

Do you want people with such a low valuation of the EOS network diluting the weight of your informed vote?

Vote Dilution

Those for whom the cost of voting exceeds the perceived value of their vote will not vote. Surely, rewarding voters would to some extent offset the opportunity cost and increase voter participation, theoretically providing us with a more decentralized voter base. It is worth remembering, however, that any time a person transitions from non-voting state to voting, the new votes dilute everyone else’s vote.

This is by design, and it is a good thing, so long as we can count on votes as representations of the people’s sentiment on the best block producer candidates. However, the ultimate effect of lowering the cost of a vote (or indeed paying voters) is to decrease the average number of “informed” votes in the network by adding a swath of voters who care less about the health and governance of the network than short term profit potential. In other words, the votes which matter for long-term sustainability are diluted. We may gain decentralization, but we do not decentralize for the sake of decentralization. Less informed votes could, except by chance, only diminish the expected quality of the block producers running the network.

Dan Larimer has recently taken this understanding a step further in suggesting not to decrease the cost to vote, but to amplify it. That we should prefer to increase the opportunity cost of a vote to promote a voter base which is both “decentralized enough” and also made up only of those who really value the long-term success of the network. To achieve this, he proposed rewarding non-voters, to ensure that apathetic profit-seekers are kept out of the governance process ([3]). His proposal can be thought of as simply the network’s way of paying for higher quality votes (with the same mechanism we pay for block producers).

Regardless of whether you agree with the specifics of Dan’s proposal, the game-theory behind it is important. We should prioritize giving influence in governance into the hands of those who can prove, on-chain, that they have the most to lose.

Effect of Low Voter Participation on Network Security

It is precisely because of vote dilution that increasing the number of voters also increases the capital threshold of attack i.e. “the cost of purchasing enough tokens to install [a] ‘malicious’ block producer” ([1]). So isn’t low voter-turnout a security concern for the network? Actually, an individual block producer has very limited ability to hinder the network. A block producer cannot change rules, double spend, reverse or censor transactions. The negative effect of a “malicious” block producer is limited to taking block producer pay from the network (“leeching”) without adding value in any other way besides block production.

Voter rewards is not an effective counter-measure to leeches, and even if it were it would not be worth the cost of the vote dilution, especially when there are other proposals to avoid leeches without these drawbacks ([5]).

Conclusion

We all want honest Block Producers who accurately represent the desire of the people, and who work for the long-term sustainability of the EOS public blockchain. I have attempted to show that voter incentives do not correct any misalignment of incentives, do not play a major role in enhancing network security, and they increase decentralization only at the cost of diminishing the expected quality of the elected block producers. It is the non-whale, non-exchange, high quality block producers who should be fighting against voter rewards the hardest.

If you appreciate posts like this and/or want this conversation to continue, please consider sharing this post.

James Mart | <Twitter>

Sources

[1] EOS New York — The Case for Voter Rewards — Part 1
https://medium.com/eos-new-york/the-missing-piece-to-the-eos-incentive-model-bd39977d243f

[2] EOS New York — The Case for Voter Rewards — Part 2
https://medium.com/eos-new-york/the-missing-piece-to-the-eos-incentive-model-part-2-727aed48f817

[3] Dan Larimer, Decentralizing in Spite of Pareto Principle:
https://medium.com/@bytemaster/decentralizing-in-spite-of-pareto-principle-eda86bb8228b

[4] Vericoin — Proof of Stake-Time
http://www.vericoin.info/downloads/VeriCoinPoSTWhitePaper10May2015.pdf

[5] Proposal to avoid BP Leeches
Introduce a cost to register as a BP (“bond”), which is returned if/when the BP is unregistered. Allow for a 15/21 BP vote to unregister other block producer candidates, which burns their bond. BPs are then responsible to identify and remove leeches. This would be effective at mitigating BP leeches, and doesn’t put the network at risk.