Algorand: Proposal for the Dutch Auction of Governance Explained in Simple Terms

A Layman’s Guide

Photo by Tim Roosjen on Unsplash

Introduction

A Proposal for Decentralizing Algorand Governance is a paper written by Algorand’s founder Silvio Micali, covering how he sees control over decision making for Algorand somewhat decentralise away from the Algorand Foundation towards a panel of governors.

This article focuses specifically on the Dutch auction proposal made on page 8 of the aforementioned paper and explains in layman's terms how governors are chosen under these proposals. If you are interested in a more generalised summary of the entire paper articles by other authors on medium are here and here.

What is the Point of an Auction?

The point of each auction is to find the people who are most keen to become governors and give them decision making power. The auction tries to capture those who are willing to earn the least interest from their govenorship on the assumption that this is because they care the most about the future of the project.

What is Being Auctioned?

Govenors are users (or more specifically wallets) that have voting rights, they can vote in proportion to the amount of tokens they commit to lock up for one year. The lockup period is to incentivise good decision making based on the idea that bad governance will lead to depreciation in the value of their tokens (and a year is a long time to hold onto a losing trade).

In traditional finance when people lock money away for a fixed period of time (for example in a fixed-term savings account) they expect an interest rate to compensate them for losing the right of withdrawal. This situation with locking away tokens is no different, therefore people are auctioning off their right to use their tokens in return for voting rights over the lockup period and reward tokens at the end of the year.

Where do the Reward Tokens Come From?

Before starting an auction the Foundation will specify a fixed total amount of tokens that will be used as reward tokens for the purpose of the auction.

Reward tokens come from the Algorand Foundation’s Treasury and are the payment made to governors for their stewardship and for locking up their existing tokens.

How Does the Dutch Auction Work?

There are multiple stages to an auction, the first stage will have the lowest interest rate and then with each subsequent stage the interest rate will increase by a fixed % amount.

During each stage people have the opportunity to pledge tokens to be committed to the governing process. Because people are willing to invest more when returns are higher the theory is that more tokens will be pledged with each subsequent stage.

The auction ends when either:

1. The interest rate of the current stage multiplied by the number of tokens pledged by people in the auction is equal to or exceeds the number of reward tokens specified at the start of the auction, or;
2. The maximum number of stages is reached and the auction expires without condition 1 (above) being met.

Govenors will then be entitled at the end of the year an amount equal to their committed tokens (the ones they pledged at auction) multiplied by the interest rate specified in the last stage of the auction.

Because it would not be possible for the Foundation to pay out more in reward tokens than the maximum stated prior to the auction, in the case of scenario 1 some tokens pledged in the final round may be rejected as unsuccessful. This is because the interest rate multiplied by the committed tokens must exactly equal the maximum number of reward tokens. For this reason there must therefore be a tie-breaking rule, the article suggests first-come first-serve, with the excess tokens committed most recently being rejected.

How the Staging Works

The interest rate in the first stage of any auction will be equal to the number of reward tokens to be distributed at this auction divided by the total number of tokens existing in circulation at the time of auction. i.e. if there are 5 reward tokens in this auction and 100 tokens existing in total circulation then the starting interest rate would be 5/100 = 0.05 == 5%.

The interest rate increases with each stage by 1 divided by the amount equal to the maximum number of stages specified at the start of the auction i.e. if the maximum number of stages was 25 then the increment would be 1/25 = 0.04 == 4%.

Using the above two examples the interest rate for the first stage would be 5% the second stage 9% the third stage 13% and so on until the final stage at 101% (if the auction was unsuccessful).

Example of an Auction

Given the following variables:

• Total tokens in circulation = 1000.
• Reward tokens in this auction = 70.
• Maximum number of stages = 20.
• The interest rate offered at each stage will therefore increase by 1/[maximum number of stages] = 5%
• There are 4 wallets A, B, C & D, each contains 250 tokens.

Stage 1:

• Interest rate = 7%

1. A pledges all 250 tokens,
2. no more pledges are made in this round.
3. [rate] x [tokens pledged] = 0.07*250 = 17.5

Stage 2:

• Interest rate = 12%
• [rate] x [tokens pledged] = 0.12*250 = 30

1. A cannot pledge any more tokens.
2. B pledges 100 tokens
3. no more pledges made in this round.
4. [rate] x [tokens pledged] = 0.12*350 = 42

Stage 3:

• Interest rate = 17%
• [rate] x [tokens pledged] = 0.17*350 = 59.5

1. A cannot pledge any more tokens.
2. D pledges 100 tokens.
3. B pledges 100 tokens.
4. no more pledges made in this round.
5. [rate] x [tokens pledged] = 0.17*550 = 93.5

93.5 > 70 so at this point the auction ends, and we assume a first come first serve tie-breaking rule is used.

At an interest rate of 17% a maximum of 411 tokens can be committed by users to fulfil the maximum reward token payout of 70: [tokens] =70/0.17.

Govening Tokens Committed by Each User:

• A: 250 (from stage 1)
• B: 100 (from stage 2, the stage 3 pledge was rejected).
• C: 0 (none pledged).
• D: 61 (411–350)

Some Criticisms of the Dutch Auction System for Algorand’s Govenance

I came across a Reddit post which makes some interesting points about the Dutch auction system as proposed by Silvio Micali so I wanted to examine them for you to help you to understand some of the relative merits (or not) of this system.

“A person who is wealthier has more resources to leverage to win the auction… This process used to win is not democratic in any way shape or form…”

Yes. This is absolutely true, but if a large investor wants to be sure of maximising their control then they need to pledge large amounts of their tokens and pledge them early. This will have the effect of reducing the interest rate payed out at auction and therefore the amount they get paid to lock up their tokens for the year.

The more of their tokens an investor has tied up long-term in a project the more likely they are to make governance decisions in the best long-term interests of the project.

Large investors have more resources to research and scrutinise proposals put to them by the Algorand Foundation. This means that they are less likely to just agree with the foundation’s recommendations decentralizing power in a meaningful way.

“By allowing the Foundation to set the parameters centralization occurs.”

Yes. Centralisation of governance but not of consensus, centralisation of consensus is dangerous because it invites fraud. If done correctly some degree of centralisation over governance creates accountability.

Currently the Algorand Foundation make all the governance decisions, this is a proposal to start pushing these decisions out to the community and so it doesn’t make much sense to say that these changes will increase centralisation of governance.

One of the most compelling reasons to invest in Algo right now is because of the team behind Algorand and their performance so far.

“You have to lock tokens for a prescribed time (1 year)… Doesn’t this scheme permit the rich to get richer?”

Yes. But only if they govern wisely, if the people who have committed tokens for the sake of governance goven badly then the coins will deteriorate in value. Govenors risk a capital depreciation exceeding the interest rate specified at auction.

“The simplest solution is to allow each wallet that has completed a KYC to participate in governance… [if] Voting is weighted in any way other than one governor, one vote, I think this is materially stupid.”

No. The know your customer process(KYC) done properly is time consuming and expensive, because there are lots of checks that need to be done to conclusively verify identity of the owner and any other controlling interest over the wallet. KYC done badly is pointless and this approach would invite fraud.

If governance was done on a one-person one-vote basis there would be both practical issues and a different moral issue. The practical concern that a malicious person with 1000 coins could open 1000 wallets with one coin each and defraud a KYC process which was not robust. The moral issue is that this person would then have more control over governance than an honest person with 1000 coins in a single wallet.

Beyond fraud there is a further moral and economics based reason why universal suffrage may not be a good model for governance, imagine 2 investors:

1. Person A: Invests a significant portion of their income every month in Algo.
2. Person B: casually invested in a few tokens.

In order to make a judgement on universal suffrage the community has to decide if it is moral that Person A should have the same voting power as Person B. They also have to consider that it is likely that Person A is much better informed about the project (because they have a much greater financial incentive) than Person B and the project would therefore be harmed by giving Person B an equal vote.

Conclusion

Page 8 of A Proposal for Decentralizing Algorand Governance looks complicated but it is actually super simple…The interest rate paid for governance is the [rate] when:

[rate] x [tokens pledged for governance] = [reward tokens available]

I hope this has answered any questions that you had when puzzling over the Dutch auction section of A Proposal for Decentralizing Algorand Governance. If you liked it then check out my other articles on Nano and Productivity.