[Phase 1] HIP-57: Harberger Taxes Seat Bicameral Governance

HIP: 57
title: Harberger Taxes Seat Bicameral Governance
author: @greenlucid
status: Phase 1
created: 2022-07-26
conflicts with: None
languages: EN

Simple Summary

Change governance to a bicameral system. Each chamber has 50% voting power. The Humanity Chamber uses 1-person-1-vote, the Stake Chamber uses 1-seat-1-vote, where each seat is an NFT allocated by means of Harberger Taxes.


Harberger Taxes are an experimental pricing mechanism that involves price discovery and is ideal for limited, public resources, such as real estate. Due to its public goods funding applications, it can serve as a solution to both bring skin-in-the-game governance, and income for the DAO simultaneously. This proposal introduces a bicameral governance system that features a Humanity Chamber, with 1p1v, and a Stake Chamber, where each seat is allocated through Harberger Taxes.


Pure 1p1v systems require no skin in the game. The registrants of a sybil resistant registry are not necesarily the consumers of the registry. One HIP, HIP-54 “Democratic Heroes” was an omen of the dangers that a DAO with voters with no skin in the game: bribing schemes to steal the treasury.

There are three ways to obtain new users:

  • The registry is used by other applications, so it has use.
  • Users are economically incentivized to register (UBI, airdrops…).
  • People are persuaded to register (family, friends, crypto-influencers and politics)

As the two main reasons to enter are not strong (UBI presents a, still decaying, 70$ yearly income, airdrops can’t realistically occur without sybil resistance and the registry doesn’t have enough attention), then the main way into the registry is getting told about it by its users. This encourages influence and popularity as the way to do governance. Popularity does not relate to good decisions. Technical issues cannot be understood by the majority of the voters, yet their criteria is what ultimately matters. Even if they trust technical people and delegate to them in those matters, then governance in technical matters is built on top of trust. This goes against the spirit of DAOs: for decentralization to be possible, trust should be minimized.

However, 1p1v does have benefits. First, it makes for an interesting political experiment to have 1p1v direct voting. It gives immediate, not negotiable utility to being a registered user. It encourages registered humans to participate and get informed about the protocol, knowing they don’t need to be represented to play a role. Anyone can create HIPs, a mechanic that hasn’t been abused so far. Lastly, it sets the grounds for a Network State.

Removing 1p1v entirely would remove these features and add barriers to participation. But, keeping the governance system purely 1p1v will result in the damaging results mentioned before. The best compromise is to reduce the voting power of 1p1v by half by creating a bichameral governance system, with a second chamber that features skin in the game.

The second issue is that the PoH DAO has a funding problem. All of the funds in its treasury were awarded by the Kleros Cooperative, and with current ETH prices, they barely cover the salary of a Product Manager and a Developer for one year. Governing is a privilege, not a burden, so having increased voting power should have a cost that ideally funds the DAO.

Regular skin-in-the-game governance solution comes in the form of token based governance. This tends to result in whales that, once they take control of the protocol, low liquidity makes for a very stale state of affairs in governance.

However, with the application of the recently created ERC-5320 Harberger Tax NFT, you could have a different type of skin-in-the-game governance. Make each seat in the chamber be an NFT subject to Harberger Taxes. For those unfamiliar with the concept, it’s a pricing mechanism that creates efficient allocation of limited resources. If Alice appraises a seat at 10, Bob can force her to sell it to him at 10. Also, Alice has to pay taxes on the seat, and they are dependant on the appraisal Alice makes. So she cannot appraise it too low or Bob will buy it from her, and she can’t appraise it too high or she will have to pay too many taxes.

The reason this mechanism is interesting for governance experiments is, these taxes will be sent to the PoH treasury, and then they can be repurposed into funding the different needs of the DAO, mainly, offering good rewards for contributors, and supporting the currently main use case of the DAO, which is, the UBI token. Unlike regular token based governance, here, the “seat whales” will be providing a healthy source of income to the DAO, and thus, will have vested interests in making the DAO succeed while benefiting the users of the registry.

So, taking into account this desire to give skin in the game to governance, and being this potentially huge source of steady income, it can be reasonably concluded that this is a good idea.


Bichameral Governance

Deploy a new strategy that follows the following bichameral scheme:

There are two chambers, the Humanity Chamber and the Stake Chamber.

The total voting power of each Chamber is 50% of the total. Within each chamber, votes can dissent, and their different weights are translated onto the full result without removing minority options. To complete this definition, here are a few examples:

Humanity Chamber: 1 for A. Resolves as (A=50%, B=0%).
Stake Chamber: 1 for A, 3 for B. Resolves as (A=12.5%, B=37.5%).
Result: (A=62.5%, B=37.5%), A wins.

Humanity Chamber: 99 for A, 1 for B. Resolves as (A=49.5%, B=0.5%).
Stake Chamber: 1 for B. Resolves as (A=0%, B=50%).
Result: (A=49.5%, B=50.5%), B wins.

Humanity Chamber: 20 for A, 80 for B. Resolves as (A=10%, B=40%).
Stake Chamber: 7 for A, 3 for B. Resolves as (A=35%, B=15%).
Result: (A=45%, B=55%), B wins.

Note, for both chambers, the inactive population does not matter.
This is a TypeScript piece of code to convey the definition more clearly, for n Chambers:

 * @param chamberVotes: Matrix where each row is a chamber,
 * and each chamber has the count of votes towards each option.
 * @returns: weight of each option after the voting.
const getWeights = (chamberVotes: number[][]): number[] => {
  const chamberSums = chamberVotes
    .map(row => row.reduce((sum, votes) => sum + votes, 0))
  let optionWeights: number[] = []
  const rows = chamberVotes.length
  const optionCount = chamberVotes[0].length
  for (let i = 0; i < optionCount; i++) {
    let weightSum = 0
    for (let j = 0; j < rows; j++) {
      weightSum += (1 / rows) * chamberVotes[j][i] / chamberSums[j]
  return optionWeights

Stake Chamber Contract

The Stake Chamber contract must be implemented in a secure way. It extends ERC-5320. The tax rate is a flat ratio of the valuation paid per year. Decreasing the valuation must be only be realized after a commit period, decreasing the valuation must not be allowed while buying. Excluding collect, interactions must revert if the funds available to pay for the taxes on the seat are not enough to go over the cooldown period. If there are not enough taxes to support the seat, the seat is then revoked. Specifically, these interactions are: buy, defund, changeValuation.

The contract must use a neutral and valuable ERC-20 currency, to buy the seats and pay taxes with.

More seats can be minted by the PoH Governor. Also, the PoH Governor can change the tokenURI of each seat, to customize them and give them an artsy, vanity aspect.

These are the parameters proposed for the contract:

currency: WETH
yearly tax rate: 20%
commit period: 3 months
number of seats: 10

This project is large scale:

  • Ideally, would have the ERC-5320 to be set and accepted first.
  • Making a secure, well tested Stake Chamber contract.
  • Investigate Snapshot. Is it possible to make a strategy that implements the weights as described in the specification? Or would two different strategies be required?
  • Adding future cosmetical value to the seats.
  • Make a frontend to interact with it.

As such, it would require, at the very least, a dedicated developer to put in 1 or 2 months of work. So, this should ideally be funded through a grant. This can be discussed throrough this HIP.


Why give each chamber equal voting weight?


Doesn’t this create plutocratic voting?

Every governance system is “plutocratic”. Pure democracy is already plutocratic in it’s own right, as the biggest capital still holds great control over the organization. In this case, the capital is “social capital”, or the ability to organize and rally crowds around an option.

This governance system mixes both economical, skin-in-the-game voting, and social voting at the same time. Both ways of governance hold equal weight. Also, this Stake Chamber model creates a rent for the whole DAO to utilize, so, even if a whale sector took ahold of the protocol, they would also be forced to steadily fund the DAO.

Why use those parameters for the Stake Chamber contract?

Having the valuation be paid as taxes in 5 years, and committing 3 months of taxes (5%) at a minimum seem reasonable defaults. They provide good rate of income to the DAO, while not being extractive enough to be attractive to investors.

Why give cosmetics to the seat NFTs?

PoH is one of the largest active DAOs there are, and the NFTs are already valuable on their own right. Giving them cosmetic and vanity value can attract newcomers, build a culture around each seat, and give them visually pleasing content for ERC-721 compliant explorers. Example: label them as “Iron Throne”, “Silver Throne”, etc.

Those cosmetics don’t necessarily need to be launched along this HIP, as this is already a large scale project. So, giving the PoH Governor a way to edit the tokenURI in the future is desirable.

Isn’t this too complicated development wise, can the DAO spare the resources needed to fund this?

That’s hard to tell, we can discuss it in this thread. I believe this would take 2-8 weeks of work for a single developer. If there’s interest on this idea, the DAO should look for a developer to implement it.

As the discussion may reject the idea, or change the requirements, it’s not easy to make an estimate of the cost of the grant, but, assuming the cost is equal to 2 months of dev time (worst case), ~15k USD. If this proposal arrives at Phase 2, a grantee should be proposed.


Assuming there was interest in this, if the scope is clear and can be estimated to take around 2-8 weeks to completion, I can propose myself to build this, since I wrote the HIP, I’ve helped build a full stack utilizing this concept and interface in a Hackathon, and I have a good idea on how to build it. However, I’m hardly available:

  • If ~2 weeks, I can take vacation and build it full time from start to finish. With the proposed scope, and discarding the frontend, there might be enough time.
  • If it needs more time, I could consider building it part time as a side project, but then it would take much longer (2.5x longer?), and for that time span, PoHv2 may be in sight, making this a risky, unstable venture.
  • If the Kleros Cooperative showed enough interest in this project, the grant could be forwarded to it and having me build it as a contractor full time.

I mainly wanted to present the idea so that the community can keep it mind, consider if it’s worthwhile, and if so, take it into account for future assumptions and projects.

To be fair, I think this project is too large scope to be feasible in the current state of affairs, but I think it’s a cool concept. Could give a lot of visibility to the DAO, given that Harberger Taxes are a well known public goods funding concept, apart from securing a source of income.


Hi Green. Thank you. it is a really interesting proposal.

It really becomes a problem for me when money is what gets you passing hips, deciding the future. inevitably brings lobbyists.

maybe if it were 49-51%?

still a fantastic topic to discuss.

inevitably brings lobbyists

This might be risky. However, if the lobbyists are paying top dollar to get their seats, then at least they are providing income to the DAO.

maybe if it were 49-51%?

Say that all humans in PoH are radically against a proposal (99% against) and an attacker holds all the seats in the Stake Chamber. Then, all it takes is for all these humans to buy one (or some) of the seats and give it a high valuation to prevent the attacker from acquiring it. They can pool the funds together, or ask for assistance for other DAOs or projects.

To get a 100% attack on the Stake Chamber you need to support all the seats at the same time.
If the attacker rebuys it from the defenders, then that’s funding the defenders can use to buy another seat. If the attacker is then forced to value all of the seats very highly, then they’re losing a lot (that, is sent to the DAO). They have to commit to pay taxes for a minimum amount, which makes sustaining this attack very taxing.


My Man, losing this is not worth all the lithium Elon saves in his freezer

losing this is not worth it

Well, there are lobbyists already. As I exposed in the Rationale section, popularity is also plutocratic and lobbyable. Highly homogenous and motivated sectors can decide the outcome, in that sense, the people rallying them are whales as well.
You would not be losing the democratic power either, since popularity still has power over half of the governance.

The reason this proposal is desirable is, the DAO rewards with governance those who are willing to fund the DAO. Whereas, right now the DAO is giving governance power to everyone, so the vast majority of the DAO population hasn’t contributed funds to the DAO in any way because there’s no incentive to do so.

1 Like

Harberger taxes is an interesting concept and with ERC-5320 I am sure it will have a role to play in DAO governance going forward.

For the PoH situation specifically, I have two concerns about its suitability to mitigate potential governance attacks:

  1. While it definitely introduces ‘cost-in-the-game’, the cost for attackers is already ‘sunk’, and unlike in PoS situations, further attacks on the DAO do not incur further costs to the attacker than is already invested at the start of the process.
  2. As mentioned by @00brujula, do people who are able to pay for the ownership/taxes of theses seats necessarily people who are interested in the best outcomes for the DAO/PoH? Partners who have a vested interest in, say, the preservation of Sybil resistance in PoH might not necessarily be willing to put down thousands or tens of thousands of coin to keep/gain a seat when push comes to shove.

Yesterday, there was a discussion in the largest, unofficial English PoH Telegram chat. Making the Seat Chamber have 50% voting power is a vulnerability, this voting power can be used to administer the taxes you’re paying. So, this just becomes a bank attack with current parameters.

Changing this bicameral system to 80% Humanity Chamber, 20% Seat Chamber eliminates this threat, and arguably makes the system more solid:

If someone attempts a bank attack and gets all seats, they’ll need to get 38% of the democratic vote. (Compared to 50%). Failing this attack results in net income for the DAO.

The Humanity Chamber still can pass anything with >62.5% vote, no matter what the Seat Chamber votes.
If someone attempts a bribing attack AND they don’t have access to the seats, (e.g. distribute the treasury between all humans), they will need >62.5% of the democratic court to fulfill the attack.

So, adding this new game to governance makes it a more secure system, because now an attacker has to hit the weakness of each Chamber. These economical weaknesses are different:

  • Humanity Chamber is susceptible to bribing attacks, biased onboarding campaigns, or propaganda.
    • Failing a bribing attack is not necessarily good. Some types, like p + e bribing are good if the proportion that accepts the bribe is a minority which is close but under the winning ratio (50%). If everyone cooperates to achieve this, the bribed proportion can split the delivered bribe with everyone else, which is good.
    • Failing a biased onboarding campaign increases the population of the registry, which is good.
    • Failing at propaganda is not necessarily good.
  • Seat Chamber can just be bought instantly. Failing the attack means the attacker commits to pay taxes if no one else buys it back, which is good. *

* To make this loss more explicit, a % could be paid instantly to the DAO (as, "transfer taxes"). Previous system is still good for the DAO, but an speculative attacker could make money on it, even if the attack fails, by waiting until the seat is bought. So, if you were going to speculate on the value of the seat anyway, “you might as well try to attack the DAO”. Making the buyer pay a fixed ratio of the seat on buy (like 5%, equivalent to 3 months of taxes) makes it harder for the attacker to quickly cut a profit on reselling the seat after the attack. The issue with this buyer-tax is that it punishes attackers and regular owners alike.

In the future, breaking governance into more Chambers like this (as this becomes a larger DAO, and more attacks are focused on it) can make this easier.

For the future, it must be taken into account that having >50% be instantly acquirable with funds is a vulnerability, so Humanity Chamber shouldn’t be broken into more Chambers with the same economical security assumptions, as, with this proposal, 20% of governance can currently be instantly bought.

Breaking the chambers onto chambers with sybil resistance (for example, quadratic voting) doesn’t break these assumptions.

Please let me know what you think. We still haven’t found a recipient for the grant, hopefully we need at least one before this passes to Phase 2. If more than one grant recipient proposes themself, we can vote first.

1 Like