Learnings from Optimism’s RetroPGF round #1 and proposals for round #2
In July 2021 the Optimism team and Vitalik Buterin started to popularize Retroactive Public Goods Funding in multiple articles and talks.
In January 2022 Optimism ran its first experiment on Retroactive Public Goods Funding distributing $1 Million of rewards among the Optimistic Ethereum ecosystem. In this article, I will recap Optimism’s RetroPGF Experiment #1 and propose improvements for a possible experiment #2.
Recap on Retroactive Public Goods Funding and Oracle design
Retro PGF is primarily concerned with creating efficient markets for public goods by retroactively allocating rewards, following a simple intuition that “it’s easier to agree on what was useful than what will be useful”.
For the desired market dynamics to take effect, individuals, projects, and companies that build public goods, and investors allocating early stage funding towards public goods, have to trust in the legitimacy, credibility, and reliability of a results oracle that allocates the rewards.
The promise is that the more credible neutral the results oracle is and the more rewards are allocated, the stronger market dynamics for public goods will become.
RetroPGF Experiment #1 Overview
In the first experiment on Retroactive PGF Optimism PBC implemented a simple results oracle design:
Phase 1 Nominations (3 weeks)
Anybody in the community can propose projects to be nominated. Badge holders* collectively filter the proposed nominee list and reject or accept nominations.
Phase 2 Voting (1 week)
Badge holders allocate their votes across nominated projects
Phase 3 Payout
Projects receive rewards proportional to how many quadratic votes they received
There were 24 badge holders selected by Optimism PBC, 18 Badge holders were part of the wider Ethereum ecosystem and 8 were part of the Optimism collective.
Badge holders were provided with a badge holder manual that outlined guidelines and rules for nominating and allocating votes to nominees.
In total $1 Million worth of ETH was distributed to public goods.
Learnings from RetroPGF Experiment #1
Below find a brief summary of the shortfalls and learnings from round 1. This is a mixture of my own impressions and notes from the Retrospective meeting of Badge holders and Vitalik
The Fairness ratio and what constitutes a public good
“Public goods” has become a common buzzword in the Ethereum community, and while more conversations and excitement happened around the topic, the community started to form its own definition and understanding of public goods, disconnected from the classic definition found in economics. Rather than defining public goods by the characteristics of the good itself, the community leans toward defining public goods as the outcome of the provision of the good as“everything that benefits the public good”.
Many of the projects that contribute to the public good are not providing public goods in the strict economic sense. Etherscan is a good example as it’s a closed source, proprietary software — definitely not a public good — but it’s accessible by anyone, free of charge, and contributes to the public good of the Ethereum ecosystem.
The Fairness Ratio
The Fairness ratio, introduced by Will Hennessy and Karl Floersch, provides a mental model for thinking about allocating rewards. It captures the idea that individuals, projects, and companies should equally create as much value as they capture.
Value Created / Value Captured = 1
Applying this to the Etherscan example, Etherscan extracts value through advertisements and different services, but in the eyes of the badge holders this extracted value was way less than the value Etherscan created. To correct this imbalance, badge holders allocated more than $24k towards Etherscan.
Conflicts of interest
As the Ethereum community is still tight-knit and closely connected, many of the 24 badge holders we’re connected or associated with nominees. While there were more extreme forms of conflict of interest, like a single badge holder voting for their own project, many badge holders faced more soft forms of conflict of interest, such as voting for projects they are friendly with or worked with in the past.
Badge holder voting strategy
In experiment #1 badge holders naturally tended to vote in their area of expertise, resulting in most votes being allocated to technology projects.
This begs in interesting question of what the strategy of a badge holder is when allocating votes:
Are badge holders supposed to allocate votes in their area of expertise or are they supposed to survey the whole ecosystem and allocate votes to most effectively benefit the ecosystem over all?
Nominee curation
In experiment #1 nomination proposals were collected via a google form. Badge holders then collectively curated the proposed nominees to end up with a set of high-quality nominations. The information relating to proposed nominees was quite slim with many failing to describe how the nominated project has benefited the public good.
In total, 98 proposals were submitted of which 78 were accepted as nominees and 20 were rejected.
Missing transparency
In researching I discovered that a lot of information relating to experiment #1 was missing from the public realm, this includes the full allocation of the $1 Million rewards, the badge holder manual, and information on the nominee curation. Hopefully, the Optimism team will make these available shortly.
Proposed improvements for RetroPGF round #2
Below a few proposed improvements to address the problems listed in the Shortfalls and learnings
Badge holder voting and expertise
Understanding the ecosystem as a whole and what projects have contributed to the public good could easily become a full-time job considering the ecosystem is ever expanding with new individuals, projects, and companies contributing.
Rather than relying on a small number of badge holders to build ecosystem expertise, I propose a system of division of knowledge, in which individuals with strong expertise in different areas are elected badge holders.
For round #2 it would be beneficial to curate a set of badge holders that are not only core contributors to the ecosystem but are also representative of ecosystem contributions overall.
This model could allow badge holders to leverage the knowledge in their area of expertise and allocate their votes with the highest degree of confidence that funds are allocated to the right projects. Adding to the legitimacy of the results oracle.
Note: There’s an interesting poll going on how to expand the badge holder set.
Improving the nomination process with optimistic curation
In Retro PGF we’re not only trying to agree on what was useful, but we’re also trying to agree on how useful it was.
The latter poses a much harder challenge than the former.
We can imagine a results oracle that breaks down the process into 2 separate phases — first assessing what projects have contributed to the public good and second quantifying the value created by the projects.
In phase 1 there is no need to quantify how much value a particular project created, the only restriction is that the value created must be above a certain threshold.
In phase 2 we quantify the value projects have created for the public good.
Designing this process to be credible neutral and decentralized is very hard to achieve, and a deep rabbit hole to explore.
The permissionless nature of proposing new nominees for a Retro PGF round allows the addition of nominees that badge holders themselves would not have come across, but it creates a curation problem of maintaining a high-quality set of nominees.
The current approach to curation is resource intensive by requiring badge holders to make consensus decisions on which proposals are accepted and which are rejected.
To fix this we could leverage existing designs from (token) curated registries, below a proposal for a simple process:
- Each nominee proposal requires a stake of x OP tokens
- During a challenge period, anybody can challenge a proposed nominee by putting up a counter stake of x OP tokens and describing how the proposal violates the Retro PFG constitution
- All the badge holders, or a random sample, are then asked to vote on accepting or rejecting the proposed nominee
- If there’s no challenge the nominee is accepted after the challenge period ends and the proposer receives back their stake + reward.
If there is a challenge and the badge holders accept the nominee, the challenger loses their stake to the proposer.
If there is a challenge and the badge holders reject the nominee, the proposer loses their stake in the challenger.
This reduces the overhead of badge holders, only requiring them to vote on disputed proposals and to voluntarily curate proposed nominees in their area of expertise rather than curating the whole list of nominees.
Optimistic curation ensures quality by requiring a small stake to nominate a project and is more community-inclusive by allowing anyone to challange a nominee.
The by-product of badge holders voting to accept or reject a proposed nominee will be a common law-like history of cases, further helping to drive understanding of what criteria and desired qualities of a nomination proposal are.
Conflicts of interest
There’s no straightforward solution to this problem. Rather than expecting to resolve conflicts of interest altogether, a good approach could be to establish strong guidelines and increase transparency.
- Add a rule to the badge holder manual that voting for one's own projects or voting for nominations that directly benefit oneself is forbidden
- Add a rule to the badge holder manual that possible conflicts of interest by voting for nominees are to be made transparent by badge holders
This approach is common in the executive branch of (democratic) government, ranging from transparency to excusing oneself from decisions if they could introduce conflicts of interest.
Below you can find a simple diagram outlining the process for a proposed Retro PGF experiment #2 incorporating the proposed improvements above.
✨🔴 Stay Optimistic 🔴✨
