FIP-ZZ: Mechanism for Fei (+Rari) infrastructure funding

I’ve been thinking about our discussion on Fei contribution funding. The key clash that has emerged here is the discrepancy in perceived subjective value of respective contributions, approaches to measuring it and hence mapping this value to compensation. One can foresee that without a systematic approach this problem will only exacerbate with the protocol growing more complex, especially in view of the coming Fei-Rari merger with consequent increasing complexity of the protocol’s infrastructure and range of contributing efforts.

What Fei is facing here is the problem of optimal allocation of public goods funding. Fei protocol, as any decentralized fat protocol, is open infrastructure.

Public goods, like national defence, urban or DeFi infrastructure, are non-excludable (if you use Fei, you can’t refrain from benefiting from efficient discord modding, for example, it’s part of what makes Fei a DeFi protocol) and non-rivalrous (your benefits do not reduce that of other Fei users).

The problem of efficient and fair distribution of funds for public goods has been a subject of extensive research.

Buterin et al. have suggested a Capital-constrained Liberal Radicalism (CLR) mechanism, aka quadratic funding, which offers a (nearly) optimal distribution of a designated budget (i.e. limited funds) against a given list of public goods.

Basically the idea is that voters deciding funds allocation can reflect in their decision not only their preference, but also the strength of their preference by buying influence at increasing marginal cost, thus a broader community can account for interests marginal on grand scale, but important in view of part of the community, while also preserving the overall democratic ethos and not sliding towards oligarchy, where a few powerful make decisions for everybody.

And then there’s a matching pool (a budget) provided either by the DAO, like Fei DAO, or some benevolent outsiders, which matches voting results with funds.

Here’s a primer on quadratic payments from V. Buterin.

It’s also been mathematically proven by Weyl et al. that under standard assumptions, a fixed set of communities and public goods as in our case, in large populations quadratic voting mechanism, of which CLR is an extension, leads to approximately optimal decisions on public goods.

Finally, a number of promising real world experiments around CLR has been carried out, of particular interest for us being Gitcoin protocol for Ethereum infrastructure funding, with purposes similar to that of Gitcoin, but implementing anti-collusion infrastructure and Optimism, which augmented CLR with badge holder governance.

So drawing from all the above here’s a scheme that might suit both Fei Labs and Fei DAO community.

A budget is allocated by a DAO decision for a period x (e.g. 3 months). By the same DAO decision a group of badge holders with a mandate to allocate these funds using quadratic voting mechanism is designated.

Badge holder governance is a middle ground between full-scale DAO, impractical if not impossible for these purposes, and a tight group like TF leads, grant committee or FL founders. Actually, the more diverse it is, the broader set of interests it represents the better. As a pilot I suggest to include into the badge holder group everybody sufficiently active within Fei, like 9TF + all current FL people: Joey, Seb, Bri, storm, klob, Elliot, Eswak + (maybe) representatives of some investors like a16z, Robot, Framework. etc. In addition a number of outside independent badge holders from the ranks of respectable DeFi+Ethereum+crypto community members could be invited to play a balancing role akin to independent board members. Badge holders could be compensated for their efforts or not, up to further discussions. This presents a way to bring alignment between interests of FL and that of the Fei decentralised community. With time the badge holder committee could and should be enlarged, for example, with discord users reaching a certain karma level as calculated by a discord bot, which in turn will bring about gradual (partial) transition of power from FL to the Fei DAO community.

Within a designated period anyone can nominate a funds recipient, be it for continuous work like modding, marketing or a fixed-goal project, with all reasoning possible.

All pre-voting discussions are held in public on discord/discourse.

Decision about funds allocation is made with the quadratic voting mechanism. All badge holders are given an equal amount of credits, e.g. 100. Each badge holder can allocate their credits reflecting their respective individual assessment of nominated project’s value and strength of this opinion (that is for a given voter marginal cost of the next vote given for the same nominee is bigger, reflecting the strength of subjective value preference: 1st vote costs 1 credit, 2nd vote for the same nominee = 2 credits (total 1+2=3), 3rd vote for the same nominee = 3 credits (total 1+2+3=6) etc.) If the number \ n of votes goes to the infinity, then the limit of the sum is \frac{n^2}{2}, hence the term “quadratic”.

The proportion of the budget received by each nominee is calculated as follows:


The result reflects the distribution not only of perceived value, but value weighed by the strength of stakeholder’s preference.

At first the mechanism could be funded from the Fei DAO Treasury. In some future iteration this logic can be encoded into the smart contract, such that a certain % of Fei DAO revenue is directed to this mechanism. Some floor level could also be designated and if Fei revenue % falls below this level, the difference is subsidised from the Treasury.

The beauty of this mechanism is that almost everything here is a variable, hence it can be gradually adapted to fit Fei evolution.

For example, principles of the badge holder committee composition are up to further discussion. At first Fei Labs representatives could have a majority and later on their voting share could go down gradually with new members joining the committee. Although I’m not in favour of this particular move, it’s technically possible. Also voting credits distribution could also be a variable, badge holders could receive credits according to some formula or in equal proportion. Distribution of people with certain expertise could also be subject to tuning: technical, mods, marketing, financial etc. Obviously, these factors will influence the eventual funds distribution.

Another class of variables could deal with the compensation structure. For a certain class of funding recipients an n-component compensation structure could be introduced as mandatory: x% retroactive + y% future-looking, or x% liquid + y% vested + z% locked. Or some floor level of payment/month. These caveats could all be integrated into the badge holder mandate by the DAO decision.

Even the key reward formula can be tweaked, as was done by Optimism, they didn’t raise the sum to the power of 2, although this led to smaller variance in funding distribution as compared to Gitcoin.

Also certain anti-spam firewalls could be introduced for nominees, like pre-approval of at least one badge holder, some (not big) deposit or proof of humanity.

Obviously, both individuals and groups like TF could be among funding seeking nominees.

Multiple variants of vanilla QV and QF aimed at mitigating certain shortcomings (particularly inefficiencies and collusion vulnerabilities) have been suggested and are being actively explored, which should become the subject of further discussion here as well.

Still in case of badge holder governance collusion considerations are slightly mitigated by reputation factors. An example of badge holder code was introduced by Optimism in the course of their experiment.