[Pre-Proposal] KPI options airdrop

Incentivisation mechanism: synthetic ERC-20 tokens called KPI options are airdropped to a selected list of the most active members of the DeFi community — thought and action leaders. The option’s redemption value is tied to an underlying Key Performance Indicator (KPI) of the protocol. Then at a certain latter date the option pays out a reward if the project’s KPI reaches a predetermined target before a given expiry date.

In short, a hypothetical scenario looks like this:

  1. Some significant amount, e.g. 50M of TRIBE either from the DAO Treasury or newly minted, is allocated as a frozen redemption collateral against the creation of 5M of KPI Options. Some amount of FEI can also be alloted.
  2. An appropriate KPI or a panel of KPIs and an expiry date are designated: e.g. FEI 24h trading volume in 6 months from now.
  3. KPI options are airdropped to selected members of the DeFi community subject to past behaviours or interactions within the DeFi space.
  4. At expiry options are allowed to be redeemed for TRIBE/FEI on a sliding scale dependent on performance of the selected KPI. For example, if 24h trading volume was chosen, and it was below some pre-set lower bound at expiry — e.g. $300M — the options would be redeemable for 1 TRIBE a piece. Conversely, if 24h trading volume was above some upper bound — e.g. $3B — they would redeem at 10 TRIBE. Any value in between would lead to redemption determined via linear interpolation. Alternatively, they’re offered to be (partially) exchanged for the next round of KPI options on preferential terms.
  5. TRIBE tokens not bound for redemption are returned to the Treasury.

Possible criteria for a whitelist of participating addresses:
For a small number (to avoid unnecessary dilution) of leading/promising protocols:

  • Addresses that participated in governance voting on Snapshot/Tally;
  • Top Discord/Discourse contributors;
  • Top Github repo contributors.

Possible participating communities: UMA, Maker, Bankless, Liquity.

Plus top FEI contributors/participants.

Possible FEI KPIs:

  • (rolling mean) 24h trading volume;
  • (rolling mean) Market Cap;
  • (rolling mean Market Cap dominance;
  • (rolling mean) Trading volume/Market Cap;
  • Market Cap rank;
  • (rolling mean) PCV size;
  • (rolling mean) Real PCV size adjusted for market fluctuation;
  • (rolling mean) User circulating FEI;
  • Number of integrations;
  • Particular integration on a certain important platform — realised in the form of indicator function (yes/no). Introduction of tools like rolling mean can serve to incentivise against short-period manipulation around expiry date.

Obviously, whitelist details should be kept secret until after a snapshot is taken to prevent everyone flooding into the path of least resistance in time to qualify.

This approach drives interest of the most active and capable part of the DeFi community in contributing to FEI’s productive organic growth without triggering immediate dump. It also builds up a strong FEI DAO community.

Plus it could also bring nice publicity.

UMA has already tested the waters and built infrastructure for this.

I think this can be realised after DIs are reinstated, but this is a serious endeavour, so I suggest we start the discussion now and if, hopefully, we converge to a decision, I’ll draft an official proposal as well as UMIP for UMA to include TRIBE and FEI to the list of acceptable collaterals, will prepare KPI Options contract and we’ll follow through.

7 Likes

Interesting way to bring more actively contributing people to the protocol :brain:

  • how would the committee establishing the whitelist be selected, would that be the core team / moderators ? should people in the committee be able the whitelist themselves ?
  • should someone whitelisted but not helping in any way be able to exercise their options ? How would that be controlled ?

If it’s a committee, for the purposes of secrecy, I’d first ask who wants to do this job. I’m not sure there will be many volunteers. We can arrange single transferable vote for candidates who registered not later than a month ago (just off the top of my head). Also we can discuss this within the community, then as we converge a Snapshot vote can be held. I don’t think the majority (headed by the team) will vote for a clearly unfair decision.

I’d reformulate the question: should FEI people be included in the first place? I’d say yes, but let’s make it a special case. The whitelist for FEI can be decided by the community discussion and a Snapshot vote. Because whatever committee composition you come up with, in any way you’d include the most proactive members (we all know their names here), there aren’t many of them here anyway and it would be unproductive and unfair to exclude any of them from the whitelist.

That’s a different kind of stimulus. We can also discuss personal options tied to personal results. Here I’m talking about a fine-tuned version of helicopter money, when a relatively broad group of (in this case qualified) people is stimulated in order to spur a rising tide, which would lift all boats. Some bets will work, other won’t, we can’t differentiate beforehand. The goal is to design the mechanism in such way that the whole effort would work. Venture funds are structured in a similar way. I don’t think specific personal productivity should be controlled here. In either way it’s hardly practically possible.

1 Like

Brilliant. This proposal would create strong financial incentives for the assigned community leaders. Indirectly, all FEI/TRIBE holders would benefit.

Another FEI KPI I could imagine is TRIBE/ETH (TRIBE’s performance relative to ETH’s performance). This chart needs urgently some green candles.

PS. The reverse chart WETH/TRIBE exists on TradingView.

@pavel ,

Interesting idea, I did not come across this. Here is the implementation proposed by Index, perhaps with your help we can start to come up with a more defined list of parameters?
For example, if we want to come up with a MCap criteria, what should it be?

We want to balance between not too high, to be realistic, and not too low to be ambitious enough. E.g.: we can get target MCap growth rate by calculating normalised mean of MCap growth rates of DeFi Pulse index constituents during the equivalent period (like the first six months). Or some of them, i.e. to pick as a benchmark a panel of provably successful projects with well-grounded ambitions.

I think its great to start this conversation thanks for bringing this up @pavel! My main concern is that I’m not sure we have seen enough of the benefits of these KPI options in other projects who have used them relative to the costs of resourcing this.

What kinds of security and engineering efforts would need to go into this beyond tuning the parameters and determining the whitelist?

Technically this will leverage UMA infrastructure – a protocol for creating ERC-20 synthetic assets. It was audited by OpenZeppelin:

Specifically, work to be done:

  1. Draft and submit for UMA DAO consideration two UMIP proposals: the first adding TRIBE as UMA’s supported collateral type (they have a mandatory template); the second defining a methodology for how the UMA Optimistic Oracle should price KPI options (they also have a template). Maybe go through a couple of improvement iterations. Generally, as far as I can judge, they welcome new quality collaterals, votes happen weekly.

  2. Fork Expiring MultiParty contract (audited – 2, 4, 5): they’ve built a forkable script for EMP deployment with tunable EMP parameters; contact UMA team prior to deployment to give a general heads up, a chance to look over the parameters chosen to catch any errors.

  3. Mint KPI tokens by locking collateral in the contract: they have a forkable frond-end for minting KPI options

  4. Airdrop: a merkle-tree distributer appaudited by Solidified.

  5. Build a claim app, needed by expiration date.

I don’t know what you’re referring to – it’s a relatively new thing, UMA created them a couple of months ago, their own first batch hasn’t expired yet. Overall, stock options proved to be a pretty efficient incentivisation tool. KPI options are a bit different, but I think the success/failure is a function of the overall mechanism design, you could find the wright config. E.g.: I think UMA made a mistake overscattering the first batch, they allotted up to $UMA 2M (=$60M at that moment) to >7k addresses. Essentially I feel it turned into yet another free lunch airdrop. Instead you want to concentrate: maybe allocate less money, whitelist far fewer addresses, pick them more thoroughly. We could debate this more thoroughly.

You want to find a proper working balance between individual options for in-house developers and crowd airdrops.

A separate topic is a market for FEI KPI options.

Also

1 Like

We’ve been thinking a lot about the best incentives structure for those working for the DAO, and we hadn’t considered this idea before so thanks for bringing it up. I think this could work and will be keeping an eye out for how the other KPI options programs turn out to help evaluate whether Fei should do this.

I have two main concerns around implementing KPI options on the Fei DAO.

My first concern is about the selection of the KPI options recipients. This will likely be a very subjective process. How do we determine who is making significant contributions to increasing the market cap of Fei (or other KPIs)? Are the engineers submitting solidity code providing more or less value than those doing the analysis which parametrizes their code? These are hard questions to answer and whatever we come up with here will be contentious with a different part of the community. I share @Eswak 's concern about options recipients not helping in any way and still benefitting massively.

You also mention that the selection of KPI options recipients might need to be kept secret until after the snapshot. This complicates the governance process.

My second concern is that many of the KPIs you suggested could be gameable. Some of them are directly gameable by individuals, like potentially using flash loans to fake trading volume. Others are indirectly gameable like market cap and user circulating FEI: What if the KPI options recipients were incentivized to artificially inflate the market cap of Fei over a certain period of time? They could advocate for governance to increase user circulating Fei at the cost of decreasing the collateralization ratio and then profit handsomely from that without necessarily benefiting the protocol in the long-term.

I think it would be cool for Fei Protocol to experiment with new ways of remunerating community members and other stakeholders, but am held back by the above issues. @pavel, would love to hear some of your thoughts on these concerns.

3 Likes

Seb, first, position: I think both your concerns are manageable, the idea is worth looking for the solution.

Now, there could be multiple use cases for KPI options: rewarding existing community members, drawing new interest from the broader DeFi community, Treasury diversification as UNI is trying now. Hence approaches to recipient selection may also vary. As I told @Eswak the original UMA’s idea was to helicopter airdrop equal amounts of options to people, who at the moment have nothing to do with the protocol, thus stimulating their interest and incentivising them to start participation. The whole premise was that if you airdrop to a reasonably big number of reasonably qualified recipients, some of them will become active contributors — you can’t know beforehand which ones, but you’re making a broad set of bets such that those who win will cover for those who miss and the overall endeavour is worth it. As we can witness the first UMA batch flopped. I gave my first thoughts in the previous reply to @joey, we can elaborate further and maybe find a better mechanism.

As for personal incentives to contributing community members: I think there is no objective way to measure each person’s contribution into Fei’s growth, as you point out these inputs may come in various forms and it’s impossible to build a measure function on such heterogenous space or build a neural network architecture to map a set of posts, commits and whatever else to [0, 1], for example.

Ideally Fei should look for a reputation system, something like the following. Community members give each other scores, say from [0, 1] for each post, proposal, commit, maybe with some weight differentiation. Then a ranking mechanism aggregates subjective individual assessments and assigns global score/ranks to each member reflecting one’s input while being:

  • accurate, such that estimated reputation scores were close enough to ground truth member’s input allowing to fairly distribute rewards;

  • resistant to manipulations like sybil attacks (creating fake identities to boost one’s own ranking) or misreporting other members as less reputable than they actually are in order to diminish their ranking;

  • computationally efficient.

Then reputation scores can be translated into KPI options, for instance.

An example of such algorithm is Personalised Hitting Time (PHT). The community is represented as a weighted digraph, where members are vertices and a score given by a member i to a member j as a reflection of i-th belief of j-th merit is a weighted directed edge h_{ij}. Then PHT is the expected number of steps that a standard random walk starting at node k will take to hit node l for the first time.

It’s been proven that PHT presents an optimal trade-off between accuracy, robustness to manipulation and computational efficiency as compared to alternatives like PageRank and PathRank. However, there is a caveat: personalized reputation mechanisms are resistant to sybil attacks (global ones are not), but to adopt a PHT for our global ranking we need to select a pretrusted set of nodes. Alternatively/complimentarily by modeling the community as a graph, Graph Neural Networks can be utilized here to learn user and input embeddings, something akin GraphRec, but that’s a long topic — we can dive deeper, if you’re open to it.

In addition there is a problem of implementation of such mechanism. Clearly current home platforms of the community, Discord, Discourse, don’t allow this experiment. An example of a platform moving in this direction is Geneva — they’re integrating communication tools like real-time chats, blog-posts, zoom- & clubhouse-like conferencing and IGTV-like broadcasting, but also to quote “… started working on a number of exciting web3 projects related to ownership and governance that’ll open up new possibilities for DAOs and other collectively-operated communities.” I’m guessing they’re working on a platform in the vein of what I’ve described. Fei could either join or build a tool of their own, it could be a very important experiment for DeFi and crypto, though obviously it’ll take resources, so it’s your call.

As for KPIs resistant to manipulation I think we could start the discussion with rolling/global median, maybe devise a robust statistic — there is a whole field called robust statistics, which has numerous tools to sift out outbreaks from the sample.

@sebastian

1 Like