Some gentle criticisms and comments on Vitalik/Gitcoin CLR
Gitcoin’s CLR matching is over. Vitalik published his evaluation of it.
Here’s my perspective, as someone who participated in it as both a recipient and a donor.
1. It’s a popularity contest. More importantly, it’s an EARLY popularity contest.
CLR is transparently a popularity contest. Check the leaderboard: the number of donors predicts your rank almost perfectly. That’s the point of the algorithm: it’s a signalling mechanism, not a donation incentivization mechanism.
But what is less apparent is that it is an early popularity contest. The number of donors in the first 3 days is probably almost as as good a predictor of your ranking as total number of donors over the entire 2 week period.
How could that be?
If you got a few contributions early, then your estimated matching number (the one in green) would be much higher than someone who got just a few less contributions than you.
This is one place where the Matthew Principle (”to everyone who has, more will be given”) holds. If you’re a donor evaluating where you want your money to go, then you are likely to give to the projects with high matching estimates - the Matthew Principle in action.
Furthermore, projects that have the most donations already get seen the most. The default ordering of grants on Gitcoin’s site is the “weighted shuffle.” That means that recipients who already have the most donors will appear higher on the page, on average. Given how many grants there were (~250 in the tech category), if you don’t get near the top early on, you probably didn’t even get seen by most potential donors for the rest of the 2 week period. The Matthew Principle strikes again.
This is particularly important because a few extra donors gets you a disproportionate amount of extra funding. Notice how in the tech category, 3x more donors got 9x more matching funding.
Several caveats apply here:
- This is somewhat an effect of the Gitcoin website design, but keep in mind that this data is all on-chain, so the data is inherently transparent and likely something that donors want to see. Gitcoin is probably making the right decision to show it to you.
- Given the sheer number of grants this round (300+ between both categories), most donors will not look through all of them, so the weighted shuffle is probably the best option.
- You could limit the number of recipients so that donors are limited to something more reasonable like 50 choices, but that inherently defeats the purpose of the mechanism.
tldr: if you didn’t get a few donors early, you had almost no chance at winning. I was a day late in starting, and had to campaign much harder than my competition because I started with a significant disadvantage. I was behind until the very end.
2. What happens when scams submit projects and get real people to donate?
Part of the point of this mechanism is that the populace can vote with their money on what they think should be funded, and thus anyone can submit something to be funded.
We saw a few trolls submit this round and get a few DAI out of it. But they were late to the game, and we’ve already established that being late is a substantial hindrance.
What happens when they organize 100 of their friends in advance to give 1 DAI on the first day?
Take it a step further, what happens when IOTA,Tron, BSV, or XRPbots submit and get 1000 real people to donate on the last day?
In a similar vein, what if Bitconnect in q4 2017 had tried to fund Carlos Matos appearing at their next event through this CLR?
It’s important to draw boundaries from the start. Negative voting can help with this - but may be an imperfect solution, particularly in the latter 2 examples.
I would suggest that we explicitly designate all Gitcoin grants to be “things that benefit the Ethereum community.”
Norms matter.
3. Estimates are marketing
No disrepect to Gitcoin, but the estimates in the image above are marketing. That is, with only minor simplifications, the estimate assumes that no one else donates after you, and that your next donation is the last donation of the round.
That’s obviously not a reasonable assumption, except for the last donation of the last donor. The graphic above that said that a $1 donation would get matched with $156 was right at the time, but my own example makes my point for me: I had 133 donors, and got way less than than 133*156.
Kudos to Gitcoin with the marketing - it does a great job of getting people to donate. It’s the best they can do, given that further explanation would be quite complicated. That’s a perfect transition to the next point:
4. Too confusing
Education takes time. But I talked to plenty of very smart people in this space who are heads down on their projects and startups who had no idea how to think about how to donate in even the most basic way. “Just give 1 DAI, it’s fine” seemed like a tough thing for most people to grasp - though this may be merely a matter of developing norms which remove the social stigma around appearing cheap by only giving 1 DAI.
It’s tough to develop an intuition for the CLR mechanism. I have read the posts and still am unclear whether there 5 people who give mostly to the same projects have their donations weighted less due to a collusion factor. So imagine your average Solidity dev (by definition: an above average human!) who has spent very little time thinking about “quadratic funding” and shows up to the site for the first time
I think the intuition will come in time (”just give 1 DAI!”) but it was harder than I thought.
5. No accountability.
CLR grants and matching donations are closer to token sales than they are to traditional grants-making organizations. CLR grants are no-strings-attached gifts. Contrast that to grants in the traditional world which usually have milestone payments or balloon payments at the end, contingent on performance.
String-free gifts aren’t intrinsically a bad thing (and do possibly some with social expectations), but it makes them different from the kind of work that you would likely fund via grants. In some sense, traditional grants are about the future, whereas these CLR grants are usually about gratitude for the past.
One final note: CLR grants are best for people who can motivate some segment of strong supporters who will go give them no-strings-attached gifts. It is not for people who whose work is appreciated by everyone, but only a little.
In the spirit of Gitcoin grants, I immediately distributed much of the donations I received (just like I did with the Week in Ethereum News 3 year anniversary NFT) and will likely do so with all matching as well, resulting in zero net personal gain.
It should be obvious that I treated this as a game. It was a game! Hopefully if you read this, you’ll have some ideas for how to play the game better yourself.