I've had two deja vu moments on Farcaster this year.

Oddly enough, I can't quite remember one of them. But the other was talking to Kenny about how communities make decisions and execute the results of those decisions onchain. Somehow, we started out talking about AirSwap and ended up talking about decentralised decision making.

Chukwuka Osakwe on Warpcast

airswap is decentralised and it wasn't about wrapping tokens. what is going on?

Farcaster

I realised it was a deja vu moment when Kenny sent me his essay on the launch of the Pics Or It Didn't Happen v2, and why that launch matters for the future of onchain communities. I knew I'd read this essay before, and seeing as Kenny only shills it when discussing this particular issue, I figured we'd had this conversation before.

poidh v2

Happy New Year! For us, 2023 will forever be remembered as the year that poidh went from an idea to a fully formed dapp with users, fans, and real ec...

https://paragraph.xyz

Pics Or It Didn't Happen

Pics Or It Didn't Happen (POIDH, henceforth) is a web3 bounty platform. People post bounties that they want someone to do, and anyone can carry out the bounty and get paid by uploading a picture proving that they've taken the required action (hence the name).

Here is a simple example.

Imagine that I want someone to eat a green apple while standing in the Eiffel Tower.

I post a bounty on POIDH with the appropriate description, and some amount of money that will be the reward, let's say 0.1ETH. Anyone, anywhere, can take a picture of themselves eating a green apple in the Eiffel Tower and upload it to the bounty page as proof of completion. The bounty creator, on verifying that the picture meets all the necessary criteria then releases the funds to the person.

Other people who are interested in the apple-eating endeavour can also add funds to the reward pool to increase the total amount of money payable to the eventual winner. Those who add funds to the pool also get to be involved in the voting process to pick the winner in this scenario. Since there are more people involved than just the bounty creator, everyone who has put in funds gets a say in deciding the eventual winner.

Vamos Atleti

So, I made a bounty.

Pretty much everything I do in web3 these days is somewhat football-oriented and this bounty was as well. It was simple enough, I just wanted someone to post a picture of an Atletico Madrid jersey with their Farcaster handle on the back of it.

I reasoned that the bounty had to be somewhat difficult to complete because the longer it ran, the more opportunity I had to get people to add funds to the rewards pool and increase the payout to the eventual winner.

Chukwuka Osakwe on Warpcast

per last conversation with kenny, a /poidh bounty for the football channel. post a picture of an atleti jersey with your farcaster handle at the back of the jersey to win some bounty money. > anyone can add funds to the rewards pool to increase the total reward and make it likelier that this bounty gets claimed. let's go and vamos atleti! https://warpcast.com/kenny/0xc0acbc72

Farcaster

Finding an Atletico Madrid jersey seemed hard enough, but one with a Farcaster handle on it? That is something that has to be made from scratch. As difficult a football-related bounty as I could think of.

In retrospect, I should have gone with just an Atleti jersey. Requiring the jersey to be made from scratch meant that there had to be enough reward money to cover the cost of making the jersey and then some more left over. Otherwise, it would make no economic sense for someone to go out of their way to make this jersey.

In the end, we had to go with just an Atleti jersey and the person's name on the back, no Farcaster handle. But, you know, you live and you learn.

Some Thoughts Then

LOOPHOLES. The first thing that surprised me was that immediately the bounty was posted, a couple of entries came in. Apparently, some people had created AI-generated images and submitted them to the bounty page.

I thought about it for a few moments and decided, hey, this is actually legit. Nowhere in the bounty did it say that the jersey had to be an irl jersey, and these guys tried to take advantage of that loophole. Of course, I meant for it to be an irl jersey, but my description didn't sufficiently capture that intent, and these people tried to seize on that loophole. Fair play to them.

This made me think about smart contracts as law. The appeal of smart contracts, I'm told, is that it is code and it will always execute when the criteria are met, no need to trust a middleman when you use a smart contract. And yet, I've always thought about how this means that you have to nail down even the slightest details when describing what criteria need to be met for a contract to be executed. And in an age of AI, LLMs, prompting, and NLP, that is even more important, I imagine.

I think about a possible future where POIDH claims are automatically evaluated by AI. You'd better be extra careful when wording your bounty (or prompting AI to write it for you). All your prize money could be captured by a smart guy somewhere, ready to exploit loopholes. Maybe we will see MEV bots for POIDH bounties at some point.

DECENTRALISATION. If you read the cast thread above, then you know that one of the things Kenny and I talked about was the kind of centralisation that comes about in DAOs when funds are concentrated in a multisig and the DAO depends on a few people to execute their decisions with funds from the multisig.

POIDH circumvents that centralisation by making it possible for groups of people to pool funds and get things done onchain together. Kenny and I put funds into the pool (but anyone could have joined) to incentivise people to complete the bounty. When a valid entry comes in, the bounty creator can trigger a vote which everyone who has put in funds can participate in to determine if the entry passes muster and wins the bounty.

Everyone who puts funds in is entitled to a vote proportional to how much funds they have put in. Say we have a total of $100 in the bounty reward pool and 5 people have contributed $20 each to the pool. They all receive equal voting rights (20%) based on the proportion of their contribution to the entire pool.

Once the bounty creator triggers the voting process for a verified claim, there is a voting period for the contributors to vote on the claim. All a claim needs to pass is to receive greater than 50% of the votes that are cast in this period. That means that even if certain contributors don't vote, it doesn't matter, as long as the claim receives more than 50% of the votes that are cast.

I don't know if this is the final boss of decentralising community action onchain, but it's a start. And it's a great start because I do agree with Kenny. The multisig model of communal action always seemed very trusting to me. It's the same old centralised, let's-absolutely-depend-on-a-small-group-of-people-with-the-real-power model all over again. We need to evolve.

SCALE. I'm very interested in seeing how POIDH handles scale as it gets bigger. I suspect that one of the reasons that DAOs eventually default to the multisig model of decision-making is scale. When you have hundreds of people in a DAO (or any kind of group, really), it is usually the case that some people are way more active than others and that decision-making and execution tend to eventually end up in the hands of the few.

Will it be the case, maybe, that as larger groups of people start to use POIDH that it becomes a lot more centralised than it is at the moment? Imagine a Farcaster channel posting bounties and all members of the channel chip in to fund the pool, which gives them voting powers. But when it comes time to actually vote, how many people will be involved in that process? Especially if evaluating the validity of claims requires some level of focus, attention, and expertise. If you have hundreds of people in the channel and the channel needs to decide on something design-related, it makes sense to leave the details of that decision in the hands of a few people who have extensive design expertise.

I imagine that eventually, it will only be some people who are invested enough to go through claims and validate them, and it is they who will vote, and eventually, we're right back where we started: centralisation.

In Conclusion

Few things irk me as much as a web3 product re-inventing the wheel. We desperately need to go in new directions, to invent new paradigms, to try out new ideas, to see how we can get closer to enabling the kind of world this technology promised to deliver on those cypherpunk mailing lists.

I'm always excited to run into and support anyone trying to do something new, and Kenny is trying to do something new with POIDH. Check him out, check POIDH out, support him, give him feedback, and godspeed to us all.

kenny 🎩 (@kenny) on Farcaster

intersubjective mechanism design @ poidh.xyz - come chill /poidh + follow @poidhbot

Farcaster