Happy Sunday and welcome to InsideAI! If you haven't done so yet, be sure to check out our Investing in AI podcast as well. This week I want to talk about DAOs - Decentralized Autonomous Organizations and give some thoughts that might be a little contrarian at the moment.
DAOs can be many things, but for those of you new to the concept, one way to think of a DAO is a group of people coming together to complete a project who aren't necessarily affiliated with each other in any "normal" way. The record of their work is controlled by a blockchain and people work on different pieces of the project and when it is done the payment or reward is distributed to everyone according to some pre-determined rule. The DAO that actually made DAOs popular was a project that had people contribute money to it, then vote on how to invest it. Unfortunately that got hacked and caused a huge rift in the project and the Ethereum community so we never got to see it's real potential.
There are two reasons I decided to write about this today. One is that as an AI investor, it seems that most VCs I speak to are blockchain crazy and think it's the most disruptive technology out there, while they tell me AI is just "basic data science with a new name." I 100% disagree. I'm short blockchain and long AI. The second reason is that I actually have some experience relevant to the DAO concept, and I think that experience backs up the statement that is the title of this post - AI, not Blockchain, will be the key to successful DAOs.
Before I tell my story, let me summarize my argument. It's simple. The problem with DAOs is the humans. People are treating DAOs like somehow if we could just have a shared public record of tasks then we could get these DAOs to work. No. That isn't the problem. Humans are the problem. When we can replace the humans with another form of intelligence, then DAOs will take off - with or without blockchain.
The story I'm about to tell originated in 2005. Jory Des Jardins actually wrote about it in Fast Company that year and you can read her article here. I was a very early blogger and my first post was in March of 2003. By virtue of being early, my blog became reasonably popular. In 2005 I had read the book "Wisdom of Crowds" and I got into an argument with a friend about if crowds could really make better decisions than individuals for areas like business. He said yes. I said no. Then I said this is something we could test.
We set up a website called "The Business Experiment" and had 1000 people sign up to participate. The idea is that everything about the business would be determined by majority vote, and equity in the project would be awarded based on participation and work for TBE (as we nicknamed it). People loved the idea of such an equitable and fair project, but what we learned would come to skewer my view of human nature forever. TBE was a massive failure, and it wasn't because we didn't have a shared public ledger (i.e. blockchain) to help us. It was because of human behavior. While there are many examples I could give, I'll stick to the top 3 events that ruined TBE.
First of all, we ran a head-to-head idea competition, with majority voting, to determine what idea TBE would pursue. The 1000 members submitted ideas and we would put two up against each other, with the winner moving on. The winning idea wasn't surprising - it was that we should build a website where small businesses could tap into the wisdom of crowds for help. But some people didn't like that idea and this was problem number one - when your favorite idea was voted out you stopped participating. By the time we got down to the final idea, fewer than 300 people were still involved even in voting, because others had quit.
Now imagine this in a work context with a blockchain DAO. You have a project and that project takes a turn you don't like or don't agree with, so you quit, and now someone else has to pick it up, come back up to speed on the right context, etc. This doesn't happen in regular corporations because when you have a job or a consulting contract or whatever, it's much harder to get out of it and find something else. The friction to change is high, so, if the project you are working on goes a little off what you think it should be, you still stay and work on it. If blockchain DAOs lower the friction of humans to come and go on projects, the churn will be high.
The second thing that happened at TBE is that we needed money to pay some basic expenses. So of course we voted on how to raise money, and the winning idea was to sell TBE branded merchandise. This was a clever and I figured with 1000 participants we could raise a decent amount just from our group. A graphic designer who was participating donated a free logo and we setup a website where people could buy TBE merchandise. We promoted it to the group, and we sold 3 items total - and one of those was a TBE coffee mug that I bought. So what people wanted was everyone else to pony up money for TBE, and not have to do it themselves. This type of free riding is going to happen in DAOs, and it isn't fixable by smart contracts or blockchain because you still have to have a human component. Unless the contract is super simple and clearly measurable, which most tasks aren't, there is a human judgment step, and that step is the weak link in the DAO.
This is why I hate paying bonuses to employees unless it's just a simple time based bonus. Every time I've designed a bonus plan and taken it to my Board, and we work through it and make sure it's super tight and crystal clear, at the end when it's time to pay the bonus, the employee always argues for a different interpretation than we had, and often makes a good case. It's hard to get these things write, and more automated contracts isn't the solution - the solution is fewer humans involved in making the decision.
The final straw that made me shut down TBE was this... two programmers agreed to build the software for the winning idea. Their software would manage the process of small business owners tapping into the wisdom of crowds. They agreed to do it for a big chunk of equity points in TBE. And then, as the software came close to completion, they told me they had been thinking about it and were really on to something and didn't want to share it with the crowd but wanted to take it and start their own company. It's tough to see how a blockchain could have enforced this. If a smart contract held back payment as a result, these guys wouldn't have cared. That wasn't our problem. Our problem was human nature. If an AI (maybe OpenAI Codex style bot) had written the software, it would not have taken it out on it's own but would have contributed the software to the project like planned.
This was all brought back top of mind this weekend when my friend Charlie O'Donnell texted me to remind me of TBE and point out it was an early DAO-like construct and that I should write about it. And I pointed out that I don't think blockchain could have fixed TBE, and I don't think blockchain is the secret to building DAOs. To build successful DAOs we need intelligent machines. We need things that can do tasks humans can do without the emotional characteristics of humans that make us bored, whiny, selfish, etc.
So that's my argument in a nutshell. As a disclaimer, this all happened 15 years ago and some of the TBE members probably read this newsletter so if you remember things differently feel free to write about it, or respond here and tell me your point of view.
In summary though, my argument is that the thing holding back DAOs is not a lack of a shared public ledger and some better smart contracts - it's that DAOs in most forms work in ways that allow the bad parts of human nature to come out, and to get rid of that, we need intelligent machines that are more rational and less emotional.
If you have opinions or thoughts on the topic, I'd love to hear them as always. Thanks for reading.