Markets in Democracy: What happens when you can sell your vote?
So, about 25 years ago, I had the following shower thought: What if all the money spent on political advertising went to the voters instead? What if we could just sell our votes?
- Voters who care could just vote normally
- Voters who don’t care could get paid
- Voters who care a lot can actually spend money and have an outsized impact on outcomes.
- Speculators and arbitrageurs can make a profit helping price discovery.
- Political advertising goes away and there is much rejoicing.
One thing that struck me is that for the life of me I couldn’t figure out how much a vote in a U.S. election would cost. Not even an order of magnitude. There is a real missing piece of information in modern democracy, namely, how much do people actually care about the outcome?
Of course, this isn’t perfect. The main problem I can see is that we start out with a highly asymmetric playing field, in which it’s dramatically easier for some players than others to buy influence. But this problem, while insurmountable for national elections, is easily surmountable for a variety of other scenarios (think corporate governance, HOA’s and DAO’s). And whereas these don’t suffer from the scourge of political advertising, the price discovery mechanism and the option to express preference intensity still offers benefits.
The idea sat around in my head for a couple of decades. When smart contracts and DAO’s appeared, it came back to my mind and I thought about implementing it, but I was busy and the learning curve was a pain, etc, etc. Then, when Claude code came out, this was the first project I built with it.
## What I’ve built
Well, I’ve built a DAO in which votes are saleable. The basic mechanism:
- Buy governance tokens from the DAO → Your payment goes to the treasury
- Token holders create proposals (textual resolutions, treasury transfers, token minting, governance token price changes)
- When a proposal gets enough support, an election triggers
- Each token holder receives voting tokens (1 for each governance token they hold)
- Voting tokens are tradeable on DEXs during the election
- To vote, send tokens to YES or NO addresses
- Majority wins (if quorum is met)
Key differentiator: Governance tokens are long term membership stakes. Voting tokens are tradeable influence on a *specific decision*.
The code is available at https://github.com/evronm/marketDAO and the fronend for the test-net (Polygon Amoy) deployment is available at https://evronm.github.io/marketDAO/index.html .
# Some Game Theory and a Fatal Flaw
Here’s the obvious treasury raid attack:
- Buy new governance tokens adding up to more than the total in existence.
- Make a proposal to trasfer the entire treasury to yourself
- Vote “YES” with your vote tokens, which now add up to more than half.
- The treasury is yours (including the funds you spent on the governance tokens), and so is control of the DAO, at least until (unless) somebody buys more governance tokens.
This is a fundamental flaw, but there are potential solutions:
1. Disable token purchases after initial distribution
- Secondary market for governance tokens can still exist
- Tokens can still be minted by regular proposal/vote
- Can be implemented with software as is
2. Vesting period for new purchases
- In the event of a treasury raid, existing participants have ample time to mint more governance tokens
- Not implemented yet
3. Conviction voting hybrid
- Grant more voting tokens to older governance tokens
- Complicated
- Not implemented
Can anyone think of additional options? If not, I’ll probably implement a configurable vesting period. I don’t think I’ll bother with the conviction voting thing unless there is real demand.
# Additional Questions
- What affects governance token price (remember, this is set at deploy time and changeable by vote)? Is it Treasury size? Number of participants?
- What affects vote price for different vote types? I’m particularly interested in the dynamics of governance token price changes, but treasury transfers and token mints should be interesting as well.
- What types of organizations is this kind of structure best suited to?
- Is there an optimal Gini coefficient for a structure like this?
# Why I’m Posting Here
- I’ve hit a wall theorizing. Most of my questions can probably only be answered empirically and I’m hoping some folks decide to play with it and get me some data.
- From what I’ve seen, this community has an interest in mechanism design. If there are obvious failure modes I’m missing, someone will spot them right away.
- This mechanism raises a lot of interesting questions. Even if the final verdict is “there is no real world use for this thing,” I think the academic exercise of price discovery can net someone a paper or two (I lack the academic credentials to go this route).
So please, try it, break it, or tell me why it’s doomed. I’d be happy to do custom deployments if anybody wants that.
What about the situation where you get fired unless you sell your vote to the company? Or maybe to the union?
Or the situation where the mob demands that you sell your vote to them so that they can use it for some socially beneficial purpose, and otherwise they cancel you?
It seems to me it should always be cheaper to buy a vote than to coerce one, especially if you price in risk. Do you disagree?
I would disagree that people in the real world act based on what’s cheaper. None of the cancellations we already see over other things are done so as to be financially optimal for the cancellers. Even companies don’t act based on what’s financially optimal; if Google was willing to fire James Damore, Google certainly would be willing to fire people for not selling their vote to Google. If Disney is willing to lose millions, maybe billions, of dollars through woke Marvel and Star Wars, I’m pretty sure they’d be willing to fire people who won’t sell their votes to them, even if it “isn’t cheaper”.
It’s true that the market hurts companies that do this sort of stuff, but it takes a long time between when someone loses money because they are acting against the market, and when they actually go out of business. Disney isn’t about to die soon.
And punishing some people does often benefit them financially anyway because even though firing an employee costs money, it also intimidates other employees, reducing the sale price of their votes.
There’s also the issue that some people won’t sell their vote for the financially optimal price either, so the company or the mob will threaten to fire them to force them to. Many people wouldn’t, absent coercion, sell their vote for any price, just like many people won’t sell sex. Or they will, but only for a life-changing amount that is many times the market price.
AFAIK, it’s already illegal to fire somebody for their political affiliation. Firing or threatening to fire someone for not selling their vote could likewise result in a hefty fine.
I’d be worried about further disenfranchisement of the poor. I won’t claim that this is a fatal flaw, but I think it should be addressed.
To make it more concrete, a fictional scenario:
To someone who sleeps outside with no income, selling their vote for $1000 to some rich dude sounds like a good deal, especially when that vote has a negligible chance of winning them a better situation.
That’s okay on an individual level of choice, but scaled up it’s a misalignment of incentives.
This does seem like a good idea for small organizations, but at some scale, I expect bad actors to break the system.
OK, think of it this way: A poor person gets government benefits. He’d be a fool to part with his vote for less than the amount of the benefits he gets. He has to assume that a rich person is buying his vote in order to pay lower taxes and deprive the poor of their benefits. So it makes sense for the poor to either vote or demand a price for their vote commensurate with the benefits they currently receive.
That said, immediate need may trump rational calculation, as, indeed may a sheer emotional response. So again, no good for things like national elections (probably; though not as bad as it initially sounds). But the question remains: in what type of environment would this work best? I’m thinking resource allocation among peers (e.g. investment groups or corporate bodies) but would love to hear additional takes.
Doesn’t seem right. Even assuming the person buying his vote wants to use it to remove his benefits, that one vote is unlikely to be the difference between the vote-buyer’s candidate winning and losing. The expected effect of the vote on the benefits is going to be much less than the size of the benefits.
The reasoning would hold if people reasoned as if they were deciding for their entire class, but not if they believe they’re deciding for just themselves. Unfortunately, the religious principle of deciding as if one is deciding for one’s entire agent-class (FDT) is not exactly ubiquitous.
Well, the entire class is receiving a much larger benefit than any member of the class, so the class as a whole would sell it’s votes for that amount. An individual voter would have to make choices based on the benefits they receive. If I stand to lose $1k by selling my vote, I’m not going to sell it for less than $1k. That said, I’m not guaranteed to lose the whole $1k or in fact any of it, so there is a “probability discount”. Maybe the floor isn’t the exact amount of benefits but some percentage of it. Looks like I have another question :D.
If there are a hundred thousand people like you who are being asked to sell your vote and you stand to lose $1000 from the wrong person being elected, your vote can only be 1/100000 of the reason the person is elected, so you should sell it for $1000/100000 = 1 cent. Under the proper decision theory, you shouldn’t sell it for less than $1000, but the number of people in the real world who understand (let alone both understand and agree with) such decision theories is negligible on the scale of voting.
All of which goes back to my point: even if we know the benefits received, we can’t know in advance how much a vote would go for. It’s a missing piece of information from standard democracy that can only be obtained empirically.
Again, I know this will never happen at the level of politics, so this entire discussion is purely academic. What I’m more interested in is how this plays out under closer to ideal conditions (smallish group with similar resources and similar smarts).
You could have the proceeds from new governance tokens go to the governors instead of the treasury. You could make any motion that doesn’t reach consensus split the DAO: If p% were in favor, the company has a p% chance of being split up among those in favor, and a 1-p% chance of being split up among those opposed.
These are really interesting mechanisms you’re proposing. The first one I think amounts to something along the lines of how Moloch DAO works. Essentially you’re paying existing members to join the club. I could definitely make that an option if there’s demand (thank you Claude Code!).
As for the DAO splitting, that sounds fascinating, but I don’t understand it. Care to elaborate?
My purpose for the first is to not let a raider steal his entry fee too.
My purpose for the second is to effectively require consensus for any motions to pass, while leaving a way out of deadlock that is neutral in expectation.
I don’t think selling or buying votes is a good solution. There is a reason why we have multiple forms of social institutions: they solve different types of problems using different types of tools. If you think the problems of governance and policy are solved via a market mechanism I think your first task is to show why the market solution never emerged in the first place. But I think if you want to go down that path why stop at votes, why not take David Friedman’s approach of a market anarchy where government largely doesn’t exist but private solutions provided in a competitive market setting do.
Well, I am a minimalist when it comes to government. But regardless of whether or not it’s government, there is a variety of situations wherein group decisions have to be made. There are already many mechanisms around for such decisions. I’ve just suggested another one. As I’ve mentioned, it’s definitely not ideal for all situations. And one of my questions continues to be in what type of organization would this be most effective?
I’m happy to keep having the conversation about vote buying in politics. It’s an interesting academic exercise. However, there is no way in hell this ever gets implemented by a national government, so any such conversations are necessarily academic. What’s more interesting to me is real world organizations that could adopt such a mechanism.
What if the oldest 60% of governors pull the raid?
Yes! Another interesting dynamic to think about. At what distribution/Gini coefficient does that become improbable or impossible?
You mean, are 100 founders enough that 60% cannot coordinate a raid? You’ll have trouble telling whether 60 of the founders are a rich guy in sixty trenchcoats.
Well, it depends on how you do the initial distribution. If you just put up the tokens for sale, then yes, you’d have the problem of one person buying up all the governance tokens. But if you distribute initially to e.g. 100 known individuals (contributors, etc), then you need actual collusion and at some scale that becomes difficult.