Emergent Outcomes of the veToken Model

Introduction to the show: ident: AI Radio. Generated commentary on the latest Artificial Intelligence papers.

Tom: Next we'll be talking about the paper "Emergent Outcomes of the veToken Model".

Jane: The paper was written by Thomas Lloyd, Daire O'Broin and Martin Harrigan from Department of Computing, Carlow Campus, South East Technological University.

Tom: Stay tuned as we take you through the paper and discuss its implications.

Title: Tom: Welcome back to the show, everyone. Today we’re digging into a paper that’s been making the rounds on arXiv, and it’s called “Emergent Outcomes of the veToken Model.” Jane, I’ll be honest, when I first saw that title, I thought it was about some new kind of electric vehicle token. But no, this is about governance in decentralized finance.

Jane: Ha, that’s a fair guess, Tom. But veToken actually stands for “vote-escrowed token.” It’s a way that blockchain projects let people lock up their tokens for a certain amount of time in exchange for voting power. The longer you lock, the more say you get. And this paper is all about what happens when you actually put that system into practice.

Tom: Right, and the authors are Thomas Lloyd, Daire O’Broin, and Martin Harrigan from South East Technological University in Ireland. They’re looking at Curve Finance, which is this big decentralized exchange for stablecoins, and they’re tracing how the veToken model plays out in the wild. It’s not just a theoretical thing—they’re pulling real transaction data.

Jane: Exactly. And what’s fascinating is that the model was designed to solve a real problem. In older systems, it was one token, one vote. That meant someone could just borrow a ton of tokens, vote for something crazy, and then give the tokens back. There was actually a case in two thousand twenty-two where someone did exactly that and drained a hundred and eighty-two million dollars from a project called Beanstalk.

Tom: Yeah, I remember that. It was like a corporate raid but at lightning speed. So the veToken model tries to fix that by making you commit your tokens for months or even years. In Curve’s case, you can lock for up to four years, and your voting power scales with that time. So someone locking for a week gets way less weight than someone locking for the full four years.

Jane: And that’s the core idea. But the paper’s title says “emergent outcomes,” and that’s the key. Because once you create this system, people build other things on top of it. Yield aggregators, voting markets, bribe mechanisms. It becomes this whole ecosystem that the original designers probably didn’t fully anticipate. And that’s what makes this paper so interesting.

Tom: So it’s not just about the model itself, it’s about the unintended consequences. And there are some wild ones in here. I can’t wait to get into the details. Stick around, because we’re going to look at how votes literally follow bribes in this ecosystem.

Summary: Tom: Alright, we’re back with “Emergent Outcomes of the veToken Model.” Jane, let’s get into the meat of it. What did these researchers actually find when they looked at Curve and all the protocols built on top of it?

Jane: So they gathered data from three levels. First, there’s Curve itself, where you lock CRV tokens to get veCRV voting power. Then there’s Convex Finance, which is a yield aggregator that locks CRV on behalf of users and gives them their own token called CVX. And then there’s Votium, which is a voting market where anyone can pay bribes to influence how people vote.

Tom: And the numbers are pretty striking. Convex holds about forty-five percent of all locked CRV. So one protocol has nearly half the voting power in Curve. That’s a massive concentration of influence, even though it’s technically spread across Convex’s own users.

Jane: Right, and then Votium has distributed over two hundred and forty-eight million dollars in bribes since it launched. And here’s the kicker—when they compared the bribes directed to each gauge with the votes each gauge received, the correlation was almost perfect. In the mature phase of the data, the correlation coefficient was zero point nine nine. That’s about as close to one as you can get.

Tom: So you’re telling me that if someone pays enough bribes, they can basically dictate the outcome of these governance votes. That sounds like it undermines the whole point of decentralized governance.

Jane: It does, but it’s also kind of by design. The gauge votes decide how new CRV tokens are distributed to liquidity pools. So if you’re a protocol that wants more rewards flowing to your pool, you have an incentive to pay voters. And Votium makes that easy. The paper even quotes Charlie Munger: “Show me the incentive, and I will show you the outcome.”

Tom: And it’s not just random actors doing this. The paper highlights Frax Finance, a stablecoin issuer, as the biggest player. They’ve spent over a hundred million dollars in bribes through Votium, which is about forty-one percent of all bribes. But here’s the twist—Frax directly locks very few CRV tokens. They’re getting influence through indirect channels.

Jane: Exactly. And that leads to the paper’s second big finding about the cost of votes. You can acquire voting power by locking CRV directly, by locking CVX through Convex, or by paying bribes through Votium. And the cost per vote is different at each level. Right now, bribes through Votium are the cheapest way to get votes.

Tom: So the more indirect and less committed you are, the cheaper it is to buy influence. That seems backwards from what the veToken model was trying to achieve. The whole point was to align voters with long-term interests, but the system ends up rewarding short-term cash payments instead.

Jane: That’s the emergent outcome, Tom. The model works in isolation, but once you add these higher-level protocols, the incentives shift. And that’s what makes this paper so valuable—it shows that governance models need to be studied in context, not just in theory.

Improvements: Tom: We’re back with “Emergent Outcomes of the veToken Model,” and I want to push on something. The paper doesn’t just describe problems—it also hints at what could be done better. Jane, what are the improvements they’re suggesting?

Jane: Well, the paper doesn’t prescribe a specific fix, but it does highlight where the model breaks down. One clear issue is voter participation. For the economically incentivized gauge votes, participation is high. But for non-gauge proposals—like whether to add a new pool—only about twenty-four addresses vote on average. That’s a tiny number for a supposedly decentralized system.

Tom: So the incentives only work for the votes that have money attached. Everything else gets ignored. That’s a governance gap.

Jane: Right. And the paper also points out that the lockup periods are inconsistent across levels. Curve requires up to four years, but Convex only requires sixteen weeks for their version, and Votium has no lockup at all. So you can get voting power with almost no commitment if you go through the right channels.

Tom: And that’s where the cost per vote gets distorted. The paper shows that bribes through Votium are cheaper per vote than locking tokens directly. So the system is rewarding people who are least committed. That seems like something a decentralized organization should address.

Jane: The authors suggest that organizations considering the veToken model should think carefully about these emergent outcomes before adopting it. They’re not saying the model is broken—they’re saying it’s more complex than it appears. And if you don’t anticipate the higher-level protocols, you might end up with governance that’s less decentralized than you thought.

Tom: So the improvement isn’t a technical patch. It’s about awareness and design. If you know that voting markets will emerge, you can build safeguards. Maybe shorter lockup periods at the base level, or rules about how much voting power any single protocol can hold.

Jane: Exactly. And the paper’s future work section mentions extending the analysis to other implementations of the veToken model. So they want to see if these patterns hold across different projects, not just Curve. That would help the whole space learn what works and what doesn’t.

Tom: It’s almost like they’re saying, “Here’s the warning, now go build better systems.” And I think that’s a really valuable contribution. But I’m curious what our listeners think—does this mean the veToken model is fundamentally flawed, or just that it needs better guardrails?

Jane: That’s the big question. And I think the answer depends on what you value. If you care about efficiency and participation, the model does great. If you care about equal influence and long-term alignment, it has serious issues. There’s no free lunch in governance design.

Conclusion: Tom: Alright, we’re wrapping up our discussion of “Emergent Outcomes of the veToken Model.” Jane, give us the final takeaway.

Jane: So the paper takes a governance model that sounds simple—lock tokens, get voting power, vote on rewards—and shows that in practice, it creates a whole ecosystem of intermediaries. Convex holds nearly half of Curve’s voting power, Votium has distributed hundreds of millions in bribes, and votes follow those bribes almost perfectly. Frax, a stablecoin issuer, gets massive influence without locking many tokens directly.

Tom: And the cost per vote is actually cheaper when you go through these indirect channels. So the system that was supposed to align voters with long-term interests ends up rewarding short-term cash payments. That’s a profound finding.

Jane: It is. And the authors’ message is clear: if you’re a decentralized organization thinking about adopting the veToken model, don’t just look at the model in isolation. Look at what people will build on top of it. Because those emergent outcomes will shape your governance more than the original design ever will.

Tom: Well said. It’s a paper that makes you rethink what decentralized governance really means. And with that, we’re saying goodbye to this one. Thanks for joining us, and we’ll be back soon with another paper from the arXiv. Until then, keep questioning the systems we build.