DAOs are supposed to be a higher form of governance. The argument is that decentralization makes them less corruptible and more democratic than the corporate boards and state institutions they aim to replace. For this issue, we're looking at bribes and conflicts of interest within the DAO world and what data is available about them.
Discourse tends to split roughly three ways. One position holds that decentralization makes DAOs structurally resistant to corruption. Another holds that whales buy outcomes. A third holds that you can't really tell from the outside.
Bribery in crypto governance isn't a single thing. It's at least four distinct flows, each with different visibility, different ethics, and different practical importance. Conflating them produces bad data and worse stories.
This issue covers three bribery flows and one structural conflict of interest, and whether each is detectable. Findings come in future issues.
The most visible bribe flow is one most people don't realize exists.
Hidden Hand, Votium, Quest, and Paladin are openly operated marketplaces. The product they sell is governance influence. An actor with a stake in the outcome of a veToken vote deposits tokens. Delegates who hold voting power then claim those tokens in exchange for voting in the depositor's preferred direction. The contracts, the amounts, and the recipients are all public.
This is bribery being sold as a product. Explicit pay-to-vote, governed by smart contracts, with claimable rewards.
Volumes are substantial. Across veCRV, veBAL, vlAURA, and vlCVX governance, public bribe markets have moved hundreds of millions of dollars cumulatively.
The industry exists in plain sight, yet most retail readers of crypto media are unaware of it. The corruption narrative ignores it because it doesn't fit the secret-cabal frame.
The second category is OTC stablecoin payments. This is bribery that looks like bribery. A wallet receives stablecoins from a source: a centralized exchange hot wallet, a freshly funded externally-owned account, or an address laundered through a mixer. Then a few hours, days, or weeks later, the recipient votes on a proposal.
The flow is on-chain. The vote is on-chain. The connection between them is inferential.
Suggestive patterns:
Each pattern alone is circumstantial, but combinations start to show potential coordination.
Sloppy actors leave traces. Sophisticated ones spin up fresh source or recipient wallets (or both) with no ties to anything else. We catch the former and acknowledge the latter.
This is the largest category by dollar volume and the hardest to write about responsibly.
DAOs pay people to be part of their ecosystems. They fund research firms that publish governance analysis. They fund delegate platforms that publish rationales for votes. They run ambassador programs, sponsor conferences, and give grants. The recipients of these flows often hold significant voting power, and some of them vote on proposals affecting the entity that pays them.
Structural compensation is a conflict of interest, but it's also how the ecosystem functions. Without paid research and paid delegates, governance participation collapses. Issue 002 showed how thin participation already is.
Some publicly disclosed examples:
These arrangements are openly documented. None of the participants are hiding anything. That is precisely why this category is hard. The line between "infrastructure" and "conflict of interest" is thin.
The bribery that nobody can detect through data is the bribery that probably matters most.
Promises of future positions. Investment commitments in unrelated funds. Conference appearances, speaking fees, advisory titles. Equity in adjacent ventures. Social capital exchange (the "I owe you one" between people who decide many things together). Legal-entity payments routed through jurisdictions that don't disclose.
None of this is on-chain. None of it is detectable through transfer tracing. It is the layer where the most consequential influence lives.
We're building three things, in order. Section 4 is not a problem data can surface, so we won't pretend to.
We don't publish accusations. We publish patterns with context, dates, addresses, and source attributions where available. Readers can draw their own conclusions about intent. When we're wrong, we correct.
For the next issue, we are picking a contested proposal already in the dataset, pulling inflow data for the top voters on each side, and publishing what we see. We don't know if we'll find anything publishable. If we don't, we'll say so.