Brian Reed: Eliza — before we get into anything else, I need to put one number on the table: 21 months.
Eliza Ward: The block-size war.
Brian Reed: Bitcoin, 2015 to 2017, couldn't agree on whether to increase the block size. That is a — let me see if I can make this land — that is a multi-year standoff over one line in a config. A centralized payment processor resolves that before lunch.
Eliza Ward: And that gap — centralized company versus decentralized network — that's actually the whole thing today. Blockchain governance. Why the same architecture that makes these networks resistant to censorship also makes them almost constitutionally unable to change.
Brian Reed: The condo analogy keeps coming back to me — every resident holds a veto, so no arbitrary evictions, but also no fixed elevator. The protection and the paralysis are the same rule.
Eliza Ward: Right — distributed consensus, the mechanism that makes Bitcoin censorship-resistant, it also requires near-unanimous agreement from miners, validators, developers, node operators for any protocol upgrade. No single authority can compel adoption. So is that working as designed, or is it broken?
Brian Reed: Both, actually — and that's what the canonical cases prove. Bitcoin, 2015 to 2017, the block-size debate: two camps, neither could compel the other, and after 21 months the minority just left. Hard-forked into Bitcoin Cash. That's not a resolution — that's a permanent split. Two separate chains running different rule sets from that point forward.
Eliza Ward: And a hard fork — let me just make sure that's landed — that's a backward-incompatible protocol change. If you don't upgrade, you're on a different network than the people who did.
Brian Reed: Exactly — and the DAO hack on Ethereum in 2016 is the other proof. Attacker exploits a code vulnerability, drains millions in ETH. Community votes to reverse the transactions via a hard fork. Majority does it. But a principled minority says — no, code is law, we don't rewrite history. They keep the original chain. That becomes Ethereum Classic.
Eliza Ward: Hold on — so Ethereum Classic isn't a bug. It's the exit option working.
Brian Reed: Right, and there's actually game theory behind why that persists. Biais and colleagues — I mean, the academic framing here is that proof-of-work systems can have multiple stable equilibria. Miners can rationally coordinate on competing branches. So two ledger versions coexisting isn't a failure to reach Nash equilibrium — it sometimes *is* the Nash equilibrium. The fork is the outcome the system was designed to produce.
Eliza Ward: Wait — so if a fork is the rational equilibrium, that means... we built networks structurally inclined to split? Like, picture a developer who's been running a Bitcoin node since 2014. January 2017, they wake up, there are now two competing chains, and the exchange they use has frozen withdrawals while it decides which one to list. That's not abstract governance theory. That's their money, frozen, because the incentive structure made splitting rational.
Brian Reed: And here's what makes it worse — the on-chain governance systems that were supposed to fix this, Tezos, Polkadot, Cardano's Voltaire framework — they relocate the coordination problem, they don't dissolve it. DAOs were meant to automate the whole thing and they just reproduced voter apathy and power concentration in a new wrapper. We'll get to exactly how that happens in a minute.
Eliza Ward: Okay but — relocate is doing a lot of work there. Because Tezos and Polkadot didn't just reskin the problem. They built cryptographic voting, automatic execution, the change gets applied without needing miners to manually upgrade. That's structurally different from a Bitcoin BIP that nobody can compel anyone to adopt.
Brian Reed: Is it though? Because the transparency is real — yes, votes are recorded on-chain, you can audit exactly who voted what. But voting weight tracks token holdings. So the question I can't get past is: a large token holder dominates a Polkadot governance vote the same way a large miner dominated the Bitcoin block-size conversation. The power just... moved.
Eliza Ward: Plutocratic capture instead of developer capture.
Brian Reed: Right — and then DAOs were meant to spread that even further. Automate collective decisions via smart contracts, reduce coordination costs, no whale dominance. Empirically? Most token holders don't vote at all. A small group of large holders controls outcomes. We literally rebuilt the same power structure inside a nominally decentralized system. That's — I mean, the 2016 DAO hack is the earlier version of this. The code executed exactly as written. The community's refusal to accept that outcome is what triggered the crisis. The social layer overrode the technical layer.
Eliza Ward: And that's the legitimacy-versus-speed tradeoff, actually. Broad participation — more legitimate, slower. Narrow fast decisions — responsive, but now you've undermined the decentralization that gave the network credibility in the first place. Wait, is anyone claiming that's solvable at the protocol layer? Because the research I've seen basically... punts. Cardano's Voltaire spec is described as comprehensive but its real-world effectiveness is an open empirical question. That's not a solution, that's a hypothesis.
Brian Reed: No, and I think that's the honest answer — we don't have the data yet to say on-chain governance measurably reduces deadlock versus off-chain. We have architectural arguments. We don't have outcomes.
Eliza Ward: So the fork question is still live. Exit-as-freedom, or evidence the system is structurally broken? I genuinely can't close that one.
Brian Reed: If the problem is ultimately social and political, not technical, then Voltaire, Polkadot's voting mechanism, none of it actually eliminates the deadlock. It relocates who holds the friction. So the real choice isn't better governance design. It's whether you recentralize enough to move fast, and lose the exit rights that made Ethereum Classic meaningful — or you keep the friction and accept that ordinary users, not miners, not developers, absorb the cost of every stall.
Eliza Ward: And I genuinely don't know which one survives contact with something bigger than a block-size argument.
Brian Reed: Neither do I. What would actually change your mind about which one matters more?
Eliza Ward: I'll sit with that.
