The PoW vs. PoS debate

A debate between Lyn Alden and Justin Drake on the foundational trade-offs between proof of work and proof of stake, hosted by Ryan Sean Adams and David Hoffman on the Bankless podcast. This video dives deep into economic security, 51% attack recovery, and the philosophical differences between commodity and equity-based money systems.

This transcript is an accessible copy of the original video transcript (opens in a new tab) published by Bankless. It has been lightly edited for readability.

Introduction (0:00)

Ryan Sean Adams: Welcome to Bankless, where we explore the frontier of internet money and internet finance. This is Ryan Sean Adams, I'm here with David Hoffman, and we're here to help you become more bankless. Guys, you are in for a treat today. We've got two experts on proof of stake and proof of work, and they are debating which is better for creating a global money system — a crypto money system — proof of work or proof of stake. Lyn Alden and Justin Drake on the episode today. We're going to talk about a few things. Number one: which provides the most economic security, proof of work or proof of stake? Number two: which provides the best deterrence and recoverability in the face of the dreaded 51% attack? Number three: which minimizes governance power for the elites — which is more "power to the people"? Number four: which is economically the fairest system to participate in? And number five — this is personally my favorite — this conversation around proof of work being more commodity money and proof of stake being more equity money.

David Hoffman: The first half of the episode just wrote itself. I seriously debated going back and making a bag of popcorn because as hosts we didn't have to do a damn thing. Lyn and Justin just took it and took this conversation in their own direction. Then we started directing the conversation in the second half, summarizing and digesting things that were said. I also really enjoyed the closing statements. Overall, I'm just very happy that in this very tribal world of crypto, we have people like Lyn Alden and Justin Drake who can come to a podcast and just talk about things without the yelling you'd find on Twitter Spaces — a very respectful conversation.

Ryan Sean Adams: Imagine having an adult conversation about crypto. Fantastic points by both sides, well articulated. I want to introduce you to repeat Bankless podcast guest, founder of Lyn Alden Investment Strategy — Lyn Alden. She's a leading expert in macro markets. She's been a proponent of Bitcoin for a long time and generally believes that proof of work is more suited to producing a new global money system than proof of stake. She also wrote an article on this topic that was widely read, very well circulated, and well argued. On the other side, we have Justin Drake, a researcher at the Ethereum Foundation and the pioneer of "ETH as ultrasound money." He believes proof of stake produces an asset with the monetary properties the 2020s will need. Justin, how are you doing?

Justin Drake: Doing great, thanks again for having me, guys.

Framing the debate: monetary premium (7:30)

Ryan Sean Adams: I want to start with the big question. I think we're trying to answer the meta question of whether proof of work or proof of stake is most amenable to the accrual of monetary premium — basically which of those two consensus protocols is most amenable to making a cryptocurrency a money. Justin, do you agree this is the correct framing?

Justin Drake: Yeah, I think that is the big question and it kind of boils down to what the consensus mechanism is all about — it's about security. We can look at a consensus mechanism through the lens of economic security, governance security, physical security, quantum security, and even maybe "meme security." Monetary premium to a large extent is about having the best memes, and I believe proof of stake has the potential for much better memes than proof of work when it comes to looking at cash flows.

Ryan Sean Adams: Lyn, do you think this is the right framing?

Lyn Alden: I think that's a good way to phrase it. In the current system, we have proof-of-work and proof-of-stake assets essentially. Equities — because you own a stake in that company — are a proof of stake. Your stake allows you to exercise some degree of control over that company. And we have normal commodities that are proof-of-work assets more or less. What we've seen over history is that either asset can acquire some degree of monetary premium. The question becomes what is most suited to a monetary premium and what is able to acquire and hold it for the very long run.

Justin Drake: I want to interject and argue why this analogy with equity and proof of stake is maybe a bad analogy. In the context of equity, you have the right to vote on anything — you can completely change the rules of the company. In proof-of-stake consensus, the consensus participants can't just arbitrarily change the rules. They can't say "we're just going to give ourselves a thousand ETH each" — that would be an invalid transaction. Ultimately the consensus participants are beholden to the community, and that is a huge difference relative to equity.

Commodity money vs. equity money (10:06)

Ryan Sean Adams: Lyn, do you want to keep going with that conversation about proof of work as commodity and proof of stake as equity?

Lyn Alden: I agree there are different types of proof-of-stake systems — some give you more control than others. If I were to focus on defining proof of work as commodity money, I'd go so far as to say that if you have a proof-of-work system with difficulty bombs or very large blocks or nodes that are hard to run, that starts to have equity-like properties too. When it comes to a commodity, producers have no influence over the properties of copper, for example. Going back to the proof-of-stake to equity analogy — shareholders in a corporation can make changes, but they can't make changes that violate the law. In proof of stake, the people who hold the capital decide which transactions are processed — that's what their stake is rather limited to.

Justin Drake: The consensus participants can try to do two things. One: a hard fork — making invalid transactions somehow valid. That violates the "law" set by users. Two: restrict which transactions are processed — what we call a soft fork, which is basically censorship. The question becomes: which system is least liable to censorship attacks? The social layer needs to intervene, and one of the massive advantages of proof of stake is that the social layer has the tools to intervene when there is censorship by consensus participants, whereas proof of work does not.

David Hoffman: Is it the case, Lyn, that you believe any proof-of-stake monetary premium is more like equity and less like a commodity? Or is there granularity there?

Lyn Alden: I think there is some degree of spectrum, but for the most part it's a rather one-directional street. There are multiple ways to become an equity but very few ways to become a commodity. Commodities by their nature are somewhat rare and immutable. The ways to create a true commodity in the digital realm are extremely limited.

51% attacks: deterrence and recovery (15:30)

Justin Drake: I want to compare proof of work and proof of stake on which is least liable to censorship attacks. One of the massive advantages of proof of stake is that the social layer has tools to intervene. In proof of work, if a 51% attacker comes in, they collect every single piece of reward, every single piece of issuance, every transaction fee. The honest miners turn off because it doesn't make sense to spend electricity without income. Then they want to liquidate, selling their hardware. The attacker can buy this hardware for pennies on the dollar and reinforce themselves. Even if the community manages a counter-attack, it would take months if not years to organize.

In proof of stake, there's a very straightforward mitigation: remove the attacker from the validator set through forced ejection. In Ethereum, if you want to double the amount of staked ETH, it takes roughly 200 days. So kicking out the attacker buys you 200 days. The community could also destroy all accrued rewards, do partial slashing, or even destroy the whole stake of the attacker. If there's 10 million staked ETH and you need another 10 million to attack, and every attack costs you 10 million — with only 120 million ETH in circulation, the attack can only happen 11 times. It's almost black and white — proof of stake is clearly superior for healing attacks.

Lyn Alden: The difference between mining and staking as sources of control is that staking requires little or no entropy. Once you hold the power, you accrue more power. In mining — whether physical commodities or digital — it's a very capital-intensive business. You have to constantly put in fresh capital to maintain your rewards. It doesn't accrue a lot of value for miners other than on the margins.

When it comes to security, it's not just about 51% attacks — it's also about bugs. Proof of stake is inherently far more complex. I'd be far more concerned about bugs than about the largest chains being 51% attacked. In a world where everything works perfectly, proof of stake has a higher cost to attack. But because proof of stake is inherently more complex, there are greater attack surfaces.

For recovering from 51% attacks: the only way in proof of stake is to soft-fork and take people's capital. If you phrase it as taking the attacker's capital, that sounds fine. But the attacker could be a regulated custodian — you're taking capital from people who had nothing to do with the attack.

Complexity, bugs, and client diversity (30:35)

Justin Drake: Complexity is not necessarily evil. Humanity thrives on it. In terms of proof-of-stake complexity — yes, it is more complex, at least 10x, maybe 100x. But it's complexity that at this point we have tamed. We have five different clients that have implemented the protocol, each with small teams of five to ten people. Client diversity enables the community to buy insurance against bugs.

On the topic of being continuously online — what we call weak subjectivity — if you've been offline, you sync up with a semi-trusted checkpoint. Bitcoin has these checkpoints all over the place. Literally in the Bitcoin Core C++ code, there are about 12 checkpoints. If there was an attacker that rewrote history from genesis, that chain wouldn't be valid because developers put subjective checkpoints in the codebase. When you download the software, you're trusting the code, trusting GitHub, trusting your operating system. There are all sorts of places where you're effectively trusting the initial checkpoint.

Lyn Alden: I would describe proof of work as inherently renewative. Every dollar that goes into the space has a set of decisions with it — what hardware to buy, what scale, what energy source. This requires an ongoing series of good decisions to maintain participation. In a pre-mined proof-of-stake system, those who bought in early have a permanent structural advantage. Proof of work requires an ongoing series of good decisions to maintain your ability to participate.

Looking at the direction of proof of work long-term, you'll see increasing integration between miners and energy producers. The electrical grid naturally has to overproduce electricity. Proof-of-work systems are a really good load balancer for stranded energy. The cheapest source of electricity is virtually zero-cost stranded energy. Right now the biggest Bitcoin mining facility in the world has about 1% of total hash rate — it's already structurally decentralized.

Fairness and the "rich get richer" argument (40:20)

Justin Drake: In proof of stake, you have perfect fairness — you put in one unit of capital and get the exact same amount of rewards no matter how big or small. In proof of work, the big fish have unfair advantages. If you're a retail miner buying one rig, you're overpaying by 2x, 3x, 4x, 5x relative to professional miners buying in bulk. Because of Moore's law, attackers who want to attack have an advantage — they can pick the latest, most bleeding-edge hardware. For them, the price of electricity is largely irrelevant, because a 51% attack only needs to run for one day or one week. 99% of the cost is in the rigs and infrastructure, not the electricity.

The fact that proof of work is inherently unfair creates advantages for an attacker. On slashing custodians — one of the tools the community has is forced ejection. There's no penalty — you're just removing them from being consensus participants. In Ethereum, that buys you 200 days. You could freeze funds for five years, do partial slashing, or a full slash. There's a wide range of tools completely unavailable in proof of work.

Lyn Alden: I would rephrase "inherently unfair" as "inherently renewative." Every dollar going in carries a set of decisions. Rather than early participants having a permanent advantage, proof of work requires an ongoing series of good decisions to maintain participation. In a proof-of-stake system where you acquire stake and it inherently provides income, you've acquired a permanent stake with no ongoing input costs. It consolidates over time into fewer owners holding larger shares.

In proof of work, you're essentially renting your ability to govern the system. Your machines degrade, your energy source could become less efficient, your hardware becomes less cutting edge. It's a constant series of decisions rather than a system that inherently benefits people who bought in first. No matter how much Bitcoin Michael Saylor has, he has zero control over what transactions are processed.

Proof of work as "proof of stake with extra steps" (50:16)

Justin Drake: I don't know anyone — and I've been in this space for almost a decade — who mines Bitcoin as an individual at this point. It's all industrial. On the other hand, with proof of stake, everyone on this call — maybe except you, Lyn — are staking as individuals. The diversity on proof of stake is much, much larger. The barrier to entry is much lower — you basically just need a computer running 24/7.

What happens with Bitcoin mining is that you have two classes of people. Retail miners might not realize they're getting wrecked — in dollar terms they're profitable because Bitcoin's price went up, but in Bitcoin terms they put in 10 bitcoins and got three back. Professional miners are buying hash rate futures, energy futures, making their own chips — the minimum investment is at least $10 million for a three-nanometer chip.

On the "rich get richer" argument in proof of stake: I see it as wealth preservation. The rich stay as rich as they were. When staking, you're paying opportunity cost — roughly 3% — and getting compensated. Net, you're doing wealth preservation. For professional miners, they hedge risk with financial products — electricity futures, hash rate futures — locking in their profit. Really, proof of work is just proof of stake with extra steps. The risk can be removed with financial products, and at the end of the day you have essentially the same financial product with the same risks and returns.

Lyn Alden: One of the big challenges of wealth concentration historically is that wealth begets more wealth. In a proof-of-stake system, if you acquire stake, it inherently provides income forever with no ongoing input costs. In proof of work, you're renting your ability to govern the system. Your machines degrade, your energy sources change, your hardware becomes less cutting edge. It's a constant series of decisions, not a system that inherently benefits those who bought in first.

On client diversity — at least last I checked, the biggest client has about 84% of Ethereum clients. You have some degree of diversification, but it's also partial illusion. In practice, one of the key things that protects against bugs is having the simplest possible code base.

NIST, quantum, and long-term fundamentals (55:04)

Justin Drake: If we want to be the internet of money, we need a time scale on the order of the internet itself — decades if not centuries. Complexity is something that has a half-life of maybe one year — it can be tamed and hardened. Over time, as lots of value is secured, systems can be trusted. The bugs will be fixed, the system will harden. What we should be looking at is long-term fundamentals. It turns out that proof of work is going to be completely disrupted by quantum computing in the next 20 to 30 years. Yes, there are short-term concerns with complexity, but if you zoom out and focus on fundamentals, there are fundamental reasons to be bearish on proof of work.

The commodity money historical argument (1:00:34)

David Hoffman: Lyn, you've kind of taken the lead as a macro commentator focusing on commodity markets. Bitcoiners tend to focus on commodity properties and how Bitcoin consumes energy as a commodity. Is this partly a moral argument — that commodity money is just a better-suited money to the world?

Lyn Alden: I approach this mostly from a macro perspective. If you look back in history, what you're giving up when you go from commodity money to stake money is that it shifts towards governance. Prior to the early 1900s, you had commodity money — proof of work in the form of gold — with layers of stake on top of it. As you shifted towards purely fiat money, you essentially shifted to proof of stake. The Federal Reserve is basically a proof-of-stake system but not on a blockchain. You have 12 regional reserve banks owned by commercial banks — their representation is based on their capital, their stake. The Federal Open Market Committee has a hodgepodge mix of federally appointed and bank-appointed officials determining policy.

Over time, that tends towards centralization. As debt builds up, the system gets increasingly captured by the government because they have to monetize debts and bail out the system. With the invention of Bitcoin, you have the reintroduction of commodity money, allowing people to opt into a different system.

David Hoffman: Baked into that argument — doesn't proof of stake represent the fiat system and proof of work represent the commodity money like gold? How much of this is a technical objective argument versus a subjective argument about fairness?

Lyn Alden: I personally approach it from a tactical risk perspective. What is the system least likely to have tail risks? What is optimized for what it's trying to do? When it comes to the fairness argument, partially that gets mixed up with different things — proof of work versus proof of stake being one aspect, and the scarcity of tokens being another.

The scarcity engine vs. the liquidity engine (1:10:31)

Justin Drake: Proof of work and proof of stake from a meme perspective are almost the exact opposite. With proof of stake, you have what I call a scarcity engine — a mechanism to turn liquid ETH into frozen ETH used as collateral. My projections say roughly half of all ETH supply will eventually be staked. For proof of work, it's the opposite — issuance and transaction fees that miners earn need to be sold to cover electricity and hardware expenses. You've created a liquidity engine, constantly market-dumping. If you ask which is the best system to create money for the internet — the one constantly dumping or the one encouraging people to hold — I think it's obvious which is more valuable from a meme perspective.

Lyn Alden: The more a system tries to do, the worse it's going to be at any one thing. As Ethereum has tried to harden its monetary policy, it's arguably lost market share in DeFi — from 97% of total value locked in late 2020 to 55% now. So far, Bitcoin doesn't really have competition from other systems. The brief periods of competition — Dogecoin memes, Bitcoin Cash — fade over time. One world is based on governance and making your ecosystem attractive. The other comes down to which is the hardest — what is the best at being money.

Governance, immutability, and self-sovereignty (1:15:36)

Lyn Alden: At the minimum, a proof-of-stake system gives the units' holders more say over which transactions are processed and which are censored. But who gets to change the protocol and how — that goes outside the proof-of-work vs. proof-of-stake debate. You can have proof-of-work systems that look more equity-like — with difficulty bombs or super large blocks so people can't run their own node.

The meme of Bitcoin is self-sovereignty. You run your own node, nodes are small, it doesn't take a lot of resources even ten years from now. Any updates are opt-in — you can retain your existing node. That maximizes self-sovereignty. If you put in difficulty bombs, the developers are going in a certain direction and you don't get to opt in.

David Hoffman: Doesn't the separation of the ASIC from BTC — requiring constant further investment — rely on the fact that returns on capital are less significant in proof of work than in proof of stake?

Lyn Alden: With commodity monies, producers rarely have much control. Gold has a very high stock-to-flow ratio — gold miners have virtually no control over the system. That's been true in the Bitcoin ecosystem. During the block size wars, the majority of miners favored the block increase and still couldn't get it through. The node operators rejected it. That combination of division of powers — where you can just run a node and become self-sovereign, and separately mining gives you temporary ability to order transactions but no permanent allocation — is what maintains a decentralized system.

The "rich get richer" rebuttal (1:25:13)

Justin Drake: On the "rich get richer" in proof of stake — I see it as wealth preservation, not wealth multiplication. When staking, you're paying opportunity cost of roughly 3% and getting compensated. Net, you're doing wealth preservation. Professional miners do the same thing — they hedge risk with futures, locking in returns that roughly match opportunity cost. Really, proof of work is just proof of stake with extra steps. The risk can be removed with financial products, and at the end of the day you have essentially the same financial product.

Lyn Alden: What makes a commodity system decentralized is the combination of small nodes and separate mining. No matter how much Bitcoin someone holds, they have zero control over transaction ordering. That's the division of powers. If you're going to approximate immutability in the digital realm as much as possible, you're designing a system that is very decentralized, very hard to force changes on, with either no changes or opt-in changes. Anything that deviates from that model — ongoing forced buy-in from a small team of developers, development hubs, foundations — is inherently more equity-like regardless of the consensus mechanism.

Closing arguments (1:30:42)

Ryan Sean Adams: This has been a fantastic conversation. This has probably been the best proof-of-stake versus proof-of-work conversation in history. Closing argument time. Justin, why is proof of stake the best way to create a crypto money?

Justin Drake: When we want to build money, what we want is monetary premium. You need a Schelling point — a coordination point to focus attention on one particular asset. We need to compare proof of work and proof of stake. In terms of economic security: for every $100 of economic security, proof of stake pays about $5 per year through issuance — roughly a 5% APR. For proof of work, the maintenance cost is roughly $100 per year. That's a 20x improvement in economic efficiency. The Beacon Chain has $32 billion of economic security. Bitcoin has about $10 billion — roughly $50 per terahash/second times 200 million terahash/second.

This efficiency unlocks the possibility of a decreasing supply — transaction fees when burnt can exceed issuance. That's a distinguishing factor from a scarcity standpoint. Qualitatively, the really big one is that we're empowering the community — the social consensus — to act as a backstop if consensus participants abuse their power. In the context of Bitcoin, I believe the community doesn't have this backstop power. To summarize: proof of stake stands out because it has much bigger security, both quantitatively and qualitatively, and that unlocks memes backed by real fundamentals.

Ryan Sean Adams: Lyn, closing arguments — why is proof of work the best way to create a crypto money?

Lyn Alden: Historically, when we look at things that acquire monetary premium, it's the things that are the hardest — the most immutable, where technology cannot come in and find a better one or increase the supply. You want an asset where the vast majority of its value is the monetary premium and very little is the utility premium. Gold is mostly held for its monetary premium; oil is entirely for utility; silver is somewhere in the middle.

If you were designing a blockchain to maximize monetary properties, you would make one whose almost entire purpose is to be money — sacrificing everything else. The perfect fork would just be a fork, not a fork and a spoon and a knife combined. The closest thing to perfect money is something extraordinarily simple. History shows complexity doesn't just work itself out. The US Air Force used eight-inch floppy disks as part of their nuclear launch process until three years ago. They kept it extremely simple, updated very slowly, disconnected from everything else. When it comes to the most critical things, we move very slowly and keep things as simple as possible.

When it comes to money suitable for putting your corporate treasury in, allocating part of your endowment, or holding 10% or more of your net worth — or managing sovereign reserves representing decades of accumulated trade surpluses — historically the best has been gold. Now we have new competitors. You'd select the most decentralized, the most immutable, the one that doesn't sacrifice any of those characteristics. That doesn't mean other systems aren't valuable — just because gold has value doesn't mean Tesla stock doesn't. They're different things doing different things. The perfect money is willing to sacrifice just about everything else in order to have the perfect attributes of money — whose main purpose is to be held and occasionally transacted with, compared to something trying to be a Swiss Army knife.

Wrap-up (1:40:14)

Ryan Sean Adams: I want to once again thank Lyn Alden and Justin Drake for joining us and having the best debate I've ever heard about proof of work versus proof of stake. Also a special thanks to Lyn for coming into what some might perceive as the "away team."

David Hoffman: Action items for you today, Bankless listeners — first, go listen to that episode again. I think it was that good. Lyn Alden has her proof-of-stake article which we referenced, and Justin Drake rattled off some numbers on ETH versus Bitcoin economic security — we'll get those in the show notes. Risks and disclaimers: none of this has been financial advice. Crypto is risky, DeFi is risky, you could definitely lose what you put in. But we are headed west. This is the frontier.