How Bitcoin is like American Idol
Everything we buy (with money) we pay for (with energy). There is no such thing as a free lunch.
In this issue:
How Bitcoin is like American Idol (reader submitted)
Have your stake and eat it too (reader submitted)
Ownership is not an afterthought (reader submitted)
How Bitcoin is like American Idol
“You say that the capital costs of proof-of-stake are just as large as the energy costs of proof-of-work, but energy is only one input into production. It is an empirical question if the capital costs of proof-of-stake are larger or smaller than the costs of proof-of-work. It isn’t a tautology. The energy to produce a bad song a good one are equal but their revenues are different. Doesn’t that break your model?” — Adapted from a conversation on Twitter
I quipped yesterday that I was doomed to unsuccessfully explain the opportunity cost of capital forever and I guess the prophecy is true because here we are. 🙂
If you haven’t already you may want to read this post about how Proof of Stake will not save us and this post comparing the externalities of proof-of-stake and proof-of-work more directly. But to summarize very briefly: proof-of-stake systems don’t eliminate energy use they simply shunt it outside the system where it is more difficult to observe but also less efficient at buying security. If you bid $10 for [something] in an open, competitive market the market will spend $10 worth of energy competing over your contract. That isn’t a tautology but it is always true.
The first thing to note is that the value of something has nothing to do with how much it cost to produce. It would be very difficult, expensive and time-consuming to dig a hole to the center of the Earth but no one wants to pay you to do that so the hole itself is worthless no matter how much it costs. The cost to produce [something] determines how much of [something] is produced (for a given price) but it is the buyers that determine how valuable it is.
That’s important because a cryptocurrency network is not a producer of network security, they are the buyer. In the metaphor of songwriting crypto networks are not writing songs but holding songwriting contests. If you are trying to commission a song and you offer a $10 reward, very little energy will be spent on your contest. Maybe your friends will submit songs they already had lying around. But if the reward you offer is $10M, people will fly in from all over the world just to compete in your contest. Lots of energy will be spent — specifically ~$10M worth.
Put another way when Lizzo sells a song for $10M she doesn’t need to spend $10M worth of energy to produce that song. But if record labels announce they plan to spend $10M buying a hit song and anyone can compete they do cause $10M worth of energy to be spent by people competing to be (or help create) the next Lizzo.
Cryptocurrency networks have to be open to be decentralized — that means anyone can compete to provide validation services to the network. That means the market for validation services will be competitive — validators that spend more of their revenue expanding will eventually push out less efficient competitors until the validators are collectively spending their entire marginal revenue competing over the contract.
In proof-of-work networks this competition is simple and straightforward. The network is just holding an auction to buy hashrate from the cheapest bidder. Miners are therefore competing to buy computation power and electricity at the cheapest possible bulk rates. The ones who can buy electricity and computers more cheaply can expand by spending their extra profits on more electricity and computers. That growth makes the market even more competitive and pushes out miners who have higher costs. The network is constantly seeking an equilibrium where the cost to mine a Bitcoin is basically the same as the cost to buy one on the open market.
In proof-of-stake the system is more difficult to reason about but the mechanics are the same. The network is holding an auction to buy capital from the cheapest providers. So validators are competing over access to cheap capital — but they will also compete in other ways.
One example we’ve talked about several times is Miner Extractable Value, the extra revenue that validators can make by selling transaction ordering services outside the mechanics of the network. If a proof-of-stake validator is skilled at extracting more MEV from the privilege of creating a block they can afford to pay more for capital to stake and still turn a profit. So validators will happily pay MEV specialists to spend energy maximizing the revenue potential of a block.
Another example is long-range attacks. Maybe validators will spend lots of money/energy buying or stealing old keys and searching for a snapshot in the history of the chain where those keys represent a majority share of the stake and hence control of the system. Most of that searching wouldn’t result in any profit but even a remote possibility of seizing control of the system would motivate a lot of energy spent checking just to be sure.
A third possibility is the competition may end up being social. Perhaps validators will spend money (and implicitly energy) traveling to conferences and to congress and schmoozing with industry leaders and politicians to create barriers to becoming a validator that prevent competition. If only validators that can satisfy the SEC’s reporting requirements are the ones who can cultivate a direct relationship with the SEC (which seems to be the SEC’s stance) then cultivating a relationship with the SEC is implicitly part of the cost of being a validator.
I don’t know exactly what expenses validators will end up incurring when they compete with each other but I do know they will keep competing until they can no longer afford to compete any harder — and at that point the total cost of validation will be equal to the total reward the network is offering to validators. Everything the network spends on validating will be spent by validators in the competition. And anything validators buy has an energy cost.
There is no such thing as a free lunch.
Have your stake and eat it too
"You say that capital used for proof of stake is locked and inaccessible, but that’s not actually true. Protocols like LIDO and EignLayer allow staked capital to be redeployed and reused." — [adapted from a conversation on the OAK discord]
LIDO and EignLayer are very clever and they legitimately do improve the network and staking so I am not casting shade. But making stake more flexible and more liquid does not change the economic reality of how proof of stake actually works. For the "stake" in proof of stake to defend the network, it has to have value. For it to have value, someone has to be willing to trade something of value for it. It is the wealth of the person who gives something up that actually secures the network.
So if you take stETH (the LIDO token) and use it to secure a loan, your wealth is no longer securing the network. You moved your wealth out of the network and into the loan. It is the person who gave you a loan (and accepted stETH as the collateral) whose wealth is now securing the network. If everyone everywhere is trying to spend their staked ETH, the price of staked ETH will be zero and the network will be undefended. Someone has to want to hold staked ETH for the system to work.
EignLayer is a system for letting capital stakers participate fluidly in multiple different staking systems — which is cool! But that doesn’t mean there is no opportunity cost of capital. Being able to re-use the same capital to secure multiple networks doesn’t make your network more secure in the same way that your business wouldn’t be more secure if the security guard got a side job protecting the business next door. If anything it raises the potential rewards for an attacker to bribe the guard so it might reduce your business/network security.
From the perspective of potential stakers LIDO potentially lowers switching costs and Eigenlayer potentially raises revenues, which means the rewards for staking are higher and the total amount of capital staked will be higher too. Since the market for validation must be competitive to be decentralized, validators will eventually spend all of those additional rewards competing with each other. Greater rewards will translate into higher total energy use by the system.
There is no such thing as a free lunch.
Ownership is not an afterthought
One of the most common objections I get to discussions about the capital cost of proof-of-stake is a variation on the idea that money is just a token and so as long as proof-of-stake doesn’t consume any physical resources then by definition no "real" wealth can have been lost. Dan Robinson famously compared proof-of-work to proof-of-stake by imagining sacrificing cows to a volcano:
The idea is basically this, society’s wealth is a big pile of stuff: goats, factories, substack newsletters, etc. We have a bunch of tokens (money, deeds, etc) that represent who owns the stuff but that’s just paperwork. The wealth is the stuff itself. If we do something that moves all the tokens around that doesn’t change the big pile of stuff then by definition we not have actually consumed any "real" wealth. It was just some fancy bookkeeping.
This line of reasoning falls apart on careful examination. Any voluntary trade between two parties increases total wealth even though nothing has changed except the mapping of ownership. Imagine if someone offered to pay you $1 to swap everything you own for something of equal market value chosen by them. If wealth was just a pile of stuff and ownership was just a bookkeeping convention you would be happy to make that trade — but of course in reality no one would ever take a deal like that. Your wealth is not just a function of the raw resale value of your physical possessions but of the specific bundling of things that you own and the value that you as the owner derive from them. Attempting to define wealth separately from the system of ownership doesn’t make sense.
It also doesn’t make sense because something doesn’t need to physically exist to have material economic value. Anything that is useful is valuable and anything that is valuable is useful — physically existing is not a requirement. Imagine a world where Google (the corporate entity) was erased but all the software libraries, engineers and data centers still existed. The world would be less wealthy even though nothing was destroyed except the knowledge of ownership. Capital costs can seem less meaningful because they are counterfactual, so they are harder to see or touch or measure — but they are just as real.
There is no such thing as a free lunch.
Other things happening right now:
Wharton economist Jeremy Siegel has some harsh words for the FED:
American investor Stanley Druckenmiller had even harsher words. Druckenmiller famously ran a fund for thirty years averaging ~30% annual returns without ever having a money losing year. When asked for advice on how to approach the economy he replied, "I brought some Cyanide pills if you want." 😬