Bitcoin Isn't Perfect ⚖️
Three Things #74: June 25, 2023
I’m first and foremost a Bitcoiner. I bought bitcoin first, used bitcoin first, studied Bitcoin first, and attended Bitcoin events first. Nothing I’ve done since including working on Ethereum or Spacemesh, my current project, changes that. Ethereum and Spacemesh are both downstream of Bitcoin both technically and socially, and are unique, individual manifestations of many of the same ideas and values behind Bitcoin. In the same way that the United States is downstream of classical Greece and Rome culturally and technically (having borrowed and refined the instruments and philosophy of democratic governance), these projects have taken many of the things that make Bitcoin great and improved upon them. I also feel that Bitcoin is still the greatest and most important project in the world, and when I’m introducing people to the ideas behind blockchain and cryptocurrency I always start with Bitcoin and encourage them to buy and hold only bitcoin in the beginning.
I try to attend as many Bitcoin events as I can. I was at dinner a few nights ago with a number of Bitcoin core contributors who asked me some thoughtful but hard questions about my work. It was the best conversation I’ve had in quite a while. The main questions were, if you consider yourself a Bitcoiner and if you care so much about Bitcoin, why not build on Bitcoin? Why build another chain? Does the world really need another chain? It gave me pause, and I’ve been turning the conversation over in my mind ever since.
To be clear I think most blockchains shouldn’t exist. Most are garbage, copy-paste scam chains that at best pay lip service to the cypherpunk ideas and values behind Bitcoin. But neither do I think that there should only ever be a single chain.
Here are the reasons I think it makes sense to launch a new chain in 2023.
Thing #1: Bitcoin Isn’t Perfect
Bitcoin is awesome. I use it all the time, rely on it quite a bit, and recommend it to others daily. What’s more, Bitcoin isn’t static. It’s improving all the time. There have been a number of recent upgrades and new technologies have been added to Bitcoin, and in general the pace of development has improved recently. I see more creative people and more interesting projects that are building on Bitcoin all the time.
Nevertheless, Bitcoin is far from perfect in its current form. It’s good for what it does—being the best money the world has ever seen—but it’s not particularly good for many other things. In fact, there’s a tension here: making Bitcoin good for other things—for example, by making the VM more expressive—might make it less good at being money. (Ethereum, which has a much more expressive VM, is much less good at being money.)
This is the main reason I don’t believe the future will or should be a single chain to rule them all. It’s a bit like suggesting that some particular country, say, the Kingdom of Elbonia has already perfected governance, so why do we need other states and countries? They should all become vassal states and Elbonia should become the seat of an empire. It shouldn’t take Bitcoiners long to work out why this would be a bad idea.
There’s no such thing as a perfect blockchain just as there’s no such thing as perfect governance and there’s no such thing as a perfect country. The world needs many approaches that optimize for different things. Different local groups of people have different needs and different values and there’s no way that one chain, one community, or one platform could account for all that diversity or satisfy all of their needs and preferences.
In blockchain there are approaches that are fundamentally different but equally valid to Bitcoin and there is real innovation happening outside Bitcoin (something that Bitcoin maximalists of the close-minded variety refuse to acknowledge). Ethereum and Solana, to name two examples, have both made interesting, innovative design choices and I believe this exploration should continue. Spacemesh intends to do the same. The only way we figure out what works is to enable lots and lots of experimentation—and while there is more experimentation on Bitcoin today than previously, the pace still isn’t fast enough because the cost of breaking Bitcoin is too high, as it should be. We need more and faster experimentation and novel chains are the best way to do this. We’re already beginning to see ideas developed on Ethereum, like NFTs and ERC-20 tokens, be adopted on Bitcoin as well, and hopefully we’ll eventually see Bitcoin adopt other useful ideas like rollups as well. Bitcoin should act like Google to the many startup chains of the world: it should foster radical innovation elsewhere and “acquire” the best ideas from the best chains as a way of avoiding innovator’s dilemma and overcoming gradual ossification.
Three particular areas of innovation come to mind. The first is decentralization. It’s pretty easy to run a Bitcoin node. Many people do this today and indeed it’s one of the most unique, important, and misunderstood aspects of Bitcoin culture. But it’s not so easy to mine on Bitcoin. You need powerful hardware, and even if you have a few spare ASICs lying around you still need to join a mining pool, which is a centralization vector.
We’re decentralization maximalists at Spacemesh, and our target is to get one million independent mining nodes on the network. Three mining pools make up 62% of Bitcoin hash power today, and six make up around 90%. We’re talking about five orders of magnitude greater degree of independent participation in consensus. It will take time to achieve that goal, but since you don’t need to join a pool to mine from home in Spacemesh the goal is ultimately feasible, and even if we fall far short of it we might still be far ahead of Bitcoin, not to mention any other network, in this respect. We’re living in a world of increasingly unclear, hostile regulation and it’s more important than ever that networks achieve a meaningful degree of decentralization in mining and consensus as well as in other ways.
Could this be done on Bitcoin? Conceivably, yes. One could invent a new type of ASIC or come up with some other clever way to scavenge hashpower from toaster ovens and microwaves, and additionally develop a new mining pool management platform that makes governance and economics of mining pools more transparent (such a thing is indeed in the works). Or one could try to convince Bitcoiners to switch to a consensus mechanism that’s friendlier to home miners. But, hard as it’s been to develop Spacemesh, it’s literally 100x easier to design the entire core protocol around such a thing in the first place as we have on Spacemesh. This is a good example of a thing that’s very, very difficult to change in a mature network.
The second area of innovation is VM design. Ethereum and other smart contract platforms have shown us the power of a flexible, Turing complete virtual machine and a reasonably robust smart contract programming language and tooling. The Ethereum base layer has achieved “escape velocity,” which means that it’s good enough to host just about any application you can imagine, and good enough to scale ad infinitum via layer two technologies like rollups. This may be true of Bitcoin someday as well. Segwit and Taproot have helped a lot already, and covenants, if they ever land, will go even further. But even this powerful combination can’t come close to enabling what EVM and other, similarly advanced, flexible VMs can achieve.
Bitcoiners will likely respond that all of that state and all of that compute doesn’t belong on chain in the first place. They’re directionally right but applications like the big three, most especially stable coins, have given the lie to this argument. We could try to retrofit Bitcoin script to support rollups, or we could just build better, more expressive platforms that enable more innovation more easily. I for one am not terribly excited about the prospect of engaging in a drawn out, bitter political war with other Bitcoiners about this sort of thing, nor am I convinced that Bitcoin should ever have such an expressive VM.
The third area of innovation I leave as the subject of the next thing.
Thing #2: Economics
If Bitcoin fails the most likely reason would be economics. To be clear, Satoshi was truly visionary and did a remarkably good job with Bitcoin’s economics considering the novelty of the idea and how little information he had at the time. Exponential decay with issuance cut in half every four years has worked pretty well for Bitcoin up to now and will probably continue to work well for a while, but it may not work well forever. The most obvious question is, what happens to Bitcoin after the mining subsidy vanishes? Research suggests that the network could become quite unstable at that point and that forks might become more common as the incentive structure for miners changes. Fees briefly spiked as high as 40% of the subsidy this year as Ordinals took off, compared to the usual rate of around 1-2%, but they’ve since settled down again. If Bitcoin is going to work economically when the subsidy runs out in a decade or two the network needs to settle on a more sustainable fee model, or else issuance needs to be increased, yet another topic that’s out of bounds for most Bitcoiners.
One nice property of proof of work mining is that most miners must regularly sell most of the coins they mine in order to pay operational costs, most notably electricity. This keeps the number of circulating coins quite high and increases velocity (as opposed to proof of stake which has exactly the opposite effect) which in turn increases price and liquidity. This is important since bitcoin is disinflationary, meaning users are incentivized to hold bitcoin rather than spend it since they expect a higher future price. Other things equal, as the subsidy falls to zero, the percentage of coins being recirculated by miners in this fashion also falls, which could have a negative impact on price and liquidity.
In general I feel that the decay in Bitcoin’s issuance is a bit too aggressive. Four years is a pretty short half life. It means that 99% of all bitcoin will be issued by about 2030. It’s worked well enough in prior years since the price of bitcoin stayed ahead of the halving of issuance such that, on average, miner revenue in fiat terms didn’t decrease. This matters because miners’ expenses are denominated in fiat. But it’s unclear how much longer this trend can continue since the bitcoin price can’t keep doubling indefinitely. I see a lot of value in running different economic experiments in blockchains that are otherwise similar to Bitcoin so that the market can determine which economic model people prefer most. Spacemesh uses a similar exponential decay model to Bitcoin but smoothed out and stretched out over a period nearly 10x as long.
Mining or otherwise participating in consensus is also too hard for everyday users on networks including Bitcoin. Mining hardware costs thousands of dollars, it’s too noisy and power hungry and generates too much heat to keep at home, and in any case you have to join a mining pool in order to mine. As discussed above, there are downsides to this model: mining pools lead to centralization of consensus power (in the extreme, a cartel could censor certain transactions) and act as intermediaries between miners and the protocol. They must be trusted to calculate rewards and distribute them fairly to home miners, with little to no transparency into how they do this or accountability if they get it wrong. I see great value in a network that allows home miners to earn coins profitably using hardware they already own, without running an energy intensive algorithm like proof of work, and without the need to join a mining pool. There are also a number of novel economic ideas and technologies here that could someday find a home in the Bitcoin ecosystem, but I don’t think it’s realistic to develop or deploy them on Bitcoin today.
Thing #3: Public Good
I consider myself very fortunate as a software developer to be able to work exclusively on permissively licensed free and open source software. I worked on proprietary software for many years and I resented the fact that when I left at the end of that process, after pouring my life energy into a project for years, I couldn’t take a single line of code with me. I’ll never do that again.
But open source is about more than my own selfish feelings. More broadly it’s about public good. Every important software engineering project today relies upon lots of open source libraries and code samples. We quite literally couldn’t be where we are today as a society without open source software; we stand on the shoulders of giants. To me, the most beautiful and most impactful thing about open source is that, even if a particular project, company, or application fails, its legacy lives on in the public domain. Who knows whether someone, somewhere, someday might find that library, code snippet, or document that you published useful and build something on top of it?
At the core of Spacemesh is a set of novel primitives for cryptography and distributed systems, things like proofs of spacetime, proofs of elapsed time, and the tortoise and hare mechanisms. We believe that these have immediate applications in Spacemesh and will enable the project to make layer one mining and consensus participation possible for the first time for millions of home miners. I feel strongly that these ideas, these technologies and primitives, should exist. They’re powerful ideas with enormous potential in Spacemesh and beyond and it’s essential that we complete their design and put them into production to prove that they work. They may one day find use cases far beyond Spacemesh, maybe even beyond blockchain and cryptocurrency. Who knows? They may even be used in the Bitcoin ecosystem one day. Proof of space time is designed as a hybrid of proof of work and proof of stake, as a greener mechanism that maintains many of the desirable security properties of proof of work without many of the downsides of proof of stake. I don’t think Bitcoin is going to move away from proof of work anytime soon, but it would be good to know that viable alternatives exist.
We could be wrong about these assumptions and there are a thousand reasons Spacemesh might fail. While I’d be sad if Spacemesh the blockchain fails, I’m consoled by the fact that the ideas and technologies we’ve developed and released to the world as a public good will live on regardless.
And this experiment too can only be run at layer one. It wouldn’t make any sense to deploy these technologies at layer two, or on top of Bitcoin. To pick one example, our consensus mechanisms are specifically designed for large scale, public, permissionless blockchains. No other application needs such a consensus mechanism—but the world does.
Success can take many shapes and forms. Successfully testing and releasing this software and these ideas into the world, regardless of what follows, is definitely one version of success.