Blockchain is a galactic algorithm

In the sense that it's literally made for a galactic civilization

Crypto, according to its advocates, is supposed to be secure to manipulation or hacking. Its advantages include fully transparent and inspectable code, decentralization that allows for zero-trust algorithms and don’t require trust in central authority, and a “code is law” ethos that governs all contracts and transactions to execute as written rather than by fuzzy and changeable human authorities.

In practice, no real cryptocurrency is particularly usable as currency. The code may be inspectable, but it’s still often full of exploitable bugs. Running your own decentralized wallet is complicated, computationally expensive, and risky, so most people just use centralized exchanges like coinbase. “Code is law” is a nice principle, but it still gets overturned by human law when there’s a strong enough social pushback. And while there may be a limit on the number of bitcoins that can ever be minted, bitcoin’s value is still to unstable to be a useful medium of exchange. Even crypto fans like Scott, when pushed to provide positive examples, typically point to stablecoins traded on centralized exchanges rather than real decentralized blockchains¹.

Still, I think there is a theoretically useful case for cryptocurrencies. They’re simply a galactic algorithm.

Classically, a galactic algorithm is an algorithm that’s asymptotically superior to the ones we actually use, but only on inputs that are too large to use in practice. For example, the Strassen algorithm for matrix multiplication has complexity O(n^2.81) (as opposed to the O(n^3) given by the basic algorithm), but it only actually starts working faster on matrices too large to be practical.

Crypto isn’t this kind of galactic algorithm. It’s the much more literal type of galactic algorithm, the kind that only works for a galactic scale civilization.

Consider an actual galactic scale (or even just multi-solar system) civilization. Lightspeed means sending lots of messages back and forth is too slow, and exchange of any sort of physical goods is even slower. And it’s probably far too big to have any sort of trusted central authority². So any sort of currency would have to be decentralized to work. It would have to have a code-is-law ethos, since the information transmission delays would take too long to renegotiate or appeal contract terms. And it would clearly need to have transparent, inspectible code.

Conversely, the weaknesses go away once we scale. If we have a whole planet full of lawyers and programmers looking through the code, we can probably get to a reasonable level of trust that we’ve caught all the bugs, issues, and exploits. We don’t have to worry about lacking the resources to set up our own wallet or exchange. and interplanetary hype cycles are probably too hard to maintain for more than a single planet to be affected at a time, given the time delays involved. We could have a traditional trust-based banking system within each planet (a single planet is small enough for that to work), and then turn to the blockchain just for interstellar commerce³.

1. There’s an argument to be made that having the backstop of a decentralized exchange to stabilize it gives the centralized exchanges the credibility they need to operate it, or makes it harder for the government to crack down on it. I don’t find it convincing though, both because a centralized exchange could still be unreliable if it wanted to (like FTX did), and because governments could still crack down on the front end of the centralized exchanges if they really wanted to.

2. Consider how long it takes to establish trust, and how easy it is to lose, on an earthly political timescale. If your last piece of information from a planet originated ten years ago, how willing would you be to trust that their central bank is still reliable now?

3. An interesting question I don’t have the answer to: Aside from proof-of-work and proof-of-stake, could we set up a stake-by-planet system, in which consensus is established by the majority of planets occupied by nodes? Probably it takes some effort to get a planet online, which could de facto work as proof-of-work while also giving effectively free access to any genuinely settled planet, but whether this is actually spoofable depends on sci-fi engineering factors I am out of my depth in describing.