Oddbean

Here’s the answer: Nothing is free. There is a high cost to enable hundreds of thousands to run Bitcoin nodes and secure the network. So high in fact, that it does prevent there from being final settlement of all global transactions, or even one transaction per person per year, on the main Bitcoin network. But this is a reality which should not be tampered with. There are many great solutions which have different sets of trade offs to solve this problem. Liquid, Fedimint, Sidechains, Lightning, Ark, Channel Factories, Zero-Conf channels, Uncle Jim setups, etc. But frankly, before we can even start addressing this issue as a serious problem, you’d need to get 250M people who are demanding an annual sovereign Bitcoin transaction. That’s not the current reality. In the future, most people will interact with Bitcoin through Bitcoin banks, and semi-trusted organizations like Liquid, Sidechains, and Fedimints. There’s trade offs to this, but even with its drawbacks it’s still magnitudes better than the current world. Consider this - in fiat, you can take custody of a small amount of money (cash) but any meaningful amount of money essentially has to be held by a third party custodian like a bank or be someone else’s liability like government debt. With Bitcoin, it’s possible to take fully sovereign self custody of a large amount of money, but smaller amounts work best with some trust reinserted into the system. People with small amounts of Bitcoin have the economic guarantees of Bitcoin secured by people who have large amounts of Bitcoin. Ultimately, if you’re someone who is talking about this problem right now, you’re early enough that you’ll always have enough Bitcoin to transact on the base layer (fees are denominated in BTC not USD). For those who adopt last, they will likely be doing so without even knowing it. Their pension will be buying Bitcoin on their behalf, or their bank or credit card company will have moved to a Bitcoin standard without them explicitly asking for it. The worst thing we could do is try to change Bitcoin to accommodate tens of billions of transactions a year for people who don’t even demand them yet, and in the process destroy the distribution of the ledger and the robustness of Bitcoin itself.

Very well thought out response. Like the contrast of custody in fiat vs Bitcoin. I'd rather be able to custody most of my money myself and increase trust for the convenience of transacting in smaller amounts than the other way around. @DarthCoin really has the concept of layers in a bitcoin world down well.

To be really fair, I’d love to keep pushing the threshold down. Right now, anything less than 250k sats (the future median net worth) is impractical to interact with solely on-chain. Some of the stuff being worked on where you could run a full node using Merkel trees or some other form of cryptographic proofs has a lot of potential. I’m not encouraging that we become complacent and stop working on better sovereignty capabilities, I’m just arguing why it’s 1. Not a huge fundamental issue at the moment 2. Not something we should solve through short sighted changes like blocksize

Are there any projections on what fees on the base layer will become? Could it cost 100k sats or 1M sats to pay the transaction fee for a single input and single output transaction? Do people think it could go much higher?

It could (probably not 1M though) but only temporarily. You’ve got to remember that Bitcoin is scarce and finite, so there is definitely an upper limit to the fees that people would be willing to pay. I actually think that fees could go down in Bitcoin terms, but go up in USD terms. Median fee in 2050 might be 20 sats/vb but that might equal $250 dollars for instance.

Thanks - i expect it’s super hard to speculate, but do think there’s any way of estimating how many transactions people will want to perform? I was tempted to make a simulation of agents and use data like number of houses bought in the world, number of business sales, etc to get an idea of how many possible transactions there could be… But it gets super messy. I suppose it won’t be just bitcoin and other systems like you mentioned will take a lot of the load. Any chance you know someone who’s explored this in depth?

There are trillions of transactions which take place annually. Bitcoin can do about 200 million layer one transactions in its current form. FedWire is a interbank settlement network run by the federal reserve, which did 200 million transactions last year. That is the best comparison to Bitcoin in terms of volume and final settlement assurances (although it’s nowhere near as final, nothing in fiat is irreversible). You can use Bitcoin however you wish, so I’m sure there will be people like you and me who buy houses or cars on L1. In 50 years though, L1 will likely be dominated by financial service providers and banks settling between each other, or fedimints and other Lightning providers opening and closing channels. You definitely won’t be buying cars on L1, and you probably won’t even buy a house on it either.