The Bitcoin Standard: Destined to Fail

The Bitcoin Standard: Destined to Fail
April 06, 2022 by Nick

In his book The Bitcoin Standard, Saifedean Ammous proposes that Bitcoin should be used, not as a system for direct peer-to-peer payments, but rather as a settlement layer between banks and large financial institutions.

I think this is a terrible idea that will not work at all the way Ammous envisions.

In this post, I will share snippets from Ammous’ book, along with commentary explaining where the critical flaws are in this proposal. Page numbers are from the Kindle edition.

Why Use Bitcoin as a Reserve Asset in the First Place?

First of all, let’s begin with the reason why the proposal occurs in the first place. Why shouldn’t Bitcoin be used as a direct peer-to-peer payment system? The reason, according to Ammous, is scaling.

As it stands, the volume of these international [payments] is far larger than what bitcoin’s blockchain can handle, and if more such payments move to bitcoin, fees will rise to limit the demand for them. Yet, that would also not spell doom for bitcoin, because sending these individual payments is not the limit of bitcoin’s capabilities. Bitcoin is money free of counterparty risk, and its network can offer final settlement of large-volume payments within minutes. Bitcoin can thus be best understood to compete with settlement payments between central banks and large financial institutions, and it compares favourably to them due to its verifiable record, cryptographic security, and imperviousness to third-party security holes.

The Bitcoin Standard, page 208

While Bitcoin has historically been understood as “a peer-to-peer electronic cash system”, Ammous argues that this understanding may need to change. Since Bitcoin cannot scale* (on the main chain) to accomodate peer-to-peer payments for billions of people, it will have to be used in some other manner. Individual payments will have to be aggregated somehow so that many payments in the economy are represented by a single transaction on the Bitcoin ledger.

Ammous goes on to suggest that a roughly appropriate scale for this aggregation would be for 850 bank-like entities to do settlement transactions among themselves, with each serving an average of 10 million customers (for a total of 8.5 billion, basically the global population). The customers would transact through other payment channels, external to Bitcoin itself, managed by their chosen bank. Only when the banks wanted to do bulk settlements between each other would they commit the aggregated transactions of their customers into a single transaction on the Bitcoin blockchain.

Bitcoin’s current capacity of around 350,000 transactions per day can allow a global network of 850 banks to each have one daily transaction with every other bank on the network.

The Bitcoin Standard, page 209

Welcome to Bretton Woods 2.0

So what is the fundamental problem with this proposal? The fundamental problem is that a similar model has been tried before and it clearly didn’t work. I am, of course, referring to the Bretton Woods system. For those unfamiliar, Bretton Woods was an arrangement made between the world’s central banks, in which US dollars would be redeemable for gold, but only by a central bank. Ordinary US citizens could not redeem their paper dollars for gold, but the central bank of a foreign nation could do that. The citizens of each nation would use paper currency issued by their respective central banks, “secure” in the knowledge that the central banks would periodically settle accounts among themselves using the “layer 1” asset of the system: gold.

We know how that went. The central banks of the world kept fractional reserves of US dollars and inflated their own currencies. The US held fractional reserves of the gold and inflated the dollar. Eventually, the chickens came home to roost and the US was revealed to not be able to redeem all the gold obligations, not by a long shot. Nixon “closed the gold window” (i.e. defaulted on the US’ gold oblications to foreign creditors) and the last link between gold and the dollar was lost.

Ammous rightly rails against all the evils in society that result from fiat money inflation (both directly and indirectly). But if the best that Bitcoin can offer us is a “Bretton Woods 2.0”, then that’s not enough to solve those societal problems. The original Bretton Woods did not solve the problem of inflation, and neither would a so-called “Bitcoin Standard” which runs on exactly the same mechanism.

What about “Free Banking” on a Bitcoin Standard?

Now, to be fair, Ammous does not expect the “Bitcoin Standard” to play out this way. He himself is critical of the Bretton Woods system. Instead, he expects Bitcoin to become the reserve asset in a system of “free banking”. The goal of free banking is to keep banks honest through competition rather than through laws. One way to keep banks from lending out more paper than they have gold in their vaults would be to pass a law that says they must have full reserves, then keep auditors dropping in on them to check that they are following the rules. But it is expensive to continuously audit all the banks, and there’s still the possibility that they might “share” some gold (ship it from bank A to bank B right after bank A was audited, so that bank B has it on hand when the auditors arrive). With free banking, we scrap those laws and let the banks keep whatever level of gold reserves they like. But since they compete with each other for customers, they will have to keep their gold reserves above a certain level, otherwise their customers might get worried, start redeeming their paper for gold and pulling their gold out of the bank. At first this might just mean that the bank with low reserves loses business to competing banks. In the worst case, it could result in a “bank run” where the bank can’t pay out its gold obligations to its customers and goes out of business. This risk is what incentivises the bank to keep their level of reserves relatively high.

Ammous argues that Bitcoin-backed banks would operate under this type of “free banking” system and that would keep them honest (i.e they would keep a high ratio of Bitcoin reserves relative to the paper currency they issue).

In a world in which no government can create more bitcoins, these bitcoin central banks would compete freely with one another in offering physical and digital bitcoin-backed monetary instruments and payment solutions. Without a lender of last resort, fractional reserve banking becomes an extremely dangerous arrangement and it would be my expectation the only banks that will survive in the long run would be banks offering financial instruments 100% backed by bitcoin. This, however, is a point of contention among economists and time can only tell whether that will be the case.

The Bitcoin Standard, page 209

A point of contention indeed. The debate between free banking and full-reserve banking is ongoing among Austrian economists (and other free-market thinkers). I won’t try to resolve it here. For the sake of argument, let’s assume that free banking with gold as the base asset works really well. Let’s assume that free banking is effective at incentivising banks to hold plenty of gold reserves and not inflate the money supply. Even if that is the case, Bitcoin would still be a terrible base asset for a free banking system.

Why do I say that? Simple: because Bitcoin, unlike gold, would be almost impossible for ordinary people to redeem. That is why I am arguing that a “Bitcoin Standard” will inevitably look a lot more like a Bretton Woods 2.0 nightmare than a free-banking utopia. When gold is the base asset, there is no hard upper limit on the rate at which a bank run can occur. If 1,000 people turn up to the bank, they can each demand their gold and walk out the door with it right away. If there are 850 banks, then 850,000 people can do this (1,000 at each bank) and all walk away with their gold. They might bury it, they might put it under the mattress, they might put it in a vault, whatever. But they can get their gold out of the bank’s vaults that same day (if the bank is indeed solvent).

The Block-Size Cap Makes Bank Runs Near Impossible

This cannot happen with Bitcoin. By Ammous’ own admission, the Bitcoin blockchain only has capacity for those 850 banks to do one transaction per day with each of the other banks in the network. Suppose they have an average of 10 million customers (to use Ammous’ own example). If 1,000 of them show up to a given bank and want to redeem their Bitcoin, that still leaves 9.9 million that want to keep using the bank as normal that day. The bank still needs to do its usual settlement payments with all the other banks in order to keep serving the remaining 9.9 million customers. So the bank that is experiencing a “run” needs to do just as many on-chain Bitcoin transactions for settlement during the run that they needed to do the day or the week before. It’s just that now there is demand for an extra 1,000 on-chain transactions from the customers trying to execute the bank run. So those customers who want to pull their funds out are going to have to bid against all the major financial institutions in order to get their on-chain transaction processed. This could be remarkably expensive. Suppose that each of the 10 million customers of the bank do an average of $100 per person in transactions every day (using USD as our unit of account, just to make the example intuitive). That is $1 billion worth of transactions that the bank wants to settle with the other banks in the network. If you as an individual want to pull your funds “out”, you have to compete against institutions that are measuring the value of their daily on-chain transactions in billions. Even if your net worth is $1 million, you’re a fairly small fish. For the bank, your whole net worth is equivalent to a 0.1% transaction fee on their standard daily settlements. You’re obviously not going to bid your whole net worth to get your money out via an on-chain transaction. That would send you broke and defeat the purpose. But the smaller the percentage of your net worth you bid, the less likely it is that your transaction will get mined in any meaningful amount of time. In fact, you will be heavily incentivised to not even try it because the cost would be so high (much higher than paying a security firm to take a truck and a few guards to the bank and collect your gold bars).

In another section of the book, Ammous even acknowledges that this precise issue was the fatal flaw in the Bretton Woods system.

… the monetary discipline of the gold standard was almost entirely lost in this world where there were no effective controls on all central banks in expanding the money supply, because no citizens could redeem their government money for gold.

The Bitcoin Standard, page 58.

At best, your practical option will be to demand that your bank transfer all your funds to another bank in which you have greater trust. A number will disappear from your account at bank A and appear in your new account at bank B. Both the banks pinky-swear that the settlement transaction they did between them that day includes all of the Bitcoin that you “own”. If your net worth is $1 million, and they do a daily settlement transaction of around $1 billion of BTC between them, who’s to say whether your $1 million was or was not included in that settlement transaction? Or even if bank A does transfer some reserves to bank B, what’s to stop bank B transferring them right back to bank A the next day? If the two are colluding in any way, there is not much chance to prove it.

Everyone would be able to see in a block explorer that bank A has (say) 1000 BTC in its cold wallet. This is meant to be the reserve asset backing all the currency used by the customers of bank A. But how could you verify that the checking accounts of all the customers of bank A add up to 1000 BTC? How could you verify that the bank has not created any extra “fiat” coins to go in the checking accounts, which are not backed by any BTC in the cold wallet? You couldn’t. The only way to verify that would be for them to publish a list of the checking account balances of all of their customers. Then the customers could add up the amounts and make sure that no accounts were left off the list. But this would be intolerable from a privacy perspective. Who wants to keep their money in a bank that periodically publishes the balances of all of its customers?

All of these forces come together to make it near impossible for customers of a Bitcoin-backed bank to redeem their funds for “specie” in any meaningful way. So even if we managed to get something like Bitcoin “free banking” established, the economic incentives mean that it would tend towards something much more like Bretton Woods. Due to the excessive cost of transacting on-chain, specie would only move between the large players who form the inner circle, but would never be available directly to individuals. This opens the way for rampant inflation through fractional reserves.

Objection: What about a privacy-preserving, cryptographic proof of reserves?

In the best-case scenario, you might be able to figure out some kind of cryptographic proof system where you use a “range proof” to verify that the total amounts of all the checking account balances add up to the amount of BTC in a bank’s cold wallet, without revealing the individual balances. But of course, solving this for a single bank is only part of the battle. What if 10 banks collude to share reserves between them? They can move reserves around during their settlement transactions to make sure that each bank appears to have enough reserves at the moment they publish their range proof, but then they would return the reserves after the “proving” is done.

Okay, let’s take it a step further. Maybe we could come up with some kind of protocol that allows us to synchronise the proof-of-reserves at a given point in time? This is much more difficult than it sounds. The whole point of this system is to allow people to transact independently of anything happening on-chain. Proving that an off-chain transaction happened at a point in time corresponding to a given on-chain block height may not possible, even theoretically. But even assuming that it is somehow possible, you have a deeper problem. This type of system would require the participating banks to opt-in. And of course, any banks that are hoping to engage in inflation through fractional reserves are not incentivised at all to participate in some kind of cryptographically-secure proof-of-reserves system.

I can already hear people jumping to say “this is free banking! If my bank refuses to provide (hypothetical) cryptographic proofs of their reserves, then I will leave and take my business to a bank that does provide that level of assurance!” Good for you. I hope that we could convince a great many people to have the same attitude if such a system ever does come to exist. However, it brings me to the second fundamental flaw in the Bitcoin free-banking proposal.

Why Didn’t We Already Have Free Banking Before Bitcoin?

If Ammous’ concept of a “Bitcoin Standard” depends on the establishment of free banking, then it is doomed before it has even gotten started. Because of course, free banking is a concept that pre-dates Bitcoin by centuries. We’ve known about free banking all along.

The problem is not that free banking hasn’t been invented or tried. Nor is the problem that gold was an inappropriate reserve asset for free banking. Indeed, as I have argued above, gold is actually more suitable than Bitcoin in that there is no artificial constraint on the ability of a bank’s clients to orchestrate a “run” on gold. It’s not as though free banking was this theoretical concept, just waiting for an asset with the properties of Bitcoin to come along in order to make it work. It worked just fine already with gold.

No, the real problem with free banking was that it had been legislated out of existence. The rulers of various States had decided that a decentralised monetary system was not in their best interests. Under the pretense of “simplifying” the lives of citizens (who had to deal with a variety of notes from different currency issuers), governments tended to legilsate themselves a monopoly on currency issuance. Once that unaccountable monopoly is established, the inflationary effects are quite predictable.

So what does this have to do with the proposed “Bitcoin Standard”? Simple: the Bitcoin Standard proposes that if we transition to free banking, Bitcoin will be useful as a reserve asset. But if we were able to trasition to free banking, then the fundamental problems of fiat currency would already be solved. We would already have checks on fiat inflation. If we could force governments to accept free banking, then Bitcoin would be superfluous. Gold would work perfectly well as the reserve asset of choice. But we cannot force them to accept free banking. Governments are always incentivised to create propaganda against free banking and try to legislate it out of existence, because this is how they establish their coveted monopolies.

If free banking requires Bitcoin, then Bitcoin has a purpose. But if Bitcoin requires free banking, and free banking works fine without Bitcoin, then Bitcoin is unnecessary. Under the Bitcoin Standard proposal, this is exactly the situation in which we find ourselves. We would need to first establish free banking because Bitcoin is not a useful asset without it. Bitcoin can’t scale for peer-to-peer transactions without a banking layer, and Bitcoin-backed free banking can’t thrive if the Bitcoin banks are monopolised or cartelised, which is a virtual certainty given that every State is heavily incentivised to create that monopoly or cartel privilege.

Also note that it would be quite easy for a State to cartelise Bitcoin banks in their jurisdiction. Want to buy some real estate in Ruritania? The Ruritanian government has passed a law requiring that settlement of real estate deals is done in their local (allegedly Bitcoin-backed) currency. You must have an account at one of their approved banks which has a license from the State to perform such real estate settlements. If you don’t do this, if you try to complete the transaction outside of the official Ruritanian banking cartel, then the Ruritanian government will not issue or recognise your title deed to the property. You will be subject to forcible eviction by the Ruritanian police. Want car insurance? You’ll have to buy your car through their banking system. For security purposes of course, so they can prove you own it and can prevent insurance fraud. It’s for your own good, citizen.

These will be the patterns that unfold and deal continuous blows to any free banking system that tries to get off the ground. If Bitcoin requires free banking as a pre-requisite, then Bitcoin is not going to help us. We need something that we can use as individuals, whether free banking is in place or not. The real promise of cryptocurrency was that it might fulfil Hayek’s conjecture:

I don’t believe we shall ever have a good money again before we take the thing out of the hands of government, that is, we can’t take it violently out of the hands of government, all we can do is by some sly roundabout way introduce something that they can’t stop.

Frederick Hayek, Source:

Could the Lightning Network Provide a Solution?

All this talk about throughput limits raises a question: could the Lightning Network provide a solution? Would using the Lightning Network enable bank customers to redeem their Bitcoin more readily and overcome the problems that I have been describing?

The short answer is “no”. Under this “Bitcoin Standard” free banking model, on-chain transactions are extremely expensive. For that reason, if you use Lightning at all, you probably only have a single Lightning channel connecting you to your bank (which acts as a liqudity hub, connecting you to the broader Lightning network). Even this is unlikely, since the cost of opening a Lightning channel would be so expensive. The vast majority of people are much more likely to just use a trusted, custodial solution provided by the bank. But for the sake of argument, let’s assume that you have a Lightning channel open to your bank.

If you start to suspect that your bank is misbehaving, issuing “synthetic” Bitcoin and keeping only fractional reserves, what do you do? You try to get out. You can’t get out via an on-chain transaction, it’s too expensive. But that also means that you can’t close your Lightning channel, because that too requires an on-chain transaction. And it only gets doubly expensive if you are going to try and open a new Lightning channel to a new bank.

Even if you had two channels already open, you would need to do an on-chain transaction to rebalance your funds from your channel at the misbehaving bank to your channel at the better bank. At the end of the day, only a transaction on the base layer is really “settled”. So you will need to do an on-chain transaction to really redeem your specie. And if that is prohibitively expensive, then you are screwed. Lightning does nothing to address that particular dilemma.

Alternatively, suppose that the large banking institutions used the Lightning network to conduct their daily settlement transactions. Perhaps this would free up space on the main chain to allow a bank run to happen more readily?

There are two problems with this scenario. First, it assumes that banks trust each other enough to forego the security of having their high-value transactions settled on chain. This is a big assumption. If Bank A thinks there is any risk that Bank B may execute a hostile channel closure and reverse a $1 billion settlement transaction, then Bank A will be incentivised to take measures to prevent that. The most obvious measure is to require settling on-chain. But that aside, the core problem still remains. By Ammous’ own estimate, the base layer can handle a maximum of 350,000 transactions per day. For an average bank of 10 million customers, this means that 3.5% of their customers can withdraw their Bitcoin in a single day if they use the entire capacity of the Bitcoin blockchain for those withdrawals. It’s a drop in the bucket. If there are 10 banks around the world suffering bank runs at the same time (hardly difficult to imagine), then they only risk 0.35% of their customers taking funds out on average each day. Once this goes on for a week and the bank can show that they have paid out every request for specie, the fears of insolvency will most likely dissipate, especially if a State propaganda machine is involved. All it takes is for a few of these banks to scratch each others’ backs by sharing liquidity and any real “run” becomes almost impossible to get off the ground. Since there is a low and hard ceiling on the rate at which a bank run can happen, the banks will have a long period where they can still appear to be fully solvent. If a bank run appears to be having no effect, and executing the bank run is costing a substantial sum to the customers, the customers are unlikely to keep it up for long.

So, in summary, using Lightning doesn’t decrease the cost of an individual customer trying to exit their bank, and using Lightning between banks doesn’t free up a meaningful amount of extra on-chain capacity in terms of enabling a proper bank run.


Let’s summarise the points made so far.

  • Free banking only works when there is the real possiblity of a bank run.
  • Small blocks effectively prevent a bank run from happening.
  • Free banking only works when you can start an honest bank without State/cartel approval.
  • If transaction fees make it cost-prohibitive to “be your own bank”, then you must necessarily trust a third party to be your bank.
  • That third party then becomes a central point of failure, a natural target for State coercion, and a likely vector for the creation of a State-approved banking monopoly/cartel.
  • Therefore, attempting to implement free banking on top of Bitcoin will almost certainly degrade into a central banking cartel that is entirely capable of implementing inflation through fractional reserves.

Historically, the fear of Bitcoin maximalists has been that bigger blocks will create centralisation due to the increased costs of operating and maintaining a full node. While this is certainly a rational concern, the ever-decreasing costs of storage and bandwidth are making it look less and less likely for bigger blocks to have a meaningful negative impact in practice.

In contrast, small blocks have a centralisation tendency that is unavoidable and even enforced by the system itself. Small blocks restrict the number of actors who can ever operate in the decentralised layer of the system. Once that cap is reached, small blocks force additional actors to operate and transact in centralised upper layers. It is only the base layer that is decentralised, so limiting access to the base layer necessarily means limiting access to decentralisation.

Understanding this, we can make a prediction. To the extent that people want to operate in a permissionless, decentralised manner, they will necessarily opt out of a global Bitcoin-backed banking system. If people accept that system, then it will fail because it will not actually save people from the inflationary evils which it was intended to defeat. If people reject it, then it will fail simply through non-adoption. That is to say, the Bitcoin Standard, as a concept, is destined to fail.

Is The Bitcoin Standard Still Worth a Read?

Yeah, totally. Ammous provides a great background to many of the theoretical issues involved in finding an ideal form of “money”. He accurately diagnoses many of the evils created by the fiat money regime. He also makes some valid criticisms of crypto projects besides Bitcoin. I agree with some of those criticisms and expand upon them in my own book. Overall, The Bitcoin Standard is a great contribution to the history of Austrian thought on cryptocurrency. I just think Ammous’ overall conclusion that Bitcoin BTC is and must be the “one coin to rule them all” is mistaken. He gets many of the problems right. But his proposed solution isn’t actually fit to address them.

If you do want to read The Bitcoin Standard, warm my capitalist heart by buying it through this affiliate link:

Footnote on Scaling On-Chain

As an aside, saying that “Bitcoin cannot scale on the main chain” is a point of significant contention. For the purposes of this article, “Bitcoin” refers to BTC, the side of the 2017 Bitcoin fork which elected to retain a hard cap of 1MB on the size of blocks in the blockchain. Given that artificial constraint, it is certainly true that the main chain cannot accomodate global payments volume. However, there was another side to that fork in 2017 (resulting in the “Bitcoin Cash” alternate chain, aka “BCH”) which removed the hard cap on block size. The BCH camp argues that the original Bitcoin technology can be made to scale on the main chain to a point where it can accomodate global payment volumes. Since 2017, that community has been actively working on developing the technology to achieve that level of on-chain scaling. However, for the purposes of this article, we are discussing Bitcoin BTC with its 1MB hard cap, which is the version of Bitcoin that Ammous is proposing to use as the base asset in the Bitcoin Standard.