Five Transactions a Second, and Other Discourtesies to Commerce
Why gold knew its station, and what happens to a currency that has forgotten its own.
There is a charming hypocrisy at the centre of the Bitcoin Core thesis, and like most charming hypocrisies it is repeated so often that the people repeating it have stopped hearing what they are saying. They tell you, with an evangelist’s confidence, that Bitcoin is digital cash; and then they tell you, with an engineer’s apology, that it cannot quite manage to be one — not today, not next year, not ever in the form they have built. The system, they explain, settles around five transactions a second. They mention this the way a guilty man mentions the weather. Five. A second. From that small arithmetic the whole edifice of the cash claim collapses, and what stands afterwards is not cash at all. It is something older, slower, more dignified, and considerably less useful in the queue at a coffee shop. It is a broken digital bullion. And bullion, to its everlasting credit, has never told you it was money.
The arithmetic, set out without flourish, kills the rhetoric. A block on the Bitcoin Core chain arrives, on average, every ten minutes. Within it sits roughly one megabyte of transaction data, swollen to perhaps two or three megabytes under SegWit’s witness discount, with a theoretical ceiling that is a few times larger and is almost never reached in practice. The arithmetic that emerges from this design — and it is a design choice, not a law of physics — is somewhere between three and seven transactions per second, depending on how charitable one is with average transaction size and how forgiving one is with mempool conditions. The figure commonly used in serious commentary is five. Five transactions per second. Per second. Across all of humanity, simultaneously, using one network.
Set this against the world that already exists. The Visa network processes, in ordinary peak conditions, in the order of sixty-five thousand transactions per second, and could, on a stress test, manage rather more. Mastercard runs in the same range. The Chinese interbank settlement systems handle volumes that make the card networks look provincial. The total volume of retail payments processed globally each day runs into the hundreds of billions. A mid-sized Asian convenience store chain, on a single Saturday afternoon, processes more transactions than the Bitcoin Core network can process in a week. There is no version of this comparison that is gentle. The number is small and the world is large.
Compute the population arithmetic and the picture becomes worse. The Bitcoin Core network, at five transactions per second, can settle approximately four hundred and thirty thousand transactions per day. The global population is around eight billion. If every human were to attempt to use the system as cash — a single purchase each — the network would require approximately fifty years to clear the queue. Double the throughput, triple it, quadruple it; the result is still measured in decades. The cap is not a bottleneck to be optimised. It is a ceiling that defines what the system is. It tells you, with the brute honesty of a number, that the system was not built for the use to which its rhetoric is committed.
Consider gold. Gold, for the better part of five thousand years, has functioned as a monetary metal. It has been mined, weighed, stamped, hoarded, taxed, plundered, and occasionally buried in cellars. It has settled debts between merchants, between cities, between empires. It has backed paper currencies that circulated in volumes gold itself could never sustain. It has done nearly everything one can imagine doing with a monetary instrument — except function, in any sustained or universal way, as the medium people actually carried about their persons to buy bread.
This is not an accident of history. It is a property of the metal. Gold is heavy. It is difficult to divide into the small units at which most commerce occurs. It is dangerous to carry. It is expensive to verify. It is awkward to authenticate. The Roman aureus, the Byzantine solidus, the gold sovereign of the British Empire — these were instruments of large transactions, of state finance, of merchant settlement, of the upper register of commerce. The peasant did not pay for fish with gold. The peasant paid with copper, with silver fractions, with credit, with promissory notes, with whatever low-value, high-velocity instrument the local economy had managed to produce.
Gold was the settlement layer. The instruments people used for daily payment were something else. This is the structure of every commodity money system that has ever functioned at scale. There is a base — heavy, valuable, scarce, secure — and there is a circulating instrument — light, divisible, redeemable, ubiquitous. The base is gold. The instrument is the bank note, the bill of exchange, the cheque, the token. The base settles. The instrument circulates. The two are not the same thing. They are not pretending to be the same thing. The Bank of England note in 1850 did not claim it was gold; it claimed it was redeemable for gold. Different claim, different category, different role.
The gold settlement model worked precisely because gold did not pretend to be cash, and the cash did not pretend to be gold. Each occupied its station. The hierarchy was explicit, the legal relations were clear, the operational distinction was a matter of daily knowledge for every literate person who handled money. Even the illiterate knew the difference; the sovereign in the pocket was not the shilling in the hand, and neither was the bank note tucked into the waistcoat. The whole arrangement was honest about its hierarchy.
The trouble with Bitcoin Core is that it has lodged itself, by deliberate engineering decision, into the settlement role — a role gold has occupied for millennia with rather more legal, custodial, and physical infrastructure to recommend it — while continuing to advertise itself in the cash role. It cannot fill the cash role. The five-transactions-per-second ceiling forbids it. It can, perhaps, fill some of the settlement role, but only awkwardly, and only by competing with an incumbent that has five thousand years of head start, a legal apparatus in every functioning jurisdiction, a custodial industry of vast scale, and an authentication infrastructure refined across continents and centuries.
This is the category error at the centre of the entire discourse. The Bitcoin Core proponent says “digital cash” and means, when pressed, “digital gold.” The two phrases are not synonyms. They are not adjacent. They sit at the opposite ends of the monetary stack, and one cannot collapse into the other without ceasing to be what it was. Cash is what circulates; gold is what backs it. The thing that settles cannot, at the same volume, be the thing that circulates. The arithmetic does not allow it. The arithmetic has never allowed it. And no white paper, however inspired, can repeal arithmetic.
If Bitcoin Core is digital gold, then by the logic of every monetary system humanity has ever constructed, there must be some other instrument that functions as the circulating cash redeemable into it. That other instrument is not Bitcoin Core. It cannot be Bitcoin Core, because Bitcoin Core has capped itself out of the role. The cash in a Bitcoin Core economy must therefore be something else — a Lightning channel balance, a custodial deposit at an exchange, a tokenised IOU from a payment processor, a stablecoin issued on a different chain. These things are not the chain. They are claims against the chain, or claims against an intermediary, or claims against a different protocol entirely. They are paper, in the older sense of the word: instruments that circulate, redeemable, in some structurally weaker way, against the base asset they reference.
This is the gold standard reconstructed in software. The user does not transact in Bitcoin. The user transacts in something that promises Bitcoin. The Bitcoin sits in cold storage, on the base chain, settling occasionally. The user’s daily activity occurs on instruments whose security model, custody model, and legal status are categorically different from the base chain’s. The whole structure is the structure of correspondent banking, dressed in new vocabulary, sold to a generation that has not read the history of the institutions whose form it has reproduced.
Now imagine — and this requires only modest imagination, because the experiment has been run in small bursts — what occurs when a population actually attempts to use the Bitcoin Core network as cash. Suppose some sufficiently large country, gripped by either ideology or desperation, mandated retail payments on the base chain. Suppose, less dramatically, that some non-trivial fraction of global commerce migrated.
The mempool is the queue of unconfirmed transactions waiting for inclusion in a block. It has a finite default size in the node software, and an infinite size in concept — transactions can keep arriving even as the queue grows. When the queue grows beyond what blocks can clear, fees rise. Miners include the highest-fee transactions first. Low-fee transactions wait. As demand rises further, the fee level required for confirmation in the next block rises with it. Anyone who has watched the Bitcoin Core network during a period of moderate stress — the late 2017 episode, the spring of 2021, the ordinal-inscription episodes of 2023 — has seen fees rise from a few cents to fifty dollars and beyond per transaction. These were not periods of mass adoption. They were minor demand shocks. The system was already at its capacity for each one of them.
Project this against actual cash demand and the picture becomes farcical. A single mid-sized retail chain, in a single afternoon, generates more transaction demand than the network can process. The fee level required to clear even a fraction of that demand would price out the underlying commerce. A two-dollar coffee paid for with a fifty-dollar fee is not a payment system. It is a curio. The arithmetic is not subtle. If the network is at five transactions per second and the demand is five thousand, then nine hundred and ninety-five out of every thousand transactions must either pay enough to outbid the others or wait. Most will wait. Most will wait long enough that the underlying commerce has been completed, refused, or forgotten.
What emerges is not a payments network. It is an auction for block space, conducted in real time, in which the participants bid for a scarce commodity — inclusion — and the price floats to whatever the marginal user will pay. Inclusion becomes the luxury. The system stops being money and starts being a queue.
The economic consequence is severe. Cash has the property of universal accessibility — anyone, regardless of wealth or status, can transact in it for ordinary purposes. The five-transactions-per-second ceiling forecloses that property. Only those whose transactions are large enough to justify the inclusion fee can transact. Small holders, small payments, small commerce — the great mass of ordinary economic activity — are priced out by the very mechanism that is supposed to serve them. A monetary instrument that excludes the small payer is not money. It is a settlement asset for the rich. Whatever else one might call it — and the proponents have a great many words for it — the one word that does not fit is cash.
And so, even were the system somehow forced into the cash role by ideology or mandate, it would fail in the same way: by reverting, under pressure, to the settlement role for which its design is suited. The pressure of mass use would drive small users off the chain, force them onto secondary instruments, and re-establish the gold-and-notes hierarchy with Bitcoin Core in the gold seat. The base chain would settle large transactions for those who could afford it; the rest of commerce would migrate to whatever circulating instrument the market produced — exchanges, custodians, Lightning hubs, stablecoins, anything that could move at the scale required. The chain would not become cash. The chain would, under stress, expel its smaller users into the arms of intermediaries.
The proponents of Bitcoin Core have an answer to all this, and the answer is the Lightning Network. The Lightning Network, they will explain, solves the scaling problem. It allows millions of transactions per second, off-chain, between participants who have opened payment channels with on-chain transactions. The base chain settles. The Lightning channels circulate. The system, they say, is whole.
This answer requires some care, because the answer is not wrong on its own terms — it is just wrong about what it has constructed.
The Lightning Network is not Bitcoin. It is a separate protocol, layered on top of Bitcoin, with a separate security model, a separate trust model, and a separate set of failure modes. A Lightning channel requires two parties to lock funds in a multi-signature output on the base chain. Those funds can then be sent back and forth between the two parties off-chain, with periodic on-chain settlements. To transact with a counterparty with whom one does not have a direct channel, one must route through intermediaries — and the intermediaries must have liquidity in the right direction, must be online, must be honest, and must not be censoring.
The Lightning Network, in its mature form, is a network of payment hubs. The hubs hold liquidity. The hubs route payments. The hubs can refuse to route. The hubs know who is paying whom, at least probabilistically, because routing leaks information about path and amount. The hubs are, in every economically meaningful sense, banks — institutions that hold funds, provide payment services, and exercise discretion over which payments to facilitate. The fact that the hubs are non-custodial in a narrow cryptographic sense — they cannot, in normal operation, simply steal the channel funds — does not change the institutional fact that they are intermediaries, and the network they form is an intermediated network. The user who routes a payment through a Lightning hub is depending on the hub’s behaviour in a manner functionally indistinguishable from the way a payer depends on a payment processor. The cryptography rearranges the failure modes; it does not abolish them.
This is not digital cash. Cash, in any meaningful sense of the word, is bearer, peer-to-peer, and does not require an intermediary to function. The Lightning Network, by construction, requires intermediaries. It can be made to work, and within its limits it does work, but it is not what was promised in the Bitcoin white paper. The white paper described a peer-to-peer electronic cash system in which transactions were sent from one party directly to another. The Lightning Network is a hub-and-spoke routing network whose security depends on watchtowers, on online presence, on the honesty of the routing path, and on the operational competence of an emerging class of professional liquidity providers. The white paper described one architecture; the Lightning Network is another. Calling the second by the name of the first is a brand exercise. It is not a technical statement.
To call this Bitcoin payments is to use the word Bitcoin the way a hotel calls a queen-size bed a king. The brand is being stretched to cover something it does not technically fit. The hotel is selling sleep; the brand is selling something else. Both serve a function. Neither is what the language promises.
The Lightning Network has its own scaling limits, its own liquidity constraints, its own privacy weaknesses, and its own custody questions. Channels can be force-closed. Watchtowers can fail. Routing can fail. Liquidity can dry up in the wrong direction. None of this is fatal to the protocol — it is a perfectly respectable engineering achievement on its own terms — but it is not the answer to the question of why Bitcoin Core cannot serve as cash. It is, in fact, the proof of the original claim. The Lightning Network exists because Bitcoin Core cannot serve as cash. If it could, no Lightning Network would be needed. The existence of the second layer is the confession of the first.
And so the structure that emerges is precisely the gold-and-notes structure of the nineteenth century. The base chain settles. The Lightning channels circulate. The Lightning hubs operate as banks. The end user transacts on instruments that are not the base chain, against intermediaries that are not the protocol, with a security model that is not the security model the white paper described. Gold and notes. Bullion and paper. Settlement and circulation. The hierarchy is the same as it was in 1850. The only difference is that in 1850 the parties were honest about which instrument was which.
It remains to ask the question the title posed: why is gold closer to cash than Bitcoin Core? The answer comes in three parts, and each is worth stating plainly.
First, gold can be transacted at any scale the holder requires, subject to physical handling. A gold coin can be passed from hand to hand without a network’s permission. There is no mempool, no fee market, no block confirmation, no protocol upgrade required. The act of payment is the act of physical transfer. It is, in the strict technical sense, peer-to-peer. The participants do not require an intermediary. They do not require a network. They require only the metal and the willingness to transfer it. This is closer to the original definition of cash — bearer, immediate, final — than anything Bitcoin Core can offer at five transactions per second through a global queue.
Second, gold’s settlement role is honest. Gold has never claimed to be the daily medium of exchange in a developed economy. The historical literature, the central bank literature, the monetary economists from Menger onwards have been clear about gold’s role as a base layer for monetary systems built on top of it. There has been no rhetorical campaign to convince the public that gold is what they should use to buy groceries. Gold’s status is openly that of a settlement and reserve asset. Bitcoin Core, by contrast, has spent fifteen years insisting on a role its design forecloses, and the resulting confusion is not gold’s fault. The honesty of gold is the honesty of an instrument that knows its own function. The dishonesty of Bitcoin Core is the dishonesty of an instrument advertised for a function its own arithmetic forbids.
Third, gold’s circulating instruments — bank notes, deposit balances, payment cards — have, for the better part of three centuries, been built around it without pretending to be it. The Bank of England note circulates as a claim on something. The cardholder uses a deposit balance as a claim on something. The hierarchy is explicit. The user knows that the instrument in hand is not the base asset; it is a claim against the base asset, mediated by institutions, with a clearly understood legal and operational structure. This is what a functional monetary system looks like. The base asset settles. The circulating instrument circulates. The two are different and the difference is acknowledged.
Bitcoin Core, by contrast, has been built into a posture of denying its own structural reality. The hierarchy is the same — base chain settles, secondary instruments circulate — but the rhetoric pretends the secondary instruments are the base chain. The Lightning balance is described as Bitcoin. The custodial exchange balance is described as Bitcoin. The stablecoin pegged to the dollar but issued on a side chain is described as part of the Bitcoin economy. The whole linguistic structure conspires to obscure the fact that the actual settlements occur on something other than what the user holds in hand. Gold, with its honest hierarchy, is more truthful to its users than Bitcoin Core has yet managed to be to its.
Suppose, finally, that we take seriously the question of what happens if the network is actually used as cash. Not by a small population of enthusiasts, not by a coffee shop in El Salvador as a publicity exercise, but by a meaningful fraction of global commerce. What follows is not a forecast. It is a description of mechanisms that have already operated, scaled to a regime in which they will operate at full force.
The first effect is fee escalation. As demand for inclusion exceeds capacity, fees rise to the level at which marginal demand is choked off. In any moderately stressed period — and the periods so far have been moderately stressed at best — fees have climbed from cents to tens of dollars per transaction. The level at which actual cash-equivalent demand would be choked off is significantly higher. A network operating at saturation under cash-like demand would price most transactions at hundreds of dollars apiece before settlement could even begin. The settlement layer that demands a hundred-dollar fee to clear a five-dollar payment is not in the payments business. It is in the auction business.
The second effect is the migration of users off the chain. Faced with fees that exceed the value of their transactions, users will move to secondary instruments — custodial accounts, Lightning channels, stablecoin networks. The on-chain user becomes a privileged minority. The network reverts to its settlement-asset character. The rhetoric of Bitcoin payments becomes detached from what most users actually experience, which is balance held with intermediary, occasionally settled to base chain when the operator chooses to do so. This is not cash. It is correspondent banking with extra steps and worse uptime.
The third effect is institutional concentration. The intermediaries that handle the user-facing payments — exchanges, Lightning hubs, custodians — gain economies of scale. Liquidity concentrates. The hubs become banks. The exchanges become payment processors. The original promise of disintermediation is reversed, in practice, by the necessity of operating around the throughput cap. The system, sold as a refuge from financial intermediaries, generates financial intermediaries with the inexorability of a thermodynamic law. The market does not respond to ideology; it responds to scarcity and friction, and where the base layer is scarce and frictional, intermediation appears to absorb the demand.
The fourth effect is regulatory capture. The intermediaries, being institutions of meaningful scale, become regulable. Anti-money-laundering requirements attach to them. Know-your-customer obligations apply. The exchanges, the Lightning hubs in their commercial form, the custodians — all become subject to the same regulatory apparatus that governs traditional banking. The base chain settles, and the settlements are largely between regulated entities transacting on behalf of users whose identities are recorded and whose activities are monitored. This is the opposite of the cypherpunk image that animated the early literature. It is also the predictable consequence of building a settlement asset and pretending it is cash. The pretence collapses; the reality is regulated. The chain has not abolished the banks; it has produced a new set of banks under different names, subject to the same regulators, performing the same intermediating function with a different vocabulary.
The fifth effect — and this is the one rarely mentioned in polite company — is that the security model degrades. Bitcoin Core’s security depends on the block subsidy and on transaction fees. The block subsidy is on a deterministic schedule toward zero. The fees must therefore grow over time, or the security budget collapses. But the fees are constrained by the throughput cap. If the chain processes five transactions per second, it can only collect fees from five transactions per second. Beyond a certain point, those fees cannot grow further without choking off the use that generates them. The chain is therefore in a bind: it needs fees to grow to compensate for the declining subsidy, but its throughput cap forbids the volume of fees that would be required without driving away the very transactions whose fees it depends upon. The eventual resolution of this bind is one of the unaddressed questions of the design, and it is not a small one. The settlement asset that cannot pay for its own security is an asset whose long-run guarantees deserve quieter rhetoric than they have so far received.
A monetary system whose long-run security budget is questionable, whose throughput cannot match the demand of its supposed use case, whose actual user activity occurs on intermediated secondary instruments, and whose institutional structure converges on the banks it was meant to replace — this is not cash. This may not even be reliably digital gold. It is, at best, a high-friction settlement asset competing against an incumbent — physical gold — whose institutional, legal, and custodial infrastructure has five thousand years of head start and shows no sign of dissolving in response to a new entrant whose arithmetic is so plainly inadequate.
The honest conclusion is the conclusion that the Bitcoin Core proponents will not, on the whole, deliver. It is that the system, as built, cannot be cash and will not be cash, and is at best a settlement-asset competing with bullion in a category that bullion already occupies. Whether it succeeds in that competition is an empirical question with a contested answer. Whether it should have been allowed to drift from its original specification into that competition is a question of intellectual honesty, and the answer there is less contested than the silence around it suggests.
Gold, to its eternal credit, has never claimed to be more than gold. The Bitcoin Core network was sold as cash, built as bullion, and operates, in practice, as the bottom layer of a stack of intermediaries that look, structurally, exactly like the banks the original literature claimed to abolish. The rhetoric and the reality have come unstuck, and the gap between them is the central fact of the present discourse — central, and unmentioned, and the more important for being unmentioned.
The choice before the Bitcoin Core proponent is therefore narrower than it might appear. Either the system accepts its station — settlement asset, bullion analogue, base layer for an intermediated stack — and ceases to describe itself as cash; or it admits that the design was wrong, that the cap was a mistake, that the cash thesis required throughput it cannot provide. There is no third option in which the chain remains capped and the cash thesis remains honest. The arithmetic is the arithmetic. Five is five. A second is a second. Humanity, however charitable it may be inclined to feel about the project, makes considerably more than five payments per second, and shows no sign of slowing down to accommodate a chain that has decided not to grow.
Gold understood its station. The Bitcoin Core network has not yet decided what its own is. Until it does, gold is closer to cash than Bitcoin Core — because gold, at least, is not pretending.