IPv4.1 Does Not Exist
To widen the field is to write a new protocol.
To widen the field is to write a new protocol. The engineers who built the world’s network had the honesty to give their successor a new number. The men who would widen the coin want theirs to keep the old name — and the old name’s reputation.
Keywords: protocol, consensus, interoperability, IPv4, IPv6, address space, integer ledger, satoshi, base unit, hard fork, backward compatibility, identity, redenomination.
There is a word that does more dishonest work than any other in the vocabulary of progress, and the word is upgrade. It is a soft word, a reassuring word, a word that promises the same thing only better, as though improvement were a matter of polishing rather than of replacing. It permits a man to believe he is keeping what he had while acquiring what he lacked. And like most words that flatter, it conceals a question it dares not ask aloud: better according to whom, and at the cost of what, and is the thing that emerges from the polishing still the thing that went in?
An upgrade, properly understood, is a change that the ancestor would still recognise as its own — a change after which the new and the old can still speak. The moment they cannot speak, the word collapses. What you hold is no longer an improved version of the thing. It is a different thing, related to the old by descent but not by identity, the way a son is related to a father: a successor, not a continuation. And a successor that insists on being called by its father’s name, that wears the father’s clothes and signs the father’s correspondence, is not a son at all. It is an impostor. The whole question of whether redefining a protocol’s base unit produces an upgrade or a usurpation turns on this single distinction, and the men who profit from the confusion have every incentive to keep it blurred.
A Protocol Is an Agreement, Not an Asset
Begin, as always, by refusing to be vague about the noun. A protocol is not a product. It is not a brand, not a logo, not a market capitalisation, not a community of the faithful. A protocol is an agreement — a set of rules that a population of strangers must share, exactly and without deviation, in order to understand one another at all. It is the grammar of a conversation conducted between machines that will never meet, trust each other on no evidence, and verify everything. The rules are not a feature of the protocol. The rules are the protocol. There is nothing else there.
From this one fact, everything follows. If the rules are the protocol, then the boundary of the protocol is the boundary of agreement. Two machines that share the rules are inside it; two that do not are outside it, and outsiders cannot be admitted by sentiment. Compatibility is not a courtesy the protocol extends. Compatibility is the protocol’s definition of itself. To remain compatible is to remain the same protocol. To break compatibility is not to wound the protocol or to extend it or to evolve it. It is to author a different one and to stand, from that instant, outside the agreement you have just rewritten.
Compatibility is not a virtue a protocol possesses. It is the only thing a protocol is.
This is why the language of “upgrading the protocol” is so often a swindle in disguise. One does not upgrade an agreement by unilaterally changing its terms and demanding the counterparties honour the old signature. One either persuades every party to adopt the new terms in the same instant — which, among millions of strangers who trust nothing, is a fantasy — or one has simply walked out of the contract and drafted a fresh one. There is no third door. The man who tells you he has upgraded the agreement while half the signatories never consented has not upgraded anything. He has founded a rival agreement and hopes you will not notice the founding.
The Engineers Who Numbered Their Honesty
The world has already run this experiment, at the largest scale in the history of engineering, and the engineers who ran it behaved with a candour that the monetary reformers would do well to study and will certainly decline to imitate.
The protocol that carries nearly every word, image, and transaction across the earth began with an address field thirty-two bits wide. Thirty-two bits affords something over four billion distinct addresses — four billion two hundred ninety-four million and change — a number that seemed, in 1981, as inexhaustible as the round figures of the credulous always seem until the day they are exhausted. The specification that fixed this width was published, plainly numbered, as the fourth version of the Internet Protocol. It did not promise infinity. It promised four billion, and four billion eventually proved too few for a planet that put a networked machine in every pocket and every doorbell.
So the engineers faced precisely the question that confronts anyone who finds his address space, or his unit space, too narrow: how does one make the field bigger? And here is the lesson, written in the dullest and therefore the most trustworthy prose, the prose of a standards document. They widened the address from thirty-two bits to one hundred twenty-eight — a space so vast that to speak its size is to leave the realm of intuition entirely. And they did not call the result the fourth version, improved. They did not call it four-point-one. They called it the sixth version, and they wrote, in the cold flat sentences of the specification, that it was “a new version of the Internet Protocol,” “designed as the successor” to the fourth. New version. Successor. Not extension. Not upgrade. Successor — the word a son uses, not the word a brand uses.
They had the courage to be called Six. The whole of their honesty fits in that single digit.
Observe what that candour cost them and what it bought them. It cost them the comfortable fiction of continuity. They could not pretend the new network was the old network grown larger; they had to admit, in the name itself, that it was a different network related to the old by purpose and ancestry but not by interoperation. And what it bought them was the one thing worth buying: the truth. Nobody who reads the number Six can be deceived about what he is dealing with. The honesty was compiled into the name.
Why You Cannot Widen the Field in Place
The reformer who wishes the coin had a smaller unit imagines that the field can simply be stretched — that one adds a decimal, or widens the integer, the way a tailor lets out a seam, and the garment is the same garment with a little more room. The network engineers can tell him, from decades of expensive experience, that the field cannot be stretched in place, and the reason is the same reason in both cases, because it is the reason that governs all agreements among strangers.
The width of the address is not a private setting on each machine. It is the shared wire format — the shape of the thing every participant must read identically or fail to read at all. A machine that expects a thirty-two-bit address and receives a hundred-twenty-eight-bit one does not see a bigger address. It sees nonsense. It cannot be coaxed, by good intentions, into understanding a sentence whose grammar it does not possess. So you cannot widen the field on some machines and leave it on others and still have one network; you have two populations that have ceased to be mutually intelligible. And you cannot widen it on all machines at the same instant, because there is no hand large enough to reach every machine on earth at once, and no authority the strangers have agreed to obey.
What actually happened, therefore, is what must always happen: the two protocols now coexist. The fourth version did not vanish when the sixth arrived; it persists, decades later, running alongside its successor, and the two communicate only through deliberate machinery built to translate between them — dual stacks that speak both tongues, tunnels that smuggle one inside the other, translators that rewrite the one into the other at the border. The translation machinery is not evidence that they are one protocol. It is the proof that they are two. One does not build a translator between dialects of the same language. One builds it between languages.
You cannot let out the seam of an agreement. You can only sew a second garment and persuade people to call it the first.
The Coin’s Address Space
Now lay the coin beside the network and watch the analogy hold at every joint, because it is not an analogy of decoration but an analogy of structure.
The unit of the coin is recorded as an integer — a count of the smallest indivisible thing, held in a fixed-width field, validated against a fixed ceiling by every participant independently. This integer field is to the coin precisely what the address field is to the network: the shared wire format, the consensus invariant, the shape every node must read identically or fail to validate at all. Every node, alone and trusting no one, checks that the integers add up and that no integer exceeds the agreed maximum. That check, performed a million times by a million strangers who never confer, is the agreement. It is the grammar of the conversation.
To “add a decimal” — to renumber the unit so that ten of the new stand where one of the old stood — is to widen this field. It is to change the shared wire format, to redefine the integer that every node validates, to alter the ceiling against which every amount is checked. And the consequence is identical to the network’s, because the cause is identical. A node running the old rule and a node running the new rule do not see a finer-grained coin. They see each other’s blocks as invalid. They reject one another. They cease to be mutually intelligible. You have not let out the seam of Bitcoin. You have sewn a second garment.
This is the entire matter, stated without ornament: renumbering the base unit is the monetary equivalent of going from the fourth version to the sixth. It is not four-point-one. It is a successor — a new protocol, born of the old by descent, incompatible with it by construction, and incapable, from the moment of its first block, of speaking to the chain it left behind. Whether one widens an address or a coin, the law is the same: the field cannot be changed in place, because the field is the agreement, and an agreement changed without unanimous consent is not amended. It is replaced.
The Swindle the Engineers Refused
Here the two stories diverge, and the divergence is not technical but moral, which is to say it is the only part that finally matters.
The engineers, having built a successor, named it a successor. They did not arrange for the sixth version to inherit the fourth version’s identity by stealth. They did not insist that an old packet and a new packet were secretly the same packet. They did not market the successor as the original-only-better and quietly hope the world would forget there had ever been a break. They numbered the break into the name and let everyone see it. Their candour was not a marketing decision; it was a precondition of the thing working at all, because a protocol that lies about its own boundaries cannot be reasoned about by the strangers who must rely on it.
The monetary reformer wants the opposite of this candour, and wants it precisely because candour would ruin him. He wishes to widen the field — to author a successor — and to have that successor inherit, without acknowledgement, the name, the history, the ledger, and above all the reputation of the protocol he is leaving. He wants the prestige of continuity while performing the act of rupture. He wants to be called the fourth version while behaving as the sixth. He wants, in a word, to be the impostor son who answers to the father’s name at the father’s funeral and signs the father’s will.
The engineers were brave enough to be a new number. The reformers want to be a new protocol with an old name — which is only a polite description of fraud.
And the tell, the unfailing tell, is the demand for the old name. A man who has built something genuinely new and genuinely better does not need to steal a dead protocol’s reputation; he is content to be judged on his own. It is only the man who suspects his successor cannot win on its merits who insists on inheriting the merits of what it replaced. The grasping after the name is the confession that the thing under the name has changed. Honesty would have renumbered it. Dishonesty keeps the number and changes everything beneath it.
The Bridge Is the Confession
There is a final symmetry, and it is the most damning, because it turns the reformer’s own machinery into testimony against him.
When the old chain and the renumbered chain coexist — and coexist they must, for the old does not vanish merely because a successor was declared — value can pass between them only through a bridge: a peg, a translator, an external mechanism built deliberately to convert the one into the other across the border that now divides them. This bridge is sold as a convenience, a feature, a sign of the new protocol’s generosity toward its predecessor. It is nothing of the kind. It is the exact monetary counterpart of the tunnels and translators that shuttle between the network’s fourth and sixth versions, and it carries the exact same meaning. One does not build a translator within a single protocol. The bridge exists because, and only because, there are now two protocols where there was one. The bridge is not the reconciliation of the schism. The bridge is the schism, rendered in machinery, signed and dated.
So every artifact the reformer points to as proof of continuity is in truth a monument to rupture. The translation layer proves two grammars. The peg proves two ledgers. The very apparatus required to make the new thing usable alongside the old is the apparatus that would be unnecessary if the new thing were the old thing. He cannot have it both ways. Either the chains are one protocol, in which case no bridge is needed and none could be built; or a bridge is needed and is built, in which case the chains are two, and the question of which one is Bitcoin has already answered itself.
Identity Is Fidelity to the Rule
What, then, decides identity, when two chains share an ancestry and contest a name? Not the size of the market, for popularity has never once made a successor into its predecessor. Not the loudness of the evangelists, for volume is the argument of those who have lost the argument. Not the custody of the brand, for a name can be appropriated as easily as a crown can be stolen, and the theft of the crown has never made the thief into the king. Identity is decided by one thing only, and it is the same thing that decides whether a packet belongs to the fourth version or the sixth: fidelity to the rule.
The chain that preserves the invariant — the integer unit, the fixed field, the consensus grammar the original strangers agreed to share — is the original protocol, however small its banner. The chain that widened the field is the successor, however large its banner, and however indignantly it insists on the old name. This is not a matter of preference or allegiance. It is a matter of definition, as cold and as final as the definition that separates the sixth version of a network protocol from the fourth. The successor may be better. The successor may be worse. The successor may inherit the earth. But the successor is not the original, and the only honest thing it can do is what the network engineers did without flinching: take a new number, and let the world see the break.
The original is the chain that kept the rule. Everything that widened the field is a new protocol wearing a stolen name — and a stolen name is the one possession that proves its owner had to steal.
The engineers compiled their honesty into a single digit and asked no one’s forgiveness for it. They understood that to widen the field is to write a new protocol, and that a new protocol deserves a new number, and that the name belongs to the rule and not to the marketing. The reformers understand this too. That is precisely why they want the old name so badly. One does not fight to inherit a reputation one has not abandoned. The fight for the name is the admission that the thing beneath it is already gone.
References & sources
Classification, stated plainly: these are standards documents and primary specifications, not peer-reviewed economics. They are cited because they are the defining rules of the protocols discussed — the only authority that bears on the claims. No citation rests on an abstract; none is invented. RFC numbers, titles, and dates were verified against the published standards.-
Postel, J. (Ed.) (1981). Internet Protocol — DARPA Internet Program Protocol Specification. RFC 791. Internet Engineering Task Force. Defines the fourth version of the Internet Protocol and its thirty-two-bit address field.
-
Deering, S., & Hinden, R. (2017). Internet Protocol, Version 6 (IPv6) Specification. RFC 8200. Internet Engineering Task Force (July 2017; obsoletes RFC 2460). Specifies the sixth version as “a new version… designed as the successor” to the fourth, widening the address field from thirty-two to one hundred twenty-eight bits, and coexisting with its predecessor through dual-stack, tunnelling, and translation rather than in-place replacement.
-
Deering, S., & Hinden, R. (1998). Internet Protocol, Version 6 (IPv6) Specification. RFC 2460. Internet Engineering Task Force (December 1998; obsoleted by RFC 8200). The earlier IPv6 specification, identifying IPv6 as the successor to IPv4 and recording the 32-to-128-bit address expansion.
-
Nakamoto, S. (2008). Bitcoin: A Peer-to-Peer Electronic Cash System. The founding specification. Establishes the integer accounting and the consensus rules every node validates independently.
-
The reference protocol’s value model. Amounts are recorded and validated as fixed-width signed integers (a count of satoshis, one coin = 100,000,000 satoshis) and checked by every node against a fixed money-range ceiling. Redefining the unit alters this shared format and the ceiling, breaking validation between old-rule and new-rule nodes — a hard fork. Primary/technical source.