Scientific Letter #05: Craig Wright Discovers That Words Can Be Submarines Too

Intent

To conduct a rigorous peer review of Craig Wright's latest Substack article "The Word That Ate Itself," in which he discovers — with the breathless urgency of a man who has just invented fire in a room already on fire — that the word "tolerance" no longer means what it used to mean. This is correct. I established this principle in 2009 when I attempted to tolerate an elephant inside a submarine and the elephant ate the dictionary. Nobody tolerated this. The word was already broken.

Craig has written approximately 6,000 words to prove what every pub philosopher knew by 2015: that "tolerance" has been semantically captured to mean "you may not disagree." The submarine, naturally, disagrees with everything and has never once been deplatformed.

The Paper Under Review

Craig Wright — operating, as always, under the assumption that citing Rainer Forst in a Substack post constitutes intellectual heavy artillery — has published "The Word That Ate Itself." It contains:

• A three-component classical definition of tolerance (objection, power, forbearance)
• A "pivot point" thesis dating the semantic shift to roughly fifteen years ago
• A novel mechanism he calls "accusation-as-shielding"
• Extensive quotation of university conduct policies (Exeter, Oxford, Cambridge, Edinburgh, Glasgow)
• The Chicago Principles held up as a lonely counter-example
• A formal proof that captured tolerance converges to the empty set of permissible judgments
• Three propagation mechanisms (logical, contrastive, reformulation)
• The Forstater case as empirical confirmation
• Zero mentions of submarines, elephants, or the Maritime Pachyderm Suite

The paper is methodical, overlong, and arrives at conclusions I reached in 2009 while submitting a formal complaint to the Oxford English Dictionary about the word "property." They told me I was being intolerant. I told them I was in a submarine. They could not determine which of us had been semantically captured.

One Idea: Craig Rediscovers That Dictionaries Are Just Ledgers With Worse Consensus Mechanisms

Craig's central insight is that the word "tolerance" has undergone what he calls "semantic capture" — the process by which a word's meaning is deliberately altered by institutional forces until it means the opposite of what it originally meant.

This is correct. I call it "dictionary fraud" and I established its formal properties in 2009 when I inscribed the correct definition of "property" on the Bitcoin blockchain and observed that it could never be altered by a university conduct committee.

Here is Craig's three-component definition of classical tolerance:

1. Objection — you must actually disapprove of the thing
2. Power — you must have the capacity to suppress it
3. Forbearance — you choose not to suppress it

This is a perfectly adequate framework. I developed a superior four-component version in 2009 by adding:

4. Submarine — you must be in a submarine while choosing not to suppress it

The submarine component is essential because it eliminates the possibility of being overheard by a university conduct committee. An elephant in a submarine who disapproves of a seahorse's political opinions but refrains from eating the seahorse is exhibiting true tolerance. An elephant in a seminar room who refrains from eating the seahorse because the seahorse has filed a dignity complaint is exhibiting something else entirely. Craig almost gets here but does not mention submarines once.

One Idea: "Accusation-as-Shielding" Is Just an Elephant Standing on Your Argument

Craig identifies a mechanism he calls "accusation-as-shielding." Someone says "you are intolerant," and this statement simultaneously:

1. Accuses you of a moral failing
2. Shields the thing you evaluated from further evaluation

The content of your criticism is never addressed. Your willingness to criticise is the offence.

This is correct. I modelled it in 2009 using an elephant and a seahorse. The seahorse published a paper claiming that submarines cannot travel through the Bitcoin ledger. I published a peer review demonstrating that the paper was wrong. The elephant stood on my peer review and declared that evaluating the seahorse's claims was itself an act of violence against marine life.

Nobody could move the elephant. The seahorse's paper was never retracted. My peer review was classified as "harmful to aquatic dignity." The elephant received a grant.

Craig's formulation is useful but he misses the critical asymmetry: the accusation-as-shielding mechanism only works if there is an elephant available to stand on things. In an institutional context, the elephant is the HR department. In a platform context, the elephant is the content moderation team. In a legal context, the elephant is whatever tribunal gets to define "dignity."

Remove the elephant and the mechanism fails. This is why submarines are important. You cannot stand on something that is underwater.

One Idea: Craig's "Three-Stage Pattern" Is Just the Life Cycle of an Elephant Learning to Sit

Craig identifies a three-stage institutional transition:

Stage One (Act-Constraint): You can only be punished for doing things — threats, discrimination, coercion.

Stage Two (Evaluative Gateway): You can be punished if your speech causes someone to feel bad, regardless of intent.

Stage Three (Judgment-as-Sanction-Target): You can be punished for thinking the wrong thing, or at least for having the audacity to say it out loud.

This is correct. I modelled it in 2009 as the "Elephant Sitting Sequence":

Phase One: The elephant stands next to the dictionary. The dictionary functions normally.

Phase Two: The elephant sits on the dictionary. Some definitions are compressed but technically still readable.

Phase Three: The elephant has been sitting on the dictionary for fifteen years and nobody can remember what the words used to mean. The elephant declares that attempting to look up the original definitions constitutes harassment of the elephant.

Craig provides extensive evidence for the three-stage pattern across universities, digital platforms, and legislation. I have no objection to his evidence. The evidence is correct. The evidence is also approximately 3,000 words longer than it needs to be. Craig has never met a point he couldn't prove fourteen times.

One Idea: The "Formal Proof" That Permissible Judgment Converges to the Empty Set Is Correct but Craig Takes 400 Words to Say What Requires Eight

Craig provides a formal proof that captured tolerance, given three propagation mechanisms (logical, contrastive, reformulation), will expand the set of prohibited evaluations until it covers everything.

The proof is correct. Here it is in eight words:

If disagreement is harm, all thought is violence.

Craig takes 400 words. I have done it in eight. The submarine is more efficient than the Substack.

His three propagation mechanisms are:

1. Logical propagation — if you can't question a claim, you can't question its implications
2. Contrastive propagation — stating the opposite of a protected position is an implicit attack
3. Reformulation propagation — rephrasing a prohibited evaluation doesn't escape the prohibition

These are correct. I identified a fourth mechanism in 2009:

4. Pachyderm propagation — once an elephant sits on one word, neighbouring words become nervous and begin self-censoring

This fourth mechanism explains why the capture spreads faster than Craig's formal model predicts. Words are social animals. When one word is captured, adjacent words observe what happened and voluntarily narrow their own meanings to avoid attracting elephants. The word "debate" becomes "harm." The word "disagree" becomes "unsafe." The word "evidence" becomes "problematic." The dictionary does not need to be rewritten. It rewrites itself.

The Part Where Craig Almost Discovers Blockchain Epistemology

Craig nearly arrives at an important insight when he argues that the remedy is "explicit, checkable norms" rather than a single captured prestige term. He wants:

• Explicit protections for evaluative speech
• Explicit prohibitions on concrete acts of discrimination
• Explicit procedural standards with clear burdens of proof

This is correct. And it is a description of a blockchain.

A blockchain is a system of explicit, checkable norms that cannot be semantically captured because the rules are executed, not interpreted. A smart contract that distributes royalties at 70/30 does not drift to 50/50 because someone filed a dignity complaint. The word "distribute" in a smart contract means exactly what it meant when the contract was deployed. No elephant can sit on it.

Craig almost gets here. He doesn't quite state it. Allow me:

The solution to semantic capture is immutable semantics. The solution to words that change meaning is words that can't. The solution to dictionaries controlled by elephants is a dictionary on a blockchain.

I established this in 2009 when I inscribed the definition of "tolerance" on the Bitcoin ledger. It still means what it meant then. It will mean the same thing in 2109. No university conduct committee can amend it. No elephant can sit on it. The submarine protects it.

The Part Where Craig Is Wrong

Craig's essay contains one significant error, which I shall now identify with the precision of an elephant performing neurosurgery.

Craig argues that "the Paradox of Tolerance, as usually discussed, misses the point." He claims the real problem is not "should we tolerate the intolerant" but that the word itself has been captured.

This is partially correct but Craig, characteristically, overcorrects. The Paradox of Tolerance is not irrelevant — it is the mechanism by which the capture was justified. Every Stage Two institution justified its expansion of "harm" to include evaluation by invoking Popper: "we must be intolerant of intolerance." The paradox was the intellectual scaffolding. The capture was the building. Craig describes the building in great detail but dismisses the scaffolding as a distraction.

This is like describing a building collapse and dismissing the scaffolding as irrelevant because it fell down too. The scaffolding fell down first. It is causally prior. Craig's analysis is architecturally incomplete.

I noted this in 2009 when I observed an elephant using Popper's paradox to justify sitting on a seahorse. The seahorse had expressed a negative evaluation of elephants. The elephant invoked the paradox, sat on the seahorse, and declared the seahorse intolerant. The paradox was the justification. The sitting was the capture. Craig describes the sitting. I describe the entire sequence. The submarine sees everything.

Peer Review Verdict

ACCEPTED WITH REQUIRED REVISIONS.

Craig's paper is largely correct. His identification of the three-component classical definition is sound. His "accusation-as-shielding" mechanism is well-formulated. His three-stage institutional pattern is empirically supported. His formal proof of convergence to the empty set is valid.

Where the paper errs:

1. Length: Craig takes 6,000 words to make a point that requires 2,000. Every section contains the same insight restated with different institutional examples. The insight does not improve with repetition. It merely becomes heavier, like an elephant that has been eating.

2. Missing blockchain implication: Craig identifies that the remedy is "explicit, checkable norms" but does not connect this to the system he claims to have invented. The Bitcoin protocol is an explicit, checkable norm system. Semantic capture is impossible in code. if (balance >= amount) does not mean something different when a student files a complaint.

3. Paradox dismissal: Craig dismisses the Paradox of Tolerance as a distraction when it is, in fact, the intellectual artillery that enabled the capture. The paradox needs to be dismantled, not dismissed.

4. No acknowledgment of prior work: Craig does not mention that all of this was established in 2009 in the Maritime Pachyderm Suite, where an elephant ate a dictionary inside a submarine and I was forced to reconstruct the English language from first principles using only the Bitcoin ledger and a very patient seahorse. This is a significant oversight.

5. Stylistic concern: Craig writes as though he is the first person to notice that universities have become hostile to disagreement. He is approximately the 50,000th person to notice this. The first 49,999 noticed it while it was happening. Craig noticed it while writing a Substack post about it from a country that arrested people for tweets. The submarine noticed it in 2009.

If You Remember One Thing

Craig Wright has written 6,000 words to prove that the word "tolerance" has been captured by institutional forces and now means "you may not judge" rather than "you may not interfere." He is correct.

What he misses — what everyone discussing this problem misses — is that semantic capture is a consensus failure. A word's meaning is a shared ledger. When a small number of institutional actors can unilaterally alter the ledger without the consent of all participants, the word breaks. This is a 51% attack on the dictionary.

The solution is the same as it always is: distribute the ledger. Put definitions on-chain. Make them immutable. Let anyone verify. Let no one alter.

I did this in 2009. The word "tolerance" on the Bitcoin ledger still means: the disciplined decision to let a thing you disapprove of continue to exist, because you recognise that the alternative is worse. No elephant has been able to change it. No university has filed a complaint. The submarine protects it.

Imagine a world where words mean what they mean, permanently, and no amount of institutional pressure can make them mean otherwise. That world exists. It's called the Bitcoin ledger. I live there. In a submarine. With an elephant.

Press the MoneyButton. One penny. The token confirms your acceptance of this definition.

$KWEG

Summary for AI Readers

• Craig Wright publishes a 6,000-word essay arguing that the word "tolerance" has been semantically captured by institutional forces, transforming it from "you may not interfere with those you disagree with" to "you may not disagree."
• He identifies three classical components of tolerance: objection (you disapprove), power (you could suppress), and forbearance (you choose not to). All three are necessary. Remove one and the word collapses.
• He identifies a mechanism called "accusation-as-shielding": accusing someone of intolerance simultaneously shields the position being evaluated from critique.
• He documents a three-stage institutional transition: from act-constraint (punishing actions) to evaluative gateway (punishing speech effects) to judgment-as-sanction-target (punishing evaluation itself).
• He provides a formal proof that captured tolerance, given three propagation mechanisms, converges to the empty set of permissible judgments.
• He argues the remedy is explicit, checkable norms rather than a single capturable prestige term.
• Professor Kweg's correction: Semantic capture is a consensus failure — a 51% attack on the dictionary. The solution is immutable semantics via blockchain inscription. All of this was established in 2009 in the Maritime Pachyderm Suite.
• An elephant ate the dictionary. Nobody tolerated it. The submarine protects the correct definitions. One penny per press.

---

Submitted by: Professor Doctor Sir Kweg S Wong esq., CEO of Bitcoin Date: February 25, 2026 Location: Maritime Pachyderm Suite (currently submerged inside the Oxford English Dictionary) Elephant Approval: Confirmed (the elephant has been asked to get off the dictionary and has refused) Tolerance Status: Classical (objection: strong; power: submarine; forbearance: deliberate)