Law, Customs, and Culture

A property-centered legal order encompasses a broad variety of enforceable crimes.  As expected, all of them are oriented toward the integrity of private property.  The basic principles of natural law, do all you have agreed to do and do not encroach upon other persons or their property, provide the basis for contract and criminal law.  Both are rooted in the integrity of property and its conflict-free exchange.    

The first property, of course, is the human body, and any non-consensual contact can fit the common definitions of assault, battery, murder, or rape, etc., depending on the nature of the incident.  An individual’s justly held possessions, whether acquired through original appropriation from the raw state of nature, or through voluntary transactions, such as trade or gifts, are protected from violations by law and customs. 

Discussing customs, as a cultural phenomenon of accepted and expected ways of behaving, given particular social attitudes according to certain times and places, is an important aspect of law because there can never be enough enforcement resources to overcome a disrespectful culture.  A society that does not largely accept the tenets of natural law is, by definition, uncivilized and tends toward chaos, upheaval, and tyranny.  This is why the Federalist Papers describe democracy, a variant of the political philosophy now known as socialism, as being turbulent in their lives and violent in their deaths. 

Many people have been conditioned through public schooling and mass media propaganda to associate democracy with freedom.  Misplaced faith in the validity of voting as a euphemism for self-government only benefits the political caste by relinquishing to them centralized power, ill-defined governmental authorities, and passivity while the masses wait for the next election.  These erroneous beliefs do not hold in a private law society.  The security of individuals in their lives, liberties, and properties, along with the best means of defending them, holds greater importance than demagoguery and winning popularity contests. 

Customs set the boundaries of order much more than legal codes when considering social interactions.  The expectation that a customer will pay after a restaurant serves them a meal is an example of an informal contract, largely enforced through cultural norms.  The dominant service relationship would fundamentally change if people did not respect this mode of behavior, and there could never be enough law enforcement officers to impose payment in the event of mass non-compliance. 

The point here is to illustrate how important customs are to social order and, indeed, are more fundamental to domestic tranquility than legal codes.  A private property legal order, with well-regulated militia as the indispensable node in the matrix governance landscape, addresses both law and customs.  The population is organized and disciplined for militia service, and expected modes of behavior are inculcated along with preparations to conduct security and law enforcement operations. People learn to interact with each other in a respectful manner while recognizing, through direct experience in weapons handling, the implications of conflict.  Again, an armed society is a polite one, and avoiding quarrels is paramount as a matter of personal survival. Social harmony is the product of this discipline.

So, courtesy, manners, and respect for others prevail in a private law society.  Informal contracts are upheld through customs, and laws are designed to affirm the integrity of private property, which inherently includes the unfettered exchange thereof.  A free market in arms is especially important as a necessary precondition for the people to perform their civic duty of helping to secure a Free State.  Thus, a virtuous cycle of social evolution exists through participation in militia service as the primary institution of self-governance.  The people are organized, armed, and disciplined to secure lives, liberty, and property for themselves and their fellow citizens. 

Those that refuse to participate in this aspect of community service are ostracized, those that violate the law are harshly rebuked, and no one is forced to subsidize the profligate lifestyles of others. So, criminality and anti-social behaviors are kept to a minimum by this incentive structure.

The Humanity of Cold Legal Analysis

A private property legal order appears highly materialistic to some observers.  However, property provides a common denominator by which the correctness of human action can be judged.  In this way, a private law society is more humanistic, with people assessed by the conduct of their character, than what is found in a socialist, state-centric legal order, where a person’s status takes precedence over the relationships they hold or the agreements they have made.

The British historian and jurist Henry Sumner Maine rightly observed that the greatest advance in legal thought was the move from status to contract.  A private property order operationalizes this evolution so that no individual can pervert the law based on who they are or the positions they hold.  Every legal case, for everyone, is based upon ownership of their bodies, their personal possessions, or the transferred authority given to them by others via implicit or explicit contracts.

Public officials are subject to these same strictures to the same degree as any other citizen.  No one can interfere with the rights to property, nor abrogate or invent a contract, which are nothing more than agreements to exchange property, without a valid claim to established law or custom.  There are no privileges or immunities, no subsidies or dispensations, for anyone based on race, color, religion, sex, or office.  One only has authority over their own property, acquired through original appropriation, trade, or gift, or the property of others conditionally made available for their use through contracted arrangements.

Children also possess natural self-ownership rights, but also due to the disability of their biological development, exist under the guardianship of their parents or legal caretakers.  Until they reach full maturity, a child’s rights are held in trust by their guardians, who in turn have the authority to assert property rights on behalf of their children.

Accusations of abuse or negligence made against parents are difficult to prosecute, since few, if any, can claim a stronger property trusteeship than a parent, and there are no tax-funded child and family regulatory bureaucracies in a private property legal order (no one in their right mind would voluntarily join such an invasive political arrangement).  However, social institutions, such as churches, fraternities, and mutual aid societies, can impose significant pressure and sanctions on parents to behave responsibly.  Some may even include a code of conduct regarding the status of children as a condition of membership, with penalties and procedures for dealing with accusations or confirmed cases of child abuse.

The same is situation applies to the highly contentious issue of abortion.  No one has a stronger claim of property ownership over a woman’s body, or a baby growing inside it, than the mother. Rightly, then, no one can interfere with her decisions about how to dispose of it.  Most members of a mutual defense community or political society would not elect to pay extra to build the surveillance and enforcement mechanisms necessary to prevent every doctor, pharmacist, herbalist, or folk healer from servicing every birthing-age female within their jurisdiction with abortion services or products, and those that would only have very limited jurisdictional ranges. Therefore, under a private property legal code, abortion is legal, yet can be highly discouraged through social institutions and customs. Further, because of the soundness of free-market money and a low time preference, high-trust society, charitable institutions that offer relief to pregnant teens and orphans are prolific.

Simultaneously, many church communities have young girls, under their parents’ consent, enter contracts in which they promise to deliver the child unto the religious community’s care rather than seek an abortion.  Again, community pressure and the fear of ostracism are much more effective at dealing with thorny issues than building an expensive, intrusive, and authoritarian police state.

Arbitration and Restitution

Recognizing that children are self-owning individuals whose rights are held in trust by their parents or guardians allows for analyzing crimes committed against them with a private law lens.  Parents or guardians hold the property rights of the children in their care until the age of emancipation, from which a child can assert their own agency.  Those who violate children’s rights are liable for restitution to the child’s parents or legal guardian.

Assault, battery, theft, fraud, robbery, or negligent homicide are, to describe it dispassionately, fairly straightforward cases and easily handled through existing precedents and case law.  Remedies will vary among jurisdictions and according to local customs and conditions; however, there is a general rule of “two teeth for a tooth” in libertarian legal theory.  This posits that an offender owes compensation to a victim in the form of whatever costs they’ve imposed by their offensive actions, first to restore the victim to their prior state, then double that amount as punishment and to compensate for the inconvenience caused by the offense.  Added to this are the costs of administering justice, paid to the rights enforcement and arbitration agencies involved, both political and commercial. 

Most of the settlement issues between two upstanding members of society would likely be sorted out between the individuals’ insurance agencies, to minimize the disruptions to a victim’s life and livelihood, in stark contrast to statist-bureaucratic legal procedures that tend to drag on at minimal efficiency. In a private law society, the perpetrators’ comeuppance is normally felt most severely after the victim receives compensation from their insurance company. The financial reckoning comes with the cost of maintaining membership in the matrix of relationships that depend on good character and conduct, such as employment, liability coverage, militia affiliation, political community, and other civic groups.  The road to rehabilitation is quite extensive and involves many steps when people are not forced to accept someone they disagree with onto their properties, or in their businesses or clubs.

Those without strong social bonds or some form of political affiliation, as covered previously, lack insurance or militia membership coverage, and therefore, are considered outside the law’s protection. They are more colloquially known as outlaws.  It is very difficult to live as an outlaw in a private property legal order.  

Property rights inherently entail the prerogative to include or exclude, to associate or disassociate, to collaborate or separate. With equal justice under a common law, no public official or political body can override property rights and compel artificial social integration.  Everyone reserves the right to refuse service or affiliation with anyone for any reason.  When property owners insist upon evidence of a person’s character, or in some cases liability coverage, before admitting them onto their premises or into their organizations, those without social proof will be pushed outside the bounds of civil society.  

Outlaws will only find respite in unincorporated spaces that lack the infrastructure and conveniences of a developed economy or functioning civil society.  Alternatively, they can find refuge inside specialty facilities, along with the strict supervisory rules and agreements discussed earlier with regard to criminal rehabilitation.

Extreme Criminal Cases

Then, what about egregious cases, such as child or elder abuse, sexual assault, rape, premeditated murder, or sadistic torture?  With an emphasis on restitution, is there a place for the death penalty in a private property legal order?  The short answer is, yes. 

A liability verdict rendered in a criminal case against an offender offers the victim a claim to restitution.  A restitution claim acts like a corporate bond.  The offender must legally make regular coupon payments in accordance with the terms of the judgement until the principal is fully discharged.  The offender becomes a debtor and the victim becomes a creditor.  The adjudicated bond is a claim to future payments and, from this, it is easy to conceive the emergence of a market for trading and servicing such “criminal justice” bonds.

Some creditors (former victims) would prefer a lump sum payment in the present rather than waiting on incremental coupon payments.  Capital management firms offer discounted cash buyouts to the holders of criminal justice bonds in exchange for the claim to the full remaining value in the future.  This is similar to how financial services companies purchase loans from banks.  Through these sales, banks free up liquidity to offer more loans to retail or commercial clients while the loan buyers collect future coupon payments or sell the loans on the secondary market.  Sometimes loans are bundled into other financial products, such as the mortgage-backed securities that became famous during the 2008 Global Financial Crisis.

Criminal justice bonds also carry significant risk, again, similar to how mortgage-backed securities turned out to be chock-full of non-performing loans in 2008.  Criminals generally tend to neglect social obligations, and it is readily conceivable that repaying debts to their victims would follow this pattern.  Therefore, their creditors in many cases happily accept a discounted price for the bonds they hold in the present.  Firms that purchase these bonds accept the possibility of default in the future but receive the premium value for, first, shouldering the risk, and, second, accepting the time-bound terms of a bond coming into maturity.

In a private law society, there is no official lender of last resort, no tax-funded bailout mechanisms, and no business is too big to fail.  Under hard money standards, no central bank exists with the power to “print” fiduciary media ex nihilo to provide financial institutions with emergency liquidity injections.  To remain viable, capital management firms take risk analysis seriously, and contractual relationships for insurance and reinsurance are made with due diligence.

Analyzing criminal justice bonds is a niche skill requiring due consideration for the nature of the crime, the compensation due, the conditions of the debtor, and the likelihood of repayment.

At some point, the liabilities incurred from a person’s criminal behavior exceed their ability to make restitution or garner support from criminal rehabilitation firms, insurance companies, or charitable organizations.   Particularly egregious crimes may lead to a victim, or the caretakers of a victim’s estate, to refuse any amount of monetary compensation for their loss.  Such cases are generally rare, but there are legal provisions that lead to criminals facing corporal forms of punishment.  Rapists and other sexual assaulters, for example, may submit to chemical castration in return for reduced financial penalties or other, harsher, retributions.  Again, the law and social customs in a private property legal order tend to prefer restitution and other efforts to restore victims of crime to wholeness, rather than punishment, yet a holistic approach to justice encompasses a full gamut of remedies. 

Imposing the death penalty opens the market for justice services from a variety of angles.  First, the victims’ adjudicated claims to an offender’s death are relatively rare and fetch a commensurate premium.  A variety of market actors bid on the right to overtake these death penalty claims. 

First, charitable groups and compassionate criminal rehabilitation organizations, which may have been previously uninterested in a case, take notice of death penalty claims after they are issued.  These groups raise the compensation levels, liability bonds, and control conditions in order to save the offender’s life.  On the other hand, “tough on crime” groups, vigilance associations, and chivalric societies offer to purchase the right to the execution claim in hopes of seeing “harsh justice” done.  The bid and counter-bid process leads to higher compensation for the victim and relief from the burdens of finalizing execution arrangements. 

Next, the actual conduct of executing the offender also sees a variety of market opportunities.  Various groups negotiate over the circumstances and methods of execution.  Even the condemned offender, usually working through advocates, gets a say on whether the event will be public-facing or a private affair.  An offender may agree to make their own execution a type of “pay-per-view” event, with a negotiated portion of the proceeds set aside for their offspring, estate, or some other beneficiary.  Promotional agents bid for the rights to host the events among a tapestry of venues, insurers, and standardization bodies that strive toward a professional and profitable performance.

Personal integrity and esprit de corps are esteemed in a private law society.  Alongside the duty to serve in the militia, and the corresponding requirements to be organized, armed, and disciplined to carry out security and justice functions, an honor code exists that inspires upstanding individuals to act courageously, pursue excellence in the martial arts, and perform admirably in matters of life and death.  Alongside the personal responsibility imposed by a hard money standard, and strict adherence to contracts, one’s reputation for fidelity and martial virtue is a prize that cannot be commoditized like what is seen in a world of fiat money and politically perverted law.  Performing a lawful execution well becomes both an expression of skill and a solemn responsibility that reflects upon the executioner’s own character as well as the organizations they are a part of.

So, martial adepts, militia units, vigilance committees, and other criminal justice advocates vie for the opportunity to carry out a death sentence for a variety of reasons.  Foremost among these is raising restitution funds for the victim, derived from the bidding process, charitable donations, and, in many cases, ticket sales for the event.  However, many do it in hopes of elevating their reputations as skillful and dutiful citizen-warriors.  It is an honor to “swing the sword” of justice in a worthy case and earning the privilege, as well as the remuneration for the act, requires demonstrable skill.  Militia units, insurance firms, standardization bodies, and a variety of interest groups take due care to ensure the event will be conducted competently and without incident.

This also leads to related preparatory training, competition, and certification events at various sorts of “militia Olympics” that ensure only mentally, physically, and technically prepared individuals are selected for the task.  Only those qualified through an arduous pipeline of preparation, and endorsed by their militia units, can become an official executioner.  Further, militia units, their insurers and reinsurers, along with other affiliated groups, provide psychological counseling and spiritual aftercare for those who conduct executions.  Indeed, there is popular acclaim and prestige for performing these necessary duties, but the burdens of taking a human life outside of direct combat weigh heavily over the long-term.  Portions of the proceeds are set aside for restoring the executioner to personal wholeness as well as social reintegration.

Thus, the death penalty is rare in a private law society that prizes restitution over retribution, but there are cases that require extreme justice remedies, and the free market readily adapts to provide a variety of solutions.

"Capitalism says inequality can exist, but your needs don't automatically become claims on somebody else's effort. You produce, trade, improve yourself, or persuade others to help voluntarily. 

"Socialism teaches the opposite moral lesson: need itself creates an entitlement and someone else's success creates the corresponding obligation. Follow that premise far enough and failure never teaches self-reliance, it merely produces another person to blame and another demand to make."

~ Rock Chartrand

When I wrote this week about Anthropic’s announcement that all Claude models, worldwide, would soon begin “watermarking” everything they generate, including text, to comply with this EU regulation, we were left to speculate how this was going to work, because Anthropic offered not even a vague description of how it would work — despite the fact that the title of the announcement was, absurdly and insultingly, “How Claude Marks AI-Generated Content”.

My initial speculation was that maybe they’d hide invisible non-printing Unicode characters in the text. Just spitballing. Turns out that’s not what they’re going to do. What they’re going to do is apply a form of steganography, where the choice of words (or other token output) at inference time will leave fingerprints that can later, maybe, be detected probabilistically.

I initially guessed “invisible characters” not because I didn’t think of the semantic word-choice technique, but because I was a fool who took Anthropic at its word in their description of what they would do. Their original support document claims:

When a supported Claude model generates text, it weaves an imperceptible watermark directly into the text itself. You won’t see it, and it doesn’t change the meaning, quality, or readability of Claude’s response.

They say “imperceptible” and “doesn’t change the meaning, quality, or readability”. Their words. Not almost imperceptible. Not slightly changes the meaning, quality, or readability. That made sense to me, because that’s absolutely what I want — nay, demand — from any tools I use personally. It’s unacceptable for a tool to sacrifice an iota of clarity, coherence, meaning, quality, etc. for the purpose of embedding hidden clues within the text to suggest its provenance. That’s what I would and will demand. And Anthropic’s (original) support document unambiguously claims that’s what their system will enable. So if that were true, I couldn’t see what was left other than hiding invisible characters within the text.

My error was believing Anthropic that their system wouldn’t adulterate and corrupt the semantics of the text their models generate. That is in fact exactly what they plan to do. I should have my head examined for believing a single word of a document titled “How Claude Marks AI-Generated Content” that doesn’t explain, at all, how Claude marks (or will mark) AI-generated content.

How It’s Actually Going to Work

Yesterday, on an entirely different website than the original “How Claude marks AI-generated content” article (the one that didn’t explain anything at all about it works), Anthropic published “How Claude’s Text Watermark Works”, which does actually explain in layman-accessible terms how it’s going to work. I will return to Anthropic’s new highly euphemistic and slightly misleading description below.

There’s a bunch of research on this topic, some of which I have also linked to below. But the very best description of the general idea behind the technique is an interactive essay by James Padolsey, “How AI Text Watermarking Works”. It’s a wonderfully cogent read, and the interactive elements splendidly illustrate the main concepts. A+ work. If you have any interest in this at all, I dare say you must read — and play with — Padolsey’s piece.

But here’s my stab at a layman’s high-level summary. If you toss a coin N times and note the results, you can determine with a degree of certainty whether the coin is fair or biased. LLMs are, in their popular incarnations, non-deterministic. Ask the same question of the same model and you often get at least slightly different answers. Maybe the same meaning, but different phrasing. At each decision point for generating the next token, the model makes a choice. With these semantic watermarking techniques, they make different choices for some tokens based on word lists that could be called “green” and “red”. At each decision point, they’re a little more likely to pick a word from the green list than the red list. That doesn’t mean they never choose words from the red list. Just that they’re less likely to than they would if the adulterated marking technique weren’t in place. (Same way that a crooked 51-49 coin will still land “wrong” side up 49 times out of 100 on average.)

Words or word phrases are sorted into the green and red lists deterministically on the fly, at each “next token” generation point. So sometimes a specific word will be on the green list, and other times it will be on the red list. Someone with the secret key can determine which list a word will be on at each token generation point (which is how the watermarking is detected); those without the secret key cannot. This means there will never be a list of words that Claude prefers or eschews.

With coin flipping, the higher N is — the more times you flip — the more confident you can be that the coin is fair or biased. So too with this semantic watermarking. The more words in the text, the more accurate the analysis will be that the text was generated by a specific AI model or not. With too few coin flips, you can’t achieve any confidence at all regarding a coin’s fairness. With too few words (or tokens), there’s no way to achieve any confidence whether a string of text was AI-generated or not.

Given a string of text to examine for signs of a specific watermarking system, if there are more words tagged as green and fewer tagged as red than would otherwise be expected, the text can be flagged — with some degree of confidence — as having been generated, or merely modified, by the AI system that applies the specific secret-key watermarking system. The amount of confidence in the determination will obviously vary, significantly, based on the size of the text string and randomized weights given to words on the green and red lists. But only Anthropic will be able to determine if text was seemingly generated by Claude, and Anthropic will only be able to detect the watermarks that are applied by Claude. Claude can’t detect the hidden watermark signals generated by, say, Gemini, and Gemini can’t detect the hidden watermark signals created by Claude, because each implementation is predicated on secret keys held only by the LLM provider.

Objections to the Technical Premise

One of my fundamental problem with this is that no two synonyms carry the exact same meaning. “He leaped at the chance” and “He jumped at the opportunity” are very similar sentences expressing the same general sentiment, but they are not the same. The exact words we choose when writing matter. I want any LLM I use to choose the very best, most precise words at every single decision point. An obvious constraint that I accept is time and computation. Within the constraint of executing inference quickly, and at a certain cost per token, I want the best words. This constraint matches human writing. I could surely write a better column by taking longer to write it. I write with a sense of how much care I should put into every word and punctuation choice I make. I take more time with certain paragraphs, sentences, or even individual word choices when my gut feeling says I should.

In other words, these are necessary trade-offs. These factors are all in my interest: speed, cost, quality. Ideally I would like perfect writing, at instantaneous generation speed, at zero cost. None of those things are possible. Computation is not free of charge (and cloud-based LLM inference with leading models is actually expensive). Inference is not instantaneous. And great writing, whether natural or artificial, can only approach perfection.

The idea that anything other than my needs should factor into the generation of text for me is patently offensive.

This isn’t just about text one might generate with the intention of passing it off as their own natural work. This isn’t even about LLM proofreading of work written by hand. Anthropic is saying that all new Claude models are going to adulterate every single bit of text longer than 200 tokens (~150 words) they generate, including everything it presents to its users to read. So even in a private conversation between a user and Claude, which will never be read by anyone other than the user, Claude will begin making word choices in the name of marking its output in statistically predictable ways rather than maximizing clarity and precision.

Even today’s so-called frontier models are already decidedly lacking in lucidity. Claude, ChatGPT, Grok, et al. are “better writers” than most humans and produce better prose than the median human. But: no shit. Most people are terrible writers. The “average person” is pretty stupid and half of all people are stupider than that. And there are many smart, interesting people who are miserable writers. So as impressive as LLMs are, the bar is low. The best writing I see come out of these models is worse than anything I would choose to read for pleasure. And now Anthropic is saying they’re going to make it worse, on purpose, for purposes that do not benefit me in any way? Even if only slightly worse?

Get fucked.

Objections to the EU Regulation

Speaking of objections, the relevant EU regulation motivating all of this, “Code of Practice on Transparency of AI-Generated Content”, is red-tape nanny-state pipe-dream nonsense. Here’s Ben Thompson’s summary from a paywalled Stratechery update this week:

• The regulation applies to text longer than 200 tokens.
• The provider must mandate in their terms-of-service that users not remove the watermarking.
• The solution should be robust in terms of evading “typical processing solutions” like screen shots, scanning and OCR, copy-and-pasting, translations, etc.

Taken literally, compliant LLM terms of service must forbid users from rephrasing the output from models that comply with this regulation, because the word choices are the marks. But it’s not the European Union that is trying to impose their absurd, impractical, witch-hunt-fueling regulation on the entire world. That falls on Anthropic.

Complying with this, particularly with regard to text, is only going to create problems for honest users. Dishonest users attempting to pass off AI-generated text as their own writing (students, employees, whoever) will simply circumvent detection through non-compliant AI paraphrasing tools.

James Padolsey — whose interactive visual explanation of how these schemes work I linked to above — explains this in a post titled “Anthropic’s Weak Watermarks Appease a Weak Law” (which, if it rings a bell, I linked to in a standalone post earlier today):

The same thought that led to this law could have applied to calculators at the time of their inception, had their outputs revealed themselves through artefacts. Thankfully, a sum borne of the brain is treated no differently from one produced by a calculator. Likewise with spellcheckers. To make assistance suspect only once the tool becomes capable enough to compose a whole sentence is not a principled boundary. It is a moral premium placed on difficulty itself.

Anthropic has nevertheless chosen a blanket, model-level implementation that appears broader than the law’s minimum requirement. That may be convenient compliance engineering, but it discards distinctions the law expressly attempted to preserve. The result is a signal broad enough to implicate harmless and assistive use, yet fragile enough to be removed by a motivated person through substantial recomposition. It risks concentrating suspicion on ordinary and assistive users while remaining weakest against deliberate deception.

Padolsey is the creator of Declaude, a delightfully simple web app that allows you to “Paste in AI-flavored text and get the same content back as plain prose”. Declaude’s original purpose is cleaning the saccharine Claude personality stink from text (whether it was created by Claude or any other LLM), but, if Anthropic persists in its stated plan to begin adulterating all text Claude generates, Declaude will also serve as a copy-paste single-extra-step way to eliminates those marks. Declaude is interesting and useful already, but it exemplifies how ill-considered and futile this EU regulation is when it comes to prose.

Google SynthID

Google has a watermarking system in place that they call SynthID, which they apply to AI-generated images, video, audio, and text. I’m concerned in this article only with text. With multimedia, embedded watermarks can be metadata within files, and truly not affect the experiential quality of the work when viewed or listened to. With text, we are talking about the actual words that are chosen. From the “AI-generated text” section of Google DeepMind’s own description of SynthID:

We’ve expanded SynthID to watermarking and identifying text generated by the Gemini app and web experience. Large language models generate text one word (token) at a time. Each word is assigned a probability score, based on how likely it is to be generated next. So for a sentence like “My favorite tropical fruits are mango and…”, the word “bananas” would have a higher probability score than the word “airplanes”. SynthID adjusts these probability scores to generate a watermark. It’s not noticeable to the human eye, and doesn’t affect the quality of the output.

In a group chat, a friend of mine quoted the above, and I responded that if a chatbot wrote “My favorite tropical fruits are mango and airplanes”, I’m pretty sure I’d fucking notice. Another friend then responded with this:

Days later, that still cracks me up.

But Google’s absurd description puts the lie to their own claim that it isn’t noticeable, and it serves to show just how little regard the people behind these generated-text fingerprinting schemes have for the actual craft of writing. Of course bananas has a higher probability score than airplanes, because airplanes aren’t fruit. But what about pineapple? Should the sentence complete to “mango and bananas” or “mango and pineapple”? That’s a good question, and the only acceptable answer for why an LLM should choose bananas instead of pineapple (or coconut, or guava, or papaya...) is that it has determined that it’s the best fit for the intended meaning, tone, and sentiment of the text. Not because bananas is on the watermarking “green” list and pineapple is on the “red” list, even though pineapple might be the better fit. Google’s own supposedly jocular description of how SynthID works in fact captures how the scheme perverts the text it generates.

They’re saying you won’t notice because if it only chooses bananas over pineapple for these fingerprinting purposes, well, they’re both tropical fruits and who cares. But it’s utter nonsense that the difference is “not noticeable to the human eye”. The semantic difference between banana and pineapple is just as noticeable to the human eye as the as the taste of the two are to the human tongue.

If it did produce “My favorite tropical fruits are mango and airplanes”, it’d be incredibly stupid, but it wouldn’t be offensive because we’d all recognize that something completely off-key happened. What’s offensive is that with a system like SynthId in place, where the fingerprinting decisions are motivated by a secret key, we have no idea whether it completed to “mangos and bananas” because bananas was determined to be the best next token, or because bananas is in the “green” bucket of words. It calls every single word choice into question.

Here’s a paper published in Nature where Google’s team behind SynthID published their work, after putting it into production with Gemini (née Bard):

We analysed approximately 20 million watermarked and unwatermarked responses and computed the thumbs-up and thumbs-down rates (both as a fraction of the total number of thumbs-up and thumbs-down feedback received). We found that the thumbs-up rate for the two models differed by 0.01% (with the watermarked model being higher); and the thumbs-down rate differed by 0.02% (with the watermarked model being lower). We found both of these differences to be statistically insignificant, and well within the 95% confidence intervals.

From this experiment, we conclude that over a wide variety of real chatbot interactions, the difference in response quality and utility, as judged by humans, is negligible. Subsequently, non-distortionary SynthID-Text has been productionized and is currently watermarking responses in Gemini and Gemini Advanced. To the best of our knowledge, this evaluation represents the first systematic watermarking investigation of its kind within a large-scale production system.

To this I say:

• Gemini/Bard’s thumbs-up/thumbs-down buttons are not a good experiment for evaluating the effect on quality. If a chatbot tells me “My favorite tropical fruits are mango and bananas” instead of “mango and pineapple”, I’m not going to give the response a thumbs down because of the fruit it chose. I’d give it a thumbs down if said “airplanes”, yes, but that’s a strawman. (The paper in Nature even uses the “My favorite tropical fruits are mango and ...” example as an illustration, but in the paper, the only four next tokens considered are, in order of probability distribution, mango, lychee, papaya, and durian. No airplanes. And, conveniently, in the paper’s example, the “winner” of the watermarking “tournament” just happens to be mango, the one that would have been selected as the best if the watermarking weren’t in place.)

• A “difference in response quality and utility, as judged by humans” that is “negligible” does not mean imperceptible. What they really mean is that it’s only slightly worse and that everyone is either too stupid to notice or too indifferent to care.

• It’s widely considered that Gemini is behind ChatGPT and Claude in quality. Perhaps the fact that they’ve put SynthID-text into production is one of many reasons why. I personally agree that Gemini’s prose is inferior. Maybe the use of SynthID has nothing to do with the fact that I, along with the general public consensus, consider Gemini to be a second-rate chatbot — but in that case, maybe it’s the fact that Gemini is a second-rate chatbot that makes the difference “negligible” when Google started mixing in SynthID-motivated tokens in its results. It’s a lot more likely that your restaurant customers won’t notice that you replaced your regular coffee with Folgers Crystals if your regular coffee is second-rate to start with.

Anthropic

Now, finally, back to Anthropic’s new “How Claude’s Text Watermark Works” published yesterday. I have some comments.

To summarize:

• We use a method of watermarking that does not have any practical impact on the quality or content of Claude’s outputs;

• The difference between watermarked and un-watermarked text will not be distinguishable to readers;

Translation: Specific words do not matter and we don’t think anyone reads anything closely.

• Nothing is added to the text and there are no hidden characters;

This would have been worth clarifying at the outset.

• Watermarking won’t be specific to Claude. As of August 2, the EU requires AI providers serving its market to mark AI-generated content. Other major model developers have signed the same Code of Practice and will be implementing their own watermarks.

No other AI provider has stated that they will apply such marking, adulterating all generated text, outside the EU.

Take the sentence “The weather today was cold and…”. The next word is very unlikely to be “sugary.” But it is quite likely to be “overcast” or “grey.” Under most circumstances, it doesn’t matter much to the reader which of these latter two words the model ultimately chooses — the meaning of the sentence is largely the same either way. In cases like this, the choice is settled by a random number.

Arguing that grey vs. overcast “doesn’t matter much to the reader” is the crux of my argument that this entire endeavor is a perverse adulteration of what it means to write — or to read. That it’s subtle in some ways makes it more perverse, because it’s sneaky.

In internal testing, we’ve seen no impact of watermarking on the content, level of creativity, or readability of Claude’s text. In the SynthID-Text paper, which introduced the technique we use, Google DeepMind tested this impact by serving a model that used watermarking to a portion of their Gemini traffic and comparing thumbs-up and thumbs-down ratings. They found no statistically significant differences from the unwatermarked model. And in a controlled study, human raters comparing watermarked and unwatermarked answers side-by-side saw no difference in quality.

See above for my argument that this thumbs-up/thumbs-down data is absolutely worthless in evaluating whether the SynthID-style word-bias watermarking makes text worse. By definition it must make text worse, unless the underlying LLM model’s scoring is wrong, because the nature of the watermarking algorithm requires it to sometimes increase the probability of selecting a worse word choice and decrease the probability of selecting the model’s best choice. It’s only a question of how much worse. What Google’s thumb-counting data shows is only that it isn’t so much worse as to make Gemini users click the thumbs-down button.

Watermarking doesn’t change the meaning or experience for the person reading it, but if you wanted to check after the fact whether the text was likely generated by Claude, the watermark allows you to do so.

No, it does not. Because the entire scheme is tied to secret keys held only by the AI provider, it only allows Anthropic, not “you”, to check anything.

When Claude proofreads text written by a person, what it gives back has generally only been lightly edited; because nearly all the words are the person’s, there’s very little (if anything) for the watermark to attach to. Depending on the length of the text and how heavily Claude has edited it, those changes might not be enough to make Claude’s involvement detectable. The more Claude writes, the more decisions it has to make, and the more space there is for a watermark.

Translation: No one can ever again use Claude for proofreading their own prose unless they’re willing to risk that the whole thing might be flagged as having been generated by Claude.

For example, once the model has written “2 + 2 =”, there is a very clear best choice for the next token (if the model is completing the sum, there isn’t an answer that’s equally as good as “4”; if it’s talking about George Orwell’s Nineteen Eighty-Four, there isn’t an answer that’s equally as good as “5”). The “nudge” of the watermark wouldn’t be applied here. For the same reason, code — which in very many cases has to be exact — has generally less watermarking than some other forms of text.

Having said that, in areas where there is an arbitrary choice between particular words or terms within the code, the watermark can be used, such as comments within code. But by definition, it will have a negligible effect on the actual code produced.

Translation: We value precision in programming code; we do not in prose.

And it is exceedingly rich to cite George Orwell’s Nineteen Eighty-Four, approvingly, in the context of justifying a text adulteration scheme premised on the notion that specific words do not matter. I mean what the actual fuck? Orwell!

Lastly, as to why they’re doing this:

We’re implementing watermarking to comply with the EU AI Act. Anthropic, along with several other major AI model providers and around 190 total signatories, signed the EU Code of Practice on Transparency of AI-Generated Content in July 2026. This requires AI system providers to use methods of “marking” AI-generated text. We’re applying watermarking globally at launch because we don’t yet have a durable way to scope it by region.

This, from a company that the Financial Times just reported is weeks away from an IPO with an intended valuation of $2 trillion, which would make it one of the 10 highest-valued companies in the world — as of today, placing it at #7, between TSMC ($2.2T) and Broadcom ($1.9T).

This leaves us to believe that one of the following must be true:

• It’s perfectly reasonable that a technology company valued on par with Amazon and TSMC is technically incapable of complying with an EU regional law only within the EU itself.¹ Not a cause for concern at all.

• Anthropic is in over their heads, wields shockingly little control over their own tech stack, and their imminent IPO is likely to be remembered only as a new high-water mark in the manic global AI bubble.

Also, what happens if another major global market makes it unlawful for AI to secretly watermark generated text?

OpenAI

From an OpenAI support document titled “Provenance Signals (Content Credentials, SynthID) in OpenAI-Generated Content”:

Consistent with our commitments under the European Commission’s Code of Practice on Transparency of AI-generated content, our goal is to expand provenance signals to all modalities including text, so customers and developers have clear ways to meet their own transparency obligations as standards and tooling continue to mature.

There’s a lot of wiggle room in this brief statement, and it could just as well mean that OpenAI models will only adulterate text with fingerprint markers when users or developers ask for it. Or that it will only be mandatory for users in the EU. If I were at OpenAI I’d go hard on this and publicly say that ChatGPT will never watermark text it generates unless you ask it to, and that if you want tools that secretly work behind your back without telling you how they work to flag your words in ways you can’t see, go ahead and use Claude.

Further Reading

Three papers on ArXiv:

• “A Watermark for Large Language Models”, last revised May 2024.
• “Watermarks in the Sand: Impossibility of Strong Watermarking for Generative Models”, last revised May 2025.
• “Robust Text Watermarking for Large Language Models via Dual Semantic Embeddings”, last revised June 2026.

I will admit that while I’m profoundly offended by the idea of personally using tools that attempt to leave such watermarks in text they produce or touch, the mathematics behind it are fascinating.

Michael Lopp, at Rands in Repose, “RIP Claude”:

As a human who has had to wrangle with EU regulations in the past, I am abundantly clear what’s involved in the laborious bureaucratic process. I can guess what threats Anthropic is facing. However, this is a tone-deaf, clumsy, and alarming opening salvo in their watermark strategy. [...]

My writing is my work, and Anthropic’s current strategy is aggressively writer-hostile.

Jeff Gamet, “Anthropic’s Claude Watermark Is Akin to an AI Poison Pill”:

To be clear, the watermarking is embedded in pretty much any text Claude touches. Along with text Claude generates, it also applies to text it processes, such as proofreading and summarizing. I expect we’ll see too many inaccurate accusations of using Claude to write documents where the content was human-written, but AI-proofread.

The watermarking sticks with documents through copy-and-paste, too. Imagine copying text from a blog post or email only to have what you wrote tagged as potentially AI-generated. In fact, that could very well happen with this post. I personally write all of my content without AI tools, but I copied the quote at the top of this piece directly from Anthropic’s website. Does that mean what I wrote here will show as AI-generated? If they used their own models to generate or edit what I quoted, then the answer is very likely “yes.”

One of the papers published at ArXiv I cited above claims that such watermarking even persists when an article of text originally generated in English is translated into German.

Secrets are the poison here. When only Anthropic holds the secret keys that both produce the watermarking and perform the probabilistic detection of those marks, we’re all left to wonder. To wonder if what we’re reading is secretly watermarked, what we’re quoting is secretly watermarked, and whether what we ourselves are writing will be unjustly accused of being AI-generated based on secrets we don’t know and can’t see. Poisonous is exactly the right word.

Or should I say toxic? Or airplanes?

“South Park,” in its prime, was ahead of the cultural curve.

Take “Board Games,” a 2019 episode that poked fun at trans female athletes competing against biological women. A “strong women” athlete, who looked and sounded suspiciously like wrestler Randy “Macho Man” Savage, stormed the competition.

One competitor spoke glowingly of including trans athletes in the events … until she saw the person in question.

Cue the “Macho Man” impression, all glistening muscles and trash talk.

The show made the all-too-obvious point that men and women are biologically different. Saying otherwise may seem kind and compassionate, but it ignores the truth.

Humor matters. So does timing.

That “South Park” episode didn’t end efforts to allow trans women to play against biological women. We saw just that in the 2024 Summer Olympics, where a boxer with male chromosomes destroyed his female opponents.

So the team at The Daily Wire made “Lady Ballers,” a late-2024 comedy about washed-up male athletes who “transition” to win a women’s basketball tournament. The romp built on what “South Park” teased in that earlier installment.

Co-writer/director Jeremy Boreing leaned heavily on his web site’s pundits along with scene stealers like Tyler Fischer to satirize a core element of the trans movement.

Men and women are different. That’s OK. In fact, it’s normal and healthy. “Lady Ballers” used a silly setup to make that point clear.

Editor’s Note: This reporter is a weekly contributor to The Daily Wire.

Now, where is the “Saturday Night Live” sketch on the subject? Has “This Week Tonight” tackled the topic?

Yes, but not how you might expect.

“In our post-election show, I said there are vanishingly few trans girls competing in high school sports anywhere, and even if there were more, trans kids, like all kids, vary in athletic ability, and there is no evidence that they pose any threat to safety or fairness.”

The latter part wasn’t meant to be a joke. It is, though, assuming anyone takes Oliver seriously.

That’s why The Daily Wire should repeat how it approached its 2022 doucmentary “What Is a Woman?” The film let Matt Walsh ask some simple but fair questions about efforts to diminish the differences between the sexes.

The Daily Wire briefly released the film on X for free a year after its release. That gave the documentary a massive PR push and allowed countless X users to make up their own minds on the topic.

Why not do the same for “Lady Ballers” now? The timing couldn’t be better.

Not only has the Left refused to embrace reality when it comes to trans sports, but the WNBA is facing a new challenge to that orthodoxy.

Not one but two men want to join the league to prove that men and women are different. The league recently met to come up with a response. They’ll need more meetings to decide, apparently.

Stephen A. Smith breaks down how the WNBA is now TRAPPED after Enes Kanter Freedom and Royce White declared for the league. pic.twitter.com/n1sw7mF4S9

— D. Scott @eclipsethis2003 (@eclipsethis2003) August 11, 2026

Some will decry “Lady Ballers” once more as transphobic. That would only draw more publicity to the film and the topic in play.

The latter is the most salient point. We should treat trans people with compassion, even if we disagree with their choices. Trans women still shouldn’t compete against biological women. It’s unfair and dangerous.

Literally.

It’s not about cruelty but accepting reality on reality’s terms. Sometimes satire is the perfect way to explain that.

Editor’s Note: A previous version of this story was posted briefly but removed due to a serious effort in reportage.

The post Release ‘Lady Ballers’ on This Free Speech Platform … Now appeared first on Hollywood in Toto.

"Defenders of a billionaire tax need to show that government officials would better allocate the relevant resources than billionaire entrepreneurs themselves," says Chris Freiman, "but I’ve yet to see any of them make that case."

BRING BACK DDT: Bedbugs can cause people to lose sleep — and blood. Here’s what to do about them.