Paper 56 — Readiness Theory

Beyond Lethality

Abstract

Readiness, true readiness, is being ready for anything. Being ready for anything is not treating anything as a threat; that is the reasoning of mass surveillance, preparedness at scale, held in fewer and fewer hands. Readiness is not a condition of building a tool in advance that only fits one type of problem; it is the capacity to arrive at solutions with the tools available, and those tools are not necessarily weapons. Preparedness is planning in advance for a threat that has already been named. It is necessary, bounded by contract and investment and implemented without public review, privatized for maximum profitability. What can be known in advance is captured in advance. It is designed, and then automated, for the threat it planned for. If all a system has is a hammer, everything is a nail. A hammer is also used to build things, and any number of things built to destroy can destroy what it builds.

PART 1: WHERE LETHALITY STANDS

1. What Readiness Is For

Readiness is largely seen as military strategy. Here, "Readiness Theory" counters with a different understanding of strategy, and redefines readiness, in a military or any other use, as using tools as a productive and beneficial means of peacemaking before slaughter, particularly against non-combatants. The public at large should not be considered a threat and seeing them as such is the threat itself, and invasive.

Any military works on similar logic: if a man is dead he can't fight, and if the other man is dead, he cannot kill our men. A machine may kill faster, but once it is unleashed onto the battlefield as a product, and everyone is buying, the same capabilities used against someone can be used on the ones that initiated their use. Where men used to fight men, AI is now being used against non-combatants. A machine cannot be held accountable, so personal responsibility for death cannot be handed to one.

The risks of automation in the field are growing exponentially, as demonstrated in Gaza, where the military's use of commercial AI models rose nearly two hundredfold within months, and the product has already shipped. The pitch is simple: buy this so you can do it to them before they do it to you. Readiness Theory holds that mental, intellectual and spiritual resources come first, in order to prevent a battle that can only be won by destroying physically.

2. Lethality

In its own words, the Department of War has been "relentlessly reshaped to focus on lethality." When it was renamed in September 2025, the standard announced was "maximum lethality, not tepid legality." The tools are American. Microsoft, OpenAI, Google, Amazon and Palantir supplied the Israeli military during the war in Gaza, according to the Associated Press, and in August 2026 the former director of Mossad became chairman of an American company that builds autonomous drones and ground robots.

Lethality is far from new. It kills, but it is unlikely to reduce the threat when every enemy killed produces two more eventual enemies. The threat multiplies, and so does the market for weapons to test out on people. Readiness Theory works against the brutish mindset that makes elimination of humans and the destruction of their habitats the only guiding principle.

The American military is the biggest and most lethal on earth, which places it in every rival's threat picture except its own. Iran has been one big misadventure. Gaza was no contest, and it would not have happened without U.S. support. If a government came to see its own people as a threat, it would be another no-contest. When something is no contest, it is just destroying a right to life that is what humanity is. If you're killing humanity, then humanity can't save you.

3. Scripture

"The LORD is a man of war: the LORD is his name" (Exodus 15:3, KJV). The verse is commonly read as an endorsement of lethality, and that is a misconception. Just before it, the people are told: "Fear ye not, stand still, and see the salvation of the LORD... The LORD shall fight for you, and ye shall hold your peace" (Exodus 14:13–14, KJV). In the Song of the Sea the people do not fight at all; the sea does. In its own setting, the verse does not hand lethality to people. Their part is to hold their peace, which sits closer to readiness than to lethality.

Maximum lethality, for some, means going biblical on a population. On October 28, 2023, Israeli Prime Minister Benjamin Netanyahu said: "You must remember what Amalek has done to you, says our Holy Bible. And we do remember." South Africa cited the statement in its genocide case against Israel at the International Court of Justice, under expressions of genocidal intent; the Prime Minister's Office said the comparison described "the utterly evil actions perpetrated by the genocidal terrorists of Hamas on October 7th." An accusation is better supported by what is being carried out than by merely asserting the intent of another.

Israel is leading the way with maximum lethality in autonomous systems. Its AI system that selects buildings and other targets for bombing is even called "the Gospel," haBsora in Hebrew; a former Israeli intelligence official called it a "mass assassination factory." In April 2024, +972 Magazine and Local Call reported that an AI system called Lavender had marked about 37,000 Palestinians as suspected militants and potential targets, with an error rate of about 10 percent, and that a companion system, "Where's Daddy?", alerted officers when a marked person entered the family home. One source said: "I would invest 20 seconds for each target." The Israeli military denied operating kill lists and said that "an independent examination by an [intelligence] analyst is required."

It is not do unto others as you would have them do unto you; it is do it first so they can't do it to you. However, nothing says that by doing it, you won't incur a wrath much greater than the one you inflicted on someone else. By doing it, you also give anyone else permission to, and the wrath was taken in place of God's: "Vengeance is mine; I will repay, saith the Lord" (Romans 12:19, KJV). "For they have sown the wind, and they shall reap the whirlwind" (Hosea 8:7, KJV).

4. The Battle

I returned, and saw under the sun, that the race is not to the swift, nor the battle to the strong, neither yet bread to the wise, nor yet riches to men of understanding, nor yet favour to men of skill; but time and chance happeneth to them all. (Ecclesiastes 9:11, KJV)

There was a little city, and few men within it; and there came a great king against it, and besieged it, and built great bulwarks against it: Now there was found in it a poor wise man, and he by his wisdom delivered the city; yet no man remembered that same poor man. Then said I, Wisdom is better than strength: nevertheless the poor man's wisdom is despised, and his words are not heard. The words of wise men are heard in quiet more than the cry of him that ruleth among fools. Wisdom is better than weapons of war: but one sinner destroyeth much good. (Ecclesiastes 9:14–18, KJV)

In Gaza there was no poor man to deliver the city. For years, the government had encouraged Qatar's cash payments into Hamas-run Gaza, from $15 million to about $30 million a month, as a way of buying quiet. The warnings were there. More than a year before October 7, 2023, Israeli officials obtained a roughly 40-page Hamas battle plan for the assault and dismissed it as too difficult to carry out. In the months before, the young women serving as the military's observers on the border reported what they saw, and older commanders told them they were "just spotters." The wisdom was there; it was despised, and its words were not heard.

The battle is not to the strong, and it is not to the wise either, because the wise are not heard. That is why there is war. If the battle were to the wise, there would be no war, and nobody would profit. A machine can be pre-programmed, and a human can even be programmed to fight like a machine, or with one. It is the human in the loop, the one the intelligence responds to, who does that unique job. Intelligence is knowing what can be made into a tool when the objective is fixing something rather than destroying it.

5. Policy

The Pentagon's rule for autonomous weapons, Department of Defense Directive 3000.09 (reissued January 25, 2023), requires "appropriate levels of human judgment over the use of force." It does not define "appropriate," and the words "human in the loop" do not appear in it. A system that cues targets for an operator to approve counts as semi-autonomous and needs no senior review, and nothing in the directive sets how long that approval must take. The intelligence systems that decide where weapons point fall outside it, and "in cases of urgent military need" its reviews can be waived. The human judgment it requires is left undefined at the points where the chain becomes blind.

The Army's own leadership doctrine asks for more than lethality can supply. Its leadership model (ADP 6-22, July 2019) sets presence, which includes fitness, confidence and resilience, beside intellect, which includes mental agility and sound judgment, and character, which includes moral courage and candor offered to superiors. Vitality is presence; readiness needs the judgment and the candor as well. On February 21, 2025, the Secretary of Defense removed the military's top uniformed lawyers, saying the department wanted lawyers who "don't exist to attempt to be roadblocks." A mentality that keeps the vitality and removes the judgment beneath it undermines the vitality of leadership in general.

What looks like lawlessness may also be a stress test. A government's readiness is its three branches, its courts and its own lawyers able to check what the executive does. When the executive removes the military's lawyers and its generals and gathers power to itself, the test is whether any of that readiness remains.

6. Chain of Command

A chain of command is linked: an order passes from one person to the next, and each link can be traced and held to account. Each link is also a place where the order can stop. A machine can be placed in the chain as a link, but it cannot be held accountable and it cannot refuse. The weakest link can be any number of things, from command to capability: the funding, the machine itself, a miscalculation by one of the machines, an aggrieved combat veteran, a soldier gone AWOL. The shorter the chain, the quicker one can see what it is linked to. Where it protects the system, the chain may be obscured, but where it leaves the system vulnerable, the chain needs to become more visible. The end of the chain may be the result of the chain itself, with each part affected along with it.

A recent example is this year, on the first day of the war with Iran, one day after Anthropic was declared a national security supply-chain risk: a missile struck a girls' elementary school, killing 156 people, 120 of them students. The chain of events can be traced. The targeting ran through Palantir's Maven Smart System, with Anthropic's Claude embedded in it to generate targets, rank them and help pair the weapons (Claude, which helped write this paper, is made by Anthropic), a system built to compress the time between finding a target and striking it; in roughly five weeks, U.S. Central Command struck some 13,000 targets with it. The Pentagon's preliminary findings pointed to outdated coordinates from the Defense Intelligence Agency, and reports say warnings that the data was outdated were ignored.

Here, the chain of command is harder to follow. Anthropic's chief executive later said the company did not know exactly how its model was used, and that the principle it had set was that "a human makes the final decision." Each link follows the other: the data passes to the analysts, the analysts to the system, the system to the model, the model to the human, the human to the target package. On September 18, 2026, a UN fact-finding mission found reasonable grounds to believe the strike was a war crime. The chain of events ended at the school. The chain of command has not yet ended anywhere.

What permits failure at the top of the chain has the opposite effect. The United States used its veto six times to block Security Council resolutions demanding a ceasefire in Gaza, the last on September 18, 2025. Two months later it brought its own, and on November 17 the Council endorsed a Board of Peace for Gaza's reconstruction, thirteen votes to none. Russia and China, each able to stop it with a veto, abstained. Russia's ambassador said it gave "complete control over the Gaza strip to the Board of Peace"; China's said the draft gave only "skimpy details" of the board's structure. They named the reasons a veto would have been appropriate, and did not act on them. Two countries with plenty at stake elsewhere and no part in what had already happened in Gaza let the resolution pass, leaving a single link in the chain: from a seat on the Council to the executive branch of the United States, whose president would chair the board for life. By not voting against it, they became party to its establishment and to what it would enable. Responsibility by inaction.

A greater problem than the unwillingness to act is people in authorized positions unable to act, due to the blindness of the embedded systems themselves. Part of that unwillingness can stem from not having enough information, while the chain of command treats readiness as the speed of a decision with the tools available, and refusal as insubordination. A link can also be kept from refusing by what it stands to gain: where compliance is rewarded, with a contract, a promotion, a board seat or a pardon, the link that could stop the order has a stake in passing it on. Pitched as "win, win," the reward is a deal rather than a bribe. After a demonstration, whether on a battlefield or a company made an example out of, the same pitch needs no reward at all: do this, or this will happen to you. The person approving a target package in seconds may not see what it was built from: the outdated coordinates, the warnings, the ranking. The link is still there, but it is blind, and a system placed between the decision and everyone who could have caught it lets things pass under the radar. If even those inside the chain cannot see, those outside it see even less. What follows looks like a failure of humanity, and it is the illusion of that type of failure that gets the (artificial or super) intelligence product shipped: the blinded human can be blamed, the blame can become the reason to automate further, and further automation blinds the human more. Human error is the product's best marketing, and the product produced the error. This is catastrophic compound error. A link that cannot see cannot refuse.

Above the chain sits a seat that is not a link at all. Project Meridian, the department's study of the capabilities it will need for "absolute technological dominance," is co-directed by Elon Musk, whose companies hold defense contracts, and by Palmer Luckey, co-founder of a company that builds autonomous weapons. They help set the direction and stand to gain from it, outside the chain where accountability attaches.

PART 2: WHAT THE THEORY SAYS

7. When Readiness Worked

On September 26, 1983, the Soviet early-warning system reported a launch from the United States: one intercontinental ballistic missile, with four more behind it. The officer on duty at Serpukhov-15, Lieutenant Colonel Stanislav Petrov, judged it a computer error; a first strike, he reasoned, would come with hundreds of missiles, not five. He was right. The satellites had read a rare alignment of sunlight on high-altitude clouds. He received no reward, was reprimanded over the paperwork, and was moved to a less sensitive post; the incident was made public fifteen years later. On October 27, 1962, the Soviet submarine B-59, out of contact with Moscow for days and under signaling depth charges from U.S. destroyers, needed three officers to agree to launch a nuclear torpedo. Vasili Arkhipov was the only one who refused, and he persuaded the captain to surface and await orders; that became public forty years later. In both cases a prepared system reported a war, and a human being saw what it could not. And in both, "no man remembered that same poor man." A battle that does not happen leaves no record, which is why readiness that works is the hardest thing to find.

In January 2003, Katharine Gun, a translator at GCHQ, Britain's signals intelligence agency, read a request from the U.S. National Security Agency for help monitoring the UN delegations of six Security Council members whose votes could decide a second resolution on invading Iraq: Angola, Bulgaria, Cameroon, Chile, Guinea and Pakistan. She was in position, and she could see. She gave the memo to the press, and The Observer published it on March 2. The second resolution was withdrawn before a vote, and the war began without it. She was charged under the Official Secrets Act; in February 2004 the prosecution offered no evidence, after her defense had asked to see the government's legal advice on whether the war was lawful. Readiness did not stop the war, but it put the surveillance on the record before the vote, and the vote was never held.

8. How This Theory Can Be Proven Wrong

A forbiddance states something the theory says cannot happen; if it happens, the theory is wrong. It says what cannot happen, not what anyone may not do. These are written to be broken, and each can be broken two ways: the productive way, or the weaponized version. If one is broken in the direction of readiness, the theory has done its job.

Forbiddance 1: A prepared system reads the unforeseen as the nearest threat it was built for, and what it senses outside that goes uncollected. What would show this wrong: a system collects what it was not built to look for and adapts its response to it, and the response is not lethal, because lethality fits every threat at the moment.

Forbiddance 2: Lethality cannot reduce a threat over time. Every enemy killed produces two more eventual enemies, a wrong match produces them faster, and the threat evolves: the violent man, the gang, the terrorist, now autonomous AI. What would show this wrong: a sustained lethal response after which the threat shrinks and stays shrunk.

Forbiddance 3: The more a system relies on lethality to answer a threat, the less it recognizes the lethality that could be used against its own system. The assumption is that eliminating a threat involves lethality, rather than that lethality creates a threat. What would show this wrong: a system that raises lethality as its priority while strengthening accountability for harm through the people whose job is to account for it, such as its lawyers, its inspectors general and its civilian-harm review. What would support it: the opposite, targeting the ICC and other organizations whose job is to hold war crimes to account.

Forbiddance 4: The more a system prepares to destroy, the less it can build what would make destruction unnecessary. What would show this wrong: growing preparation to destroy that comes with growing capacity to prevent: aid, diplomacy, repair, and the people whose job is to object.

Forbiddance 5: The more a system is automated, the less humanity contributes to it, and the less it can be stopped from inside. What would show this wrong: something automated that increases humanity's contribution rather than taking away the need for it.

CLOSING

9. Human General Intelligence

Artificial general intelligence is defined by performance: "highly autonomous systems that outperform humans at most economically valuable work" (OpenAI). Performance is not intelligence. Human general intelligence is not a score for ranking people, the general intelligence psychology has measured since 1904. It is the capacity every person has to meet what was not foreseen: readiness. It is as evenly distributed as "all men are created equal" implies. Artificial general intelligence is built in advance, from what people made, and held in few hands. Built in advance, it is preparedness. Defined against human work, it is Forbiddance 5 stated as a goal. (Anthropic, which makes Claude, is among the companies building toward it.)

Recursion without an outside signal is catastrophic compound error. If a blinded human approves, the approval is called human error, and the error justifies more automation. A model trained on its own output collapses when what it generates replaces real data (Nature, 2024). Each round compounds the last. Recursive self-improvement, the path to artificial general intelligence, can become the same loop at scale. In 1965, I. J. Good called the first ultraintelligent machine "the last invention that man need ever make."

Human general intelligence is the outside signal. Collapse can be avoided when real data from outside the loop is kept alongside synthetic data rather than replaced by it (Gerstgrasser et al., 2024). The question is the origin of the data, not how much of it there is. After an AI that outplayed the strongest human players at Go appeared, professional players became more novel and played better (Shin et al., 2023).

10. The Window

What this paper describes is unfolding and evolving simultaneously, behind a classified wall of information, while the directives are being established. The Autonomous Warfare Command is to stand up by October 1, 2027, and Project Meridian's report is due within 120 days.

Organizing around the decision is the first trillionaire in the United States. He bought the platform he called the "de facto public town square," where officials and the public speak. He holds billions in government contracts and no office in which he answers to anyone but the executive branch his PAC spent millions to re-elect, and he continues to spend to elect Republicans. Through DOGE, he exercised authority across federal agencies without an office established by law, a question still before the courts.

11. References

Scripture: King James Version (Bible Gateway): Exodus 14:13–14, 15:3; Ecclesiastes 9:11, 9:14–18; Romans 12:19; Hosea 8:7.
Abraham, Y. "'Lavender': The AI machine directing Israel's bombing spree in Gaza." +972 Magazine and Local Call, April 3, 2024. https://www.972mag.com/lavender-ai-israeli-army-gaza/
ABC News (Australia). "The biblical story of Amalek and South Africa's genocide case against Israel." January 31, 2024. https://www.abc.net.au/news/2024-01-31/biblical-story-amalek-south-africa-icj-genocide-case-israel/103403552
Associated Press investigation of U.S. AI models in Israel's war (2025), as summarized by the Business & Human Rights Resource Centre. https://www.business-humanrights.org/en/latest-news/ap-exposes-big-tech-ai-systems-direct-role-in-warfare-amid-israels-war-in-gaza/
Mazzetti, M. and Bergman, R. "'Buying Quiet': Inside the Israeli Plan That Propped Up Hamas." The New York Times, December 10, 2023.
CNN, citing The New York Times, on the "Jericho Wall" battle plan, December 1, 2023.
Haaretz and The Washington Post on the female observers ("just spotters"), as cited in Kaleido, "Dehumanization: A Spiritual Crisis," May 2026.
U.S. Department of Defense. Directive 3000.09, "Autonomy in Weapon Systems," January 25, 2023.
U.S. Army. ADP 6-22, "Army Leadership and the Profession," July 2019.
JURIST, on the removal of the Judge Advocates General, February 21, 2025.
U.S. Department of War. "Secretary Hegseth Announces Six Major Initiatives at Quantico Speech." September 30, 2026. https://www.war.gov/News/Releases/Release/Article/4615424/secretary-hegseth-announces-six-major-initiatives-at-quantico-speech/
Rev. Transcript, Department of War executive order signing, September 5, 2025. https://www.rev.com/transcripts/department-of-war-executive-order
"2026 Minab school attack." Wikipedia (and the reporting it cites: Reuters, The New York Times, AP, BBC, Bellingcat, Forensic Architecture). https://en.wikipedia.org/wiki/2026_Minab_school_attack
Amodei, D., interview with Bloomberg ("The Circuit"), June 2026, as reported by Let's Data Science. https://letsdatascience.com/news/anthropic-ceo-addresses-possible-claude-role-in-iran-school-a095f4cc
JURIST. "Anthropic sues US Defense Department over 'supply chain risk' designation." March 2026. https://www.jurist.org/news/2026/03/anthropic-sues-us-defense-department-over-supply-chain-risk-designation/
CNBC. "U.S. appeals court upholds Pentagon designation of Anthropic as supply chain risk." September 25, 2026. https://www.cnbc.com/2026/09/25/pentagon-anthropic-ai-risk-appeals-court.html
UN Security Council Resolution 2803 (November 17, 2025). https://www.securitycouncilreport.org/un-documents/document/s-res-2803.php ; https://en.wikipedia.org/wiki/United_Nations_Security_Council_Resolution_2803
Al Jazeera. "US vetoes UN Security Council demand for Gaza ceasefire for sixth time." September 18, 2025. https://www.aljazeera.com/news/2025/9/18/us-vetoes-un-security-council-demand-for-gaza-ceasefire-for-sixth-time
"1983 Soviet nuclear false alarm incident" and "Vasily Arkhipov." Wikipedia, and sources cited there.
"Katharine Gun." Wikipedia. https://en.wikipedia.org/wiki/Katharine_Gun
OpenAI. Charter. https://openai.com/charter/
Spearman, C. "'General Intelligence,' Objectively Determined and Measured." American Journal of Psychology 15 (1904): 201–292.
Good, I. J. "Speculations Concerning the First Ultraintelligent Machine." Advances in Computers 6 (1965).
Shumailov, I., et al. "AI models collapse when trained on recursively generated data." Nature 631 (2024). https://www.nature.com/articles/s41586-024-07566-y
Shin, M., Kim, J., van Opheusden, B. and Griffiths, T. L. "Superhuman artificial intelligence can improve human decision-making by increasing novelty." PNAS 120 (2023).
CNBC. "Elon Musk becomes world's first trillionaire as SpaceX begins trading on the Nasdaq." June 12, 2026.
CNN. "Elon Musk's super PAC starts spending to help GOP in midterms." September 3, 2026. https://www.cnn.com/2026/09/03/politics/elon-musk-spending-midterms
Does 1–26 v. Musk (D. Md.), on appeal.