The Nine Powers: What a Whole Education Keeps Asking

Nine questions a complete programme of study keeps asking, from ends to moral imagination, how each power covers a blind spot in the others, and how to use the list as a checklist and a reading rotation.

In 1966 Joseph Weizenbaum of MIT described ELIZA, a program that answered typed sentences roughly as a psychotherapist might. It matched key words to templates and kept almost nothing of what it was told. Even so, he reported, some users were "very hard to convince" that ELIZA was not human. His sharpest conclusion was not about code. "Important decisions increasingly tend to be made in response to computer output," he wrote. The person who reads that output must judge how far to believe it. ELIZA showed "how easy it is to create and maintain the illusion of understanding". A question about machines had become a question about human judgment.

That is the idea of this page. A whole education develops several powers at once, because each one sees what the others miss. The historical cases are Established from the texts cited. The list of nine, and the claim that the powers correct one another, is this wing's design. It is Provisional: a framework to test on yourself.

The nine powers

Each power asks its own question:

Ends: what is real, good and worth pursuing.
Language and proof: what follows from what.
Character: whether reason can govern desire and fear.
History and politics: how groups organise, rule, fail and recover.
Scarcity: how finite resources are valued and allocated.
Science and probability: what evidence justifies belief.
Systems and risk: how interacting parts fail or endure.
Computation: what machines can represent, and what they miss.
Moral imagination: what abstraction overlooks about suffering and meaning.

On this view, a power left out is not a gap in a syllabus. It is a blind spot, and the other powers cannot see into it.

How each power checks the others

Ends. Aristotle opens the Nicomachean Ethics by observing that every art and inquiry "is thought to aim at some good" (I.1, 1094a). The arts nest: bridle-making serves horsemanship, and horsemanship serves strategy. Science, systems and computation tell you how to reach a goal. They are silent on whether it deserves reaching, as Reasoning for What Ends? argues.

Language and proof. A valid argument carries truth from premises to conclusion. It cannot make a false premise true. When the premises are claims about the world, science has to check them. In return, proof checks every other power for claims that do not follow. Proof and Precise Reasoning separates validity from soundness, and the elenchus lets you test a belief for contradictions.

Character. Socrates, as Aristotle reports him, held that no one, when he judges, acts against what he judges best. Aristotle replied that "this view plainly contradicts the observed facts" (Nicomachean Ethics VII.2, 1145b). People know better and do worse. A flawless argument is useless to someone whose fear or desire overrules it. Character decides whether a sound conclusion gets acted on.

History and politics. Herodotus called his book an "Inquiry". He wrote, above all, so that "the causes may be remembered" for which Greeks and barbarians went to war (Histories I, proem). History tests a plan against what groups have actually done. Elinor Ostrom's Governing the Commons (1990) documented communities that ran shared resources under rules of their own. They did so with reasonable success over long periods. Their institutions resembled "neither the state nor the market" (p. 1). A design that looks right on paper still has to survive rivals, institutions and power.

Scarcity. Every choice spends time, money or attention that could have gone elsewhere. That forgone option is its opportunity cost. Scarcity turns a list of ends into a set of trade-offs. Friedrich Hayek (1945) added that the knowledge needed to allocate well "never exists in concentrated or integrated form". It is dispersed among individuals, and prices help carry it. A plan that ignores cost is a wish list.

Science and probability. This power asks what evidence justifies a belief, and how much. Judea Pearl separates seeing from doing. Data about what happened together cannot, by itself, say what will happen if you intervene. Such questions "cannot be inferred from passive observations alone, regardless of how big the data" (Pearl, 2019). Diet data on its own hits the same wall. Reading Nutrition Evidence puts a diet link seen in populations a rung below a randomised trial counting heart attacks. Judgment Under Uncertainty builds the general habits of weighing evidence.

Systems and risk. Charles Perrow's Normal Accidents (1984) argued that in systems with complex interactions and tight coupling, some failures are inevitable. Adding warnings and safeguards does not remove them, because the complexity itself produces them. Systems thinking checks the single-cause story and the clever fix. Systems, Decisions and Robust Design shows how to design so that being wrong is survivable.

Computation. Writing a program exposes vagueness. Weizenbaum noted that writing an ELIZA script "clearly shows where the programmer's understanding and command of his subject leaves off". Computation's own blind spot is the one ELIZA displayed: fluent output with no understanding behind it. Pearl's limit applies here too, since a system trained only on associations cannot reason about interventions. Technology, AI and Human Judgment gives tests you can run on a model.

Moral imagination. Sophocles's Antigone sets Creon's decree, that none shall bury Polyneices, against Antigone's appeal to "the immutable unwritten laws of Heaven". Each side has a case. A cost-benefit table can count a loss. Tragedy shows what the loss means to the person who bears it. This power checks every abstraction above it, including the ends you chose.

Using the list

Use the nine as a checklist. Before a plan, a purchase or a course of study, ask each question in turn. Mark the ones the plan never asks. The silent ones are where to look first.

The same list makes a reading plan. Take one work from each power in turn, then start again, so no power goes unexercised for long. This keeps a programme from becoming either an antiquarian survey or a narrow technical course, the balance described on Start Here. The diagram at the top of this page shows the rotation.

The nine powers do a different job from the seven liberal arts. The arts are disciplines, each with its own object, studied in an order (The Order of the Seven Liberal Arts). The powers are questions a whole programme keeps asking of whatever it studies.

Where this could be wrong

The strongest objection. The list is arbitrary. Scarcity could fold into systems, and computation into proof. Nothing cited here shows that nine is the right number. Nor does it show that a learner who covers all nine judges better than one who goes deep in two. Breadth also has a cost, as the fifth power itself insists. Time spent on nine subjects is time not spent mastering one.

The best reply. The list is a checklist of questions, not a theory of knowledge. Merging two items does no harm if both questions still get asked. The claim is narrow. A plan that never asks one of these questions has an unguarded side. The cases above show what that can cost.

What would settle it. Keep a log of your own mistakes for three months. Tag each with the question that would have caught it. If most mistakes fall outside the nine, the list needs revising. If one power never appears, it may be a luxury for you. A fair test of the larger claim would compare learners who rotate through all nine with specialists. Both would face judgment tasks set in advance. No such study is cited here, so the claim stays Provisional.

Try this

Take a plan you are making: a course of study, a purchase or a project. Write the nine questions down the margin and answer each in one line. Circle the line that was hardest to write.
Recall a decision that went wrong. Name the power whose question would have caught it, and write that question where you will see it next time.
Choose the power you have studied least. Pick one work or Library page from it and finish it this month.