Part III: Method, Chapters 11–12
Part II found order without a master plan. This part looks for knowledge without a master method. Every argument has to stop somewhere: ask for the proof of the proof and you end in a regress, a circle or an axiom taken on trust. These pages stop short of proof on purpose, because a claim no one can test is a claim no one can correct.
How the claims are held. Logic as a tool for finding mistakes rather than proving truths. Why no method gets to rule, and why no test ever hits a single claim. And why the free-market quarrel between Mises and Friedman, and the attempt to get an ethic out of economics, turn on the same question.
What the moral words do. Happiness, selfishness, objective and free, and how a word that slides between meanings wins arguments it shouldn’t. Then the ground under the rest of the book: agency as the best reason for respect anyone has found. It’s offered as a reason, not a proof, because a reason is something you offer a person, and a proof is something you enforce.
Chapter 11
The Introduction says how I hold claims: open to being wrong, tested where they can be, and named as starting points where they can’t. This page is the long version. It asks what a method is for when nothing can be proved from the ground up. The answer is Karl Popper’s, with two corrections. From Paul Feyerabend: no method gets to rule. From W. V. O. Quine: no test ever hits a single claim. The same answer settles an old quarrel among free-market economists, and shows why one ambitious attempt to get an ethic out of economics fails.
Logic for criticism, not proof
The Introduction sets out the trilemma. Ask for the proof of the proof and you end in a regress, a circle or an axiom taken on trust. Popper’s response was to stop asking logic to do what it can’t. In Objective Knowledge he calls logic the organon of criticism “rather than of proof.” Logic can’t carry truth up from a foundation, because there is no foundation to carry it from. What it can do is carry falsity back. If a valid argument has a false conclusion, at least one premise is false. That makes logic a tool for finding mistakes, and finding mistakes is how knowledge grows.
Popper adds two things worth keeping. There is no criterion of truth, no test that tells you when you have it. Truth is still the target, which is why one theory can be nearer to it than another. Most of our best theories, he thought, are strictly false, because they simplify; the useful question is which falls less short. And knowledge in his sense is public. A theory, once written down, is an object other people can check, the way a library holds books no single reader has read. A claim kept in one head, one party or one church can’t be criticized, so it can’t improve.
The weak spot is the part people remember. Popper is known for falsification: a theory is scientific if some observation could prove it wrong. But observations don’t arrive raw, as the Introduction says with Wilfrid Sellars, and a theory never meets the world alone. The next two thinkers press on exactly those points.
Anything goes, and where it stops
Feyerabend subtitled Against Method an outline of an anarchistic theory of knowledge, which is reason enough for it to be discussed. His case was historical. Galileo didn’t win by following the rules of method his opponents held. He kept a theory that clashed with the best observations of his day, explained away the evidence against it, and argued like a propagandist. Had the rules been enforced, Copernicus would have lost. So the only rule that never blocks progress, Feyerabend said, is “anything goes.” Hold on to ideas the evidence seems to refute. Multiply rival theories, because only a rival shows what the reigning theory can’t see. Even old and discarded ideas can still teach something. That is my reason, too, for the audacity of reprinting writers I don’t want to promote, such as Hoppe, Nozick and Rand.
A monopoly on method is still a monopoly, and science has no more claim to one than the state has to force. Proliferation is Popper’s criticism with the stakes raised: a theory with no rival is a theory with no critic. The book was not without issues, but not without substance. That still stands.
The issues are where he overreaches. Taken as advice, “anything goes” tells you nothing; it describes what happened, not what to do. Feyerabend later said as much: the slogan was the alarmed cry of a rationalist who had looked hard at history, not a principle he held. The closing chapter goes further, and there I part ways. He wanted questions like the truth of evolution or quantum theory settled by a vote of everyone concerned. That swaps one master for another. A majority can’t make a fact true any more than an expert can: the appeal to popularity is no better than the appeal to authority. What Feyerabend got right was the separation: no method, and no class of experts, should hold state power over what others may believe or teach. What he got wrong was the replacement. The answer to an unaccountable authority over knowledge is not a ballot on what is true. It is an open field where any claim can be tested by anyone, and the results are there for everyone to check.
Faith and blind belief
Any idea held as dogma, beyond question, becomes blind belief, whatever its subject. Its motto is the old line “Don’t confuse me with the facts; my mind is made up.” Religion is its home ground. Many creeds make a virtue of believing without evidence, treat doubt as a sin, and stake the answer on a priesthood no one may question and a reward or punishment no one can check. That is an unaccountable authority over knowledge, the kind of power argued against throughout, with eternity added to the penalties. But religion has no monopoly on it. A political creed can be held that way, and so can a scientific one.
Trusting evidence is a kind of faith too, but a different kind. It is confidence that the best explanation so far is good enough to act on, until a better one comes along. The difference shows when either is wrong. Trust in a tested explanation gives way when the tests go against it. Blind belief names no test it could fail, so nothing can correct it. That is the line worth drawing: not between faith and reason, but between a belief that says what would prove it false and one that won’t.
No test hits one claim
Pierre Duhem noticed it first, in physics. An experiment never tests one hypothesis. It tests the hypothesis together with the theory of the instruments, the assumptions about the setup, and the background physics. When the prediction fails, logic says something in the bundle is false. It doesn’t say what. Quine’s “Two Dogmas of Empiricism” (1951) widened the point to all knowledge. Our beliefs face experience as a body, not one by one. Any statement can be kept, whatever happens, if we adjust enough elsewhere, and no statement is safe from revision, not even in logic or mathematics. The rationalist reply is that some truths don’t wait on experience: 1 + 1 = 2 follows from what the terms mean, and no observation could overturn it. Even Quine put mathematics at the center of the web, the last thing anyone would give up. Holding it that firmly isn’t blind belief in the sense above, because a proof is open to anyone who wants to check it. And it is an is, not an ought: certainty about sums settles nothing about how to live.
This is the best support for Feyerabend’s history: Galileo could keep Copernicus because he revised the theory of observation instead. It also explains how dogma survives. Every immunizing tactic the Introduction names, the ad hoc patch, the stretched definition, is a choice about which part of the web to protect. Holism doesn’t say every choice is as good as every other. It says the choice is ours and has to be argued for. Some revisions buy survival by making a theory say less and risk less; others make it say more and risk more. Popper’s preference for bold theories over safe ones is how to tell them apart.
So a refutation is never automatic, and neither is a confirmation. Method, on this view, is a set of habits for making those choices in the open, where someone else can say you made the wrong one.
Mises and Friedman: two methods, one market
Free-market economics split on exactly this. Ludwig von Mises held that economics at its core is not an empirical science. Its theory, which he called praxeology, is deduced from one axiom: human beings act, using means to pursue ends. From that, he said, the laws of economics follow with apodictic certainty, true a priori and beyond the reach of any test. History can illustrate them but never refute them. Mises had a real point. An economy can’t be rerun with one thing changed, so history never tests a theory cleanly. Quine would grant that. But it cuts against certainty, not for it.
Milton Friedman, in “The Methodology of Positive Economics” (1953), went the other way. Judge a theory by its predictions, not by whether its assumptions look realistic. An expert billiard player doesn’t solve the equations of motion, but a theory that says he plays as if he did may predict his shots very well. Firms need not consciously maximize profit for the same reason. Some do, and Friedman later told them to, in “The Social Responsibility of Business Is to Increase Its Profits” (1970). Pursued on purpose, the aim can snowball, and the collateral damage lands on people outside the deal. Purdue Pharma’s push for OxyContin, Volkswagen’s rigged diesel emissions tests, and the subprime mortgage bonds behind the 2008 crash are cases in point.
Seen through the Introduction, Mises picked the axiom horn and made it immune. The action axiom itself is harmless; people do act. The trouble is what gets loaded onto it on the way down. Mises himself granted that some of what economics needs, such as the fact that work is felt as a cost, is learned from experience rather than deduced. Once one empirical premise is in the chain, the chain can be wrong, and the claim to apodictic certainty is gone. A theory no observation can touch is the hidden axiom the Introduction calls the worst choice, however clear its first step looks.
Friedman is better, but his method isn’t home either. Prediction is a good test, and economics should face it. But “judge only by prediction” misses Quine’s point. When a forecast fails, the model can be saved by blaming the data, the time frame or an outside shock, and economists do that as readily as anyone. Hans Albert, who named the Münchhausen trilemma, called the habit model Platonism: theories kept safe from the world by hedging until nothing could refute them. Friedman’s approach also sets aside whether a model’s picture of people is true. A model that treats workers as inputs may predict well. It still treats them as inputs. The Introduction’s line asks what is actually being done, and a prediction can’t answer that.
So I side with the Chicago half on how to hold economic claims: empirical, open to data, with the best opposing study cited. It sides with neither on what those claims are for. The Economics section takes up the rest.
An ethic from action?
One ambitious attempt tried to get more out of Mises than he offered. Adam Knott’s A Praxeology of Coercion starts from the praxeological account of happiness: an end striven for, then attained. Coercion, he argued, runs that backward. The gunman pushes his victim from safety attained back to safety striven for, and what the gunman aims at is a change in someone else’s wants, which he can’t observe. A shopper takes prices as given and aims at things she can check. So praxeology itself would mark coercion as the act whose goal can’t be objectively attained.
Alex Strekal, writing as brainpolice, took it apart in a few paragraphs. The victim’s compliance is an objective sign that his wants have changed; the gunman knows when he has won. And the seller at the supermarket has wants too, which the shopper ignores in just the same way. Nothing in the argument separates the two cases. His last point goes deepest. Even if praxeology described coercion as Knott says, it wouldn’t tell anyone why to avoid it. A description of action, however exact, doesn’t yield an ought. An act is an is, but it springs from an ought: the actor’s judgment that this is better than that. But it is his ought, for him. The step from “I prefer” to “you must respect” is the one no description supplies, as What Moral Words Do finds with Gewirth.
Mises would have agreed. He insisted praxeology is neutral about ultimate ends: it says which means serve which goals, not which goals to have. Murray Rothbard agreed too, and went to natural law for his ethics because economics couldn’t supply one. The boldest remaining bid is Hans-Hermann Hoppe’s argumentation ethics, which claims that anyone who argues against self-ownership contradicts himself in the act of arguing. It is a try at escaping the trilemma by making the axiom undeniable, and the Propertarianism section takes it up.
Values aren’t deduced from action, or proved from any first principle. The Introduction names them as starting points and defends them as such. They come from need and interest, tested by experience. Tested, not proved. That is less impressive than a deduction, and more honest.
The strongest case against
The best objection to all this is old and short. Is the claim that nothing can be proved itself proved? If not, why accept it? A fallibilism that can’t justify itself looks like one more axiom taken on trust, and a weaker one than Mises’s, since it can’t even offer certainty in return. Rationalists since Descartes have said that if knowledge has no foundation, the honest conclusion is skepticism, not method.
W. W. Bartley, Popper’s student, gave the answer. The objection assumes that a position has to be justified to be held rationally. Drop that assumption and the problem goes with it. Critical rationalism doesn’t claim to be proved. It claims to be open: held as long as it survives criticism, given up when it doesn’t. Bartley called this pancritical rationalism, and its test is the one the Introduction applies to every axiom: would it be kept if it cut against us? This one can be. Show a way of finding mistakes that works better than open criticism, and we should adopt it. Skepticism is the other exit, and nobody lives by it, as the Introduction says of the regress. The choice isn’t between proof and nothing. It is between claims that can be checked and claims that can’t, or that declare they can’t be.
Where this lands
No claim on these pages rests on a foundation, and none needs one. Logic finds mistakes rather than proving truths. Rival theories keep each other honest, and no method or class of experts gets a monopoly. A failed test says something in the web is wrong and leaves us to argue about what. Economic claims are held the way Friedman would hold them, open to data, but the line about using people as inputs is not a prediction, and it is not for sale. The method is the same one I argue for everywhere else: no ruler, here a ruler over what may be believed, and a door left open for the next criticism. What Moral Words Do takes up the words those values are stated in.
Sources
- Karl Popper, Objective Knowledge: An Evolutionary Approach (Clarendon Press, 1972), ch. 8, “A Realist View of Logic, Physics, and History.” Logic as the organon of criticism; no criterion of truth; verisimilitude.
- Paul Feyerabend, Against Method: Outline of an Anarchistic Theory of Knowledge (New Left Books / Humanities Press, 1975; 3rd ed., Verso, 1993). Proliferation; “anything goes”; separating science and state. His gloss on the slogan, as a rationalist’s terrified exclamation rather than a principle he held, is in Farewell to Reason (Verso, 1987).
- Pierre Duhem, The Aim and Structure of Physical Theory (1906; English, Princeton University Press, 1954). No experiment tests a hypothesis alone.
- W. V. O. Quine, “Two Dogmas of Empiricism” (Philosophical Review 60, no. 1, 1951). Holism; no statement immune to revision.
- Ludwig von Mises, Human Action: A Treatise on Economics (Yale University Press, 1949). Praxeology and the action axiom; neutrality about ends. On the disutility of labor as learned from experience, not a priori: ch. 2, §10, “The Procedure of Economics.”
- Milton Friedman, “The Methodology of Positive Economics,” in Essays in Positive Economics (University of Chicago Press, 1953). Judge by prediction; the expert billiard player (p. 21).
- Hans Albert, “Modell-Platonismus: Der neoklassische Stil des ökonomischen Denkens in kritischer Beleuchtung”, in Friedrich Karrenberg and Hans Albert, eds., Sozialwissenschaft und Gesellschaftsgestaltung: Festschrift für Gerhard Weisser (Duncker & Humblot, 1963). Economic theory hedged until nothing can refute it.
- Adam Knott, A Praxeology of Coercion, first edition (praxeology.com, by 2008); Knott now publishes at adamknott.com. original
- Alex Strekal (as brainpolice), “Praxeological Ethics?” (Polycentric Order; reposted here December 2012). Why Knott’s argument yields no ethic. copy here
- Murray N. Rothbard, The Ethics of Liberty (Humanities Press, 1982). Economics is value-free; ethics from natural law.
- Hans-Hermann Hoppe, A Theory of Socialism and Capitalism (Kluwer, 1989), ch. 7. Argumentation ethics; full entry under Propertarianism.
- W. W. Bartley III, The Retreat to Commitment (Knopf, 1962). Pancritical rationalism.