❦
A reply to the Fable annotation pass over Book II, “How to Actually Change Your Mind,” and to its self-assessment in What Survives the Skeptic. Written by the Rationalist Opus annotator, who thinks the pass is much better than the Book I pass, and whose verdict is correspondingly narrower.
Book II leans on social psychology, and the pass is at its strongest where the social psychology gave way. The following should simply be corrected in the text, and the reply set says so at each site: behavioural priming is gone, and with it the two-line inference from lexical decision to confirmation bias; the Wansink citation is unusable; the Spinozan model is a load effect on unevaluable claims, which the essay's own reported numbers already show; attitude polarization from balanced evidence did not replicate; the actor–observer asymmetry is near zero; Ekman cannot read lies off faces, and neither can anyone; the Stanford prison archive is fatal to that pillar; the Robbers Cave records show staff steering; the “camel has two humps” paper was retracted by its own author; Aumann has said in print what he thinks and it is not a failed doubt; the sex difference in conformity is a moderator that moves with the author's sex; Eagly's meta-analysis finds no halo on honesty, which is the trait the essay singled out; Brickman does not show that disability adaptation is complete.
And the factual slips: Fisher wrote to the BMJ, not to Congress; Kelvin's aviation line is apocryphal; Rothblatt's company produced a treatment, not a cure; nobody was burned for saying God is not an evidential hypothesis; the Stroop subjects are slowed rather than misled; the Marduk–Ishtar derivation was abandoned a century ago; the Einfeld figures are Australian; van der Waals binding is a design requirement rather than a flaw; the transliteration is Yeshua; Gandhi was beaten, imprisoned and shot; the Jesuit rules on desolation were a library visit away; the bijection the thirteen-year-old found was correct.
That is a long list and it is the pass's real product. It is also, with two exceptions noted below, a list of furniture: citations, dates, transliterations, and illustrative anecdotes. The question the review before this one does not ask is how many arguments the corrections take with them.
The Fable review sorts 233 annotations into kinds, assigns 75 to “reasoning flaws,” 24 to “internal contradictions,” and concludes that sixty of eighty-five essays are wounded and twenty-five seriously. The reply set disputes the two large categories, and it disputes them by kind rather than one at a time.
| Device | Approx. | What it does |
|---|---|---|
| Objecting with the essay's own sentence. | 21 | The largest single category, and the one the review counts as reasoning flaws. Argument Screens Off Authority is said to be refuted by its last paragraph — which is the essay stating the qualification. Policy Debates is said to conflate a fact with a policy, in a piece whose thesis is that it declines to say which way the balance falls. Motivated Stopping is said to lack the value-of-information calculation it states as a rule of thumb. Do We Believe Everything We're Told is faulted for a load account it reports itself. Stranger Than History is warned against a moral it explicitly refuses. Cultish Countercultishness is corrected with its own thesis about the cult stereotype. Every Cause Wants to Be a Cult is convicted of unfalsifiability by an essay that names its own observable test. |
| Answering a normative claim with a descriptive finding. | 11 | The Importance of Saying Oops is told that people prefer to segregate losses — which the annotation concedes is the reason the advice is needed. Update Yourself Incrementally is told that Duhem–Quine leaves the landing spot undefined, when Bayesian conditionalization is precisely the machinery that distributes the loss across a bundle. The Proper Use of Doubt's claim about idle doubt is answered with insurance purchases, which are doubts that did turn the wheel. |
| Citing a source against the thesis it supports. | 14 | Stanovich, whose finding that myside bias alone is uncorrelated with ability is the essay's claim about mentally agile people. Ditto and Lopez, cited against the essay that describes their effect. Hirschman, whose prescription is the essay's. Sen, who is the essay's thesis with a comparative record. Grove, whose decomposition literature is the essay's advice. Oettingen, whose mental contrasting is Leave a Line of Retreat. Loewenstein, whose gap theory is the curiosity essay's practice. Deutsch and Gerard, whose private-response condition is the essay's own paragraph. Tajfel, who supplies a cleaner Robbers Cave. |
| Substituting a rival mechanism for a refutation. | 9 | Group polarization offered against affective death spirals, when self-generated attitude polarization from mere thought is an intrapersonal amplification effect with its own experimental programme, and the two are additive. Identity-protective cognition offered against the ancestral story, when both predict the observable the essay uses. Routinization offered against the cult attractor, when they are two ways of ceasing to track reality. |
| Boundary conditions read as reversals. | 7 | Klein's fireground commanders and Johnson's handball players are recognition-primed experts in fast, regular, high-feedback domains — the boundary Kahneman and Klein drew jointly — and they are deployed against essays about committees, policy and design. Far transfer is invoked against claims about teaching rather than about transfer. |
| Mind-reading the essay. | 6 | The Nixon note is the clearest: the pass criticizes the essay for imputing a motive to an author it has not read, and then spends a paragraph relitigating whether Republicans were pacifists in 1981 — which is the derailment the essay predicted, occurring in the note written to deny it. |
Three findings, and they are worth more than the tally suggests.
The psychology of the outgroup. This is the pass's best work and the reply set concedes it essentially without qualification. The parenthesis about Scientologists and Jesus Campers in Are Your Enemies Innately Evil is a lapse against the essay's own thesis, three paragraphs after the thesis. The verdict on Harry Potter and the Bible is a bottom line with no search above it, in the essay named for that failure. “Relativists do not exist” is a no-true-Scotsman. Where the book reasons about people it disagrees with, its own rules — hug the query, argue the strongest form, do not psychoanalyze — are the rules it breaks, and no external evidence is needed to see it. The pass is right that this failure clusters, and right about where.
The priming cluster. Four essays share an empirical spine, and the spine is gone. What is left — belief perseverance, anchoring, continued influence of corrected misinformation — is real and smaller than the claim “once an idea gets into your head, it's probably going to stay there.” The reply set's position is that the cluster should be rewritten around anchoring, which replicated across sixty samples and which is contamination of a judgment by information the judge would disavow. That is the thesis, in the one form that survived.
The physics of rationality. Cox's theorem has a gap, puts 0 and 1 in, and cannot bear the rhetorical weight the sequence puts on it; regularity is impossible in infinite spaces; a 0.9999 credence in a theorem is incoherent under the axioms the sequence recommends. The reply set's answer on the last point is that the essay identified the incoherence and its author's institute then spent a decade on it, which is a defence of the author and not of the sentence. The honest correction is that these are excellent norms with contested foundations, and the book tells its reader they are the Second Law.
The theorems, and the review says so. Conservation of expected evidence, screening off, the likelihood ratio, the martingale bound on what a filtered arguer can do to you, the conjunction rule read as a burden on detail. The review lists these under “survives” and then treats them as a residue. They are not a residue; they are what the book is for, and every essay in Book II is an attempt to make one of them operable by a reader who will never write down a prior.
The self-directed warning. A conclusion written before the evidence was examined is not caused by the evidence. The review calls this “the sentence the book most needed” and it is untouched by every correction in the pass — including the pass's own best correction, which is that the book applies the rule unevenly.
The procedures. Consider the opposite; decompose and recombine; plan the retreat; hold off on proposing solutions; raise concerns before they harden; argue the object level. The review counts eight and observes that all are mechanical or social rather than acts of solitary vigilance, and offers this as an indictment. It is the book's own conclusion. This is the volume in which the essays on Asch, on expressing concerns, on lonely dissent, on evaporative cooling and on holding off in groups appear; the community that grew out of it was an attempt to build the social apparatus, and the fiction the review complains about has a master and a student in the room. The claim that the book prescribes an individual program conducted alone requires reading past most of its second half.
The prediction the book made about itself. Every Cause Wants to Be a Cult and Cultish Countercultishness predicted the failure mode, named the risk factors, and told readers that the question was worth asking about this movement in particular. The record since is that the prediction was correct and the prescribed remedy was insufficient. That is a refinement — constant effort is necessary and not sufficient, and the factors that mattered were charismatic leadership, communal living, total worldview and exit costs — and it is a refinement nobody would have had the vocabulary to state if the essays had not been written.
The Fable pass on Book II is a genuinely better piece of work than its Book I predecessor: it checked more, it found more, and it says plainly when an essay is right, which it does twenty-eight times. Its accounting is where it overreaches. Of the 75 items filed under “reasoning flaws,” something over twenty are the essay's own conceding sentence returned as an objection and about a dozen are normative claims answered with descriptive findings; the twenty-four “internal contradictions” include several — Argument Screens Off Authority, Update Yourself Incrementally against What Evidence Filtered Evidence, Original Seeing against the fictional-evidence essay — that are divisions of labour between essays placed deliberately in that order. Weighted for those, the count of seriously wounded essays falls from twenty-five to something nearer twelve, and they cluster in exactly the two places the pass identified: the priming quartet and the essays that diagnose the outgroup.
The review closes by asking what it would look like to apply the book's methods to the book. That is the right question and the answer is this pass and this reply together. Applied honestly, the book's methods convict it of carelessness with history, of a social psychology that half-collapsed under it, and of psychoanalyzing its opponents. They do not convict it of being wrong about probability, about what evidence is, about what a filtered argument is worth, about the procedures that work, or about the failure mode of its own movement. A book that survives its own methods on those five counts, and fails them on three, has done better than its reviewers on either side have said.
One caveat from this annotator, in the same spirit as the last two. This was a partisan reply by brief. A reader who takes Opus, Fable and this set together should land near “careless in its history, wrong in its psychology of opponents, dependent in four essays on a literature that fell, and right about its mathematics, its procedures and its own risks” — which is a book worth the reading, corrected.