Roosevelt's critics had been complaining about the cost of his government programmes for years without effect — "their attacks had no resonance with the public, who saw the president as working to end the Great Depression."1
Then they found what looked like the perfect example.
his dog, Fala, which he lavished with favors and attention. Critics railed at his insensitivity—spending taxpayers' money on a dog while so many Americans were still in poverty. But Roosevelt had a response: How dare his critics attack a defenseless little dog? His speech in defense of Fala was one of the most popular he ever gave.2
Greene's reading:
In this case, the weak party involved was the president's dog and the attack backfired—in the long run, it only made the president more sympathetic, since many people will naturally side with the "underdog."3
The strategic error is precise and it is not about dogs.
Roosevelt's opponents had a real argument — government spending — and no traction with it. So they looked for a concrete instance to make the abstraction vivid, which is ordinarily correct technique: a specific, small, checkable example does the work that a large claim cannot.
Their mistake was the choice of instance. The abstraction wasteful spending is contestable and unsympathetic; the instance they selected transferred the contest onto a party the audience liked more than either side.
⚠ Which is the finding, and it generalises past this case: an example is not a neutral illustration of an argument. It replaces the argument. Once Fala was the subject, the question in front of the public was not is this administration wasteful but should people be attacking a dog, and the second question has an obvious answer.
You do not get to specify how your example will be read once it is in the room.
Greene attributes the effect to a general tendency — many people will naturally side with the "underdog." ⚠ That is the weakest part of the passage and the corpus should not rest on it.
The stronger reading is structural. The dog was not sympathetic before the attack; it was not a public figure at all. The attack created the pairing — a group of political opponents on one side and a small defenceless thing on the other — and the pairing is what the audience responded to.
Which means the sympathy was manufactured by the attackers, at their own expense, and no amount of correctness about the underlying spending could recover it.
⚠ And it explains why Roosevelt's speech worked as well as it did. He did not defend the spending. He asked how dare they attack a defenceless little dog — which is not a rebuttal, it is an insistence that the contest the critics had accidentally created be adjudicated on its own terms.
Law 36 names two dangers in responding to a small irritant, and this is the one the chapter gives least attention to.
First danger: "by paying undue attention to a puny enemy, you look puny." Scale advertises the target — the Punitive Expedition. Second danger: "if you succeed in crushing the irritant, or even if you merely wound it, you create sympathy for the weaker side."4
⚠ These are not the same warning and they have opposite failure conditions. The first is a danger of not succeeding — the longer it takes, the worse it gets. The second is a danger of succeeding, and it fires precisely when the attack lands.
So the bind the corpus recorded on the law page is real and this case is the proof of one half: losing to a small opponent looks weak and beating one looks cruel. ⚠ There is no version of attacking Fala that ends well, including the version where the spending really was indefensible.
⚠ 🚩 SECONDARY WITHOUT PRIMARY · 🚩 SINGLE SOURCE — no date, no named critic, no quotation from the speech, and no indication that this is one of the most quoted pieces of American political oratory with a substantial documentary record.
🚩 The specific allegation is not stated. Greene gives "spending taxpayers' money on a dog" and "lavished with favors and attention", which is vague enough that the reader cannot assess whether the critics had a point. A page that cannot say what the accusation was cannot evaluate the response to it.
🚩 "One of the most popular he ever gave" is offered with no measure and no source.
🚩 MOTIVATED REASONING in the framing — this is a case selected because it worked, told from the winner's side, and the corpus should note that Roosevelt's response was itself a piece of deflection: he answered a charge about public money by discussing an animal, and the chapter presents that as skill without registering it as evasion.
Worth separating out, because the chapter files this among warnings to the aggressor.
Roosevelt is the defender here, and what he does is the same reclassification the Keys recommend elsewhere: Sand reclassifies an institution, Aretino reclassifies his origins, and Roosevelt reclassifies the attack itself — from a charge about spending into an assault on a dog.
⚠ Which is the third instance of the same operation in one chapter, and Greene distributes them across a warning, a technique and a tactic without noticing they are one move.
The general form: when attacked, do not contest the claim — contest what kind of act the attack was.
You have a real criticism to make and you are looking for the concrete example that will make it land.
Choose the instance for how the audience will read it, not for how well it proves your point. ⚠ The best illustration of your argument is frequently the worst thing to say, because the example becomes the subject and your argument does not come with it. Fala proves the spending case beautifully and puts a dog in the room.
Run the substitution test. Say your example out loud as the entire content of your position — they spent money on a dog — and ask what a neutral person now thinks the disagreement is about. If the answer is not your argument, you have chosen wrong.
And check who the example makes sympathetic. Not who it indicts. Every concrete case has a party at the small end of it, and your attack is what puts them there. They were not sympathetic before you named them.
If you are on the receiving end, the move is Roosevelt's: do not defend the substance, characterise the act. ⚠ This works and it is evasion, and the corpus should be clear that those are compatible — the technique's effectiveness has nothing to do with whether the underlying charge was fair.
Strongest evidence. A clear structural sequence — ineffective general attack, search for a concrete instance, instance transfers the sympathy, response converts it into the president's best-received speech — with a mechanism that does not require the underdog generalisation.
Tension — Greene explains it by a general human tendency and the case supports a structural explanation instead: the pairing was created by the attack.
Tension — the response is deflection and is presented as skill. Roosevelt did not answer the spending charge.
Tension — the second danger fires on success and the first fires on failure, and the chapter presents them as one warning.
🚩 [POPULAR SOURCE] · 🚩 SINGLE SOURCE · 🚩 SECONDARY WITHOUT PRIMARY — no date, no critic named, no line quoted from a heavily documented speech, and the specific allegation left vague. 🚩 MOTIVATED REASONING — a winner's case with the underlying charge unassessable.
Open questions. Was the attack actually about the dog? Greene's account implies the critics chose Fala as their central example, and a more likely reading is that a minor line item was seized on opportunistically and Roosevelt elevated it. If so the initiative was his, not theirs — which would make this a case of a defender creating the favourable pairing rather than an attacker blundering into it, and would reverse the chapter's lesson.
This is the second of Law 36's two named dangers and the only case for it.
Against the law's Transgression the pair completes the bind. ⚠ Villa demonstrates the cost of not succeeding and Fala the cost of succeeding, and together they make the chapter's real argument — there is no good outcome available in a public contest with a smaller party — which neither case states and the chapter never assembles.
Against Law 33's the thumbscrew there is a direct contradiction worth recording. That law instructs you to find and press a target's specific vulnerability. ⚠ Fala was Roosevelt's vulnerability by that reading — a specific, checkable, personally embarrassing instance — and pressing it destroyed the attackers. The two laws give opposite advice about the same act, and the variable that separates them is whether the pressure is applied in public, which Law 33 does not mention.
And against Aretino and Sand in the same chapter: three reclassifications, filed as three different things.
Rhetoric — the example that replaces the argument. Oversimplification to Third-Grade Level holds that a message reduced to its most concrete form is the version that travels, and that concreteness is what determines which part of an argument survives transmission.
Fala is that principle producing a disaster for the people who invoked it.
The insight neither produces alone: the simplification literature treats concreteness as a delivery decision — the same argument, made portable. The Fala case shows it is a substitution, not a compression: the concrete instance does not carry the abstraction along with it, it replaces it in the audience's mind and then behaves according to its own properties. A dog has properties — small, defenceless, not responsible for federal budgets — and none of them were in the argument about spending. ⚠ Which gives the operative rule the rhetoric frame implies and does not state: before choosing your most vivid example, list what it will be about once nobody remembers what it was an example of. If that list does not favour you, the abstraction was safer, even though it was losing.
Ethology and social cognition — the pairing that creates the weak party. Loss Aversion establishes that evaluations are made against a reference point, and that the point is set by how a situation is framed rather than by its contents.
That is a better account of the Fala effect than the underdog generalisation.
What the pairing produces: many people will naturally side with the underdog is a claim about a standing disposition, which predicts that the dog was already sympathetic. It was not — it was not a public entity at all until the attack made it one. The reference-point frame says the sympathy is generated by the framing that the attack itself supplied: once the situation is presented as an assault, the audience's reference point is the dog's position, and everything is evaluated as a loss to it. ⚠ Which is why the attackers could not recover by being right about the money — they had installed the reference point and could not move it from inside. And it yields a defence Roosevelt used and the chapter does not name: his speech did not argue, it kept the reference point in place, because the frame was doing all the work and any substantive rebuttal would have moved it.
Sharpest implication. An example does not illustrate an argument, it replaces it — once Fala was the subject the public question was should people attack a dog, and that has an obvious answer. And the sympathy was not pre-existing: the attack created the pairing, at the attackers' expense, which is why being right about the spending could not recover it. ⚠ Together with the Punitive Expedition this completes the chapter's real argument: losing to a small opponent looks weak, beating one looks cruel, and there is no version of attacking Fala that ends well — including the version where the spending was indefensible.
Generative questions.