Far too often, debates have been “won” simply by pointing to well-known logical fallacies, as if identifying a formal fallacy in an argument proved that the argument was wrong. It doesn’t seem to matter what the fallacy is. It can be affirming the consequent, denying the antecedent, slippery slope, poisoning the well, the archetype fallacy or whatever. They all seem to share a common thread: observing that an argument is fallacious in the formal, logical sense of the word, and fallaciously concluding that it should therefore be ignored altogether (or worse, turned into a a mini logic-lesson, as I myself have been guilty of on occasion).
This fallacious reasoning about fallacies is summed up nicely, albeit unintentionally, by this site, quoted approvingly by McQ, which has this to say:
A fallacy is a kind of error in reasoning. The alphabetical list below contains 171 names of the most common fallacies, and it provides explanations and examples of each of them. Fallacies should not be persuasive, but they often are.
Wrong. Fallacies should never be mistaken for conclusive proofs, but that doesn’t mean they should never be persuasive. Indeed, they often should be. Here’s why: the word fallacy, as used by logicians, does not mean the same thing that it typically means when used by laymen. A good lay definition of “fallacy” is as follows:
- A false notion.
- A statement or an argument based on a false or invalid inference.
- Incorrectness of reasoning or belief; erroneousness.
- The quality of being deceptive.
Obviously, anything that matches those definitions should never persuade anyone of anything, and is rightly described as an error in judgment when it does. However, none of these definitions match that of a logical fallacy, which merely means that the conclusion does not logically follow from the argument. In other words, all this definition means is that it is possible – not necessarily probable – that the argument can be true while the conclusion is false.
Take, for example, a criminal trial. Every one of them, without exception, involves inferential leaps which, however reasonable, would be deemed “fallacious” under the rules of geometric logic. The fact that the defendant’s fingerprints are on the gun and his DNA is found on the victim may be very damning indeed, but they are not absolute proof that he is the killer. Maybe both were planted. Maybe he did fired the gun before or after the murder, e.g., at the shooting range, and had consensual sex with the victim, only to have The Real Killer steal his gun, rape the victim while wearing one of Mike Nifong’s super-condoms, and then shoot her while wearing rubber gloves. Maybe the cops had access to the suspect’s DNA from some other source, and simply planted it at the scene. Maybe The Real Killer used a different gun, and the matches between the bullet and the barrel of this gun are the product of a bizarre coincidence. Even if the whole thing was witnessed by 10 people and captured on videotape, maybe all 10 of them are lying and the videotape forged. Maybe this, maybe that – a lot of things could have happened, which is why we only require proof beyond a reasonable doubt, and not proof beyond any doubt whatsoever, as some bored logicians imply that we should. Case in point:
Poisoning the well is a preemptive attack on a person in order to discredit their testimony or argument in advance of their giving it. A person who thereby becomes unreceptive to the testimony reasons fallaciously and has become a victim of the poisoner. This is a kind of ad hominem.
[Prosecuting attorney in court] When is the defense attorney planning to call that twice-convicted child molester, David Barnington, to the stand? OK, I’ll rephrase that. When is the defense attorney planning to call David Barnington to the stand?
This is an excellent example of poisoning a well that should be poisoned. No, it doesn’t follow logically that if Barnington is a twice-convicted child molester, he must be about to lie on the stand now, which presumably is the conclusion this prosecutor wants the jury to draw. But does anyone seriously doubt that a twice-convicted child molester is more likely than the average witness to lie on the stand? If so, try this simpler example, instead:
[Prosecuting attorney in court] When is the defense attorney planning to call that twice-convicted perjurer, David Barnington, to the stand? OK, I’ll rephrase that. When is the defense attorney planning to call David Barnington to the stand?
Formally, this is precisely the same “fallacy” as appeared in the previous example. Just because Barnington has twice been convicted of lying on the stand, doesn’t mean he’s going to lie on the stand this time around. He might not. But then again, he might, and given his track record he’s a lot more likely to than the average joe, so it’s not unreasonable to ask a jury to take this into account. Yet if we were to follow this author’s reasoning to its logical conclusion, we’d have to ignore Barnington’s past entirely, on the theory that anything which doesn’t mathematically prove Barnington is lying should not be considered at all. That may be “logical,” in the most formal sense of the word, but it’s also insane.
The reality is that when discussing arguments on anything other than logical proofs, some “fallacies” are truly fallacious in the common dictionary sense – i.e., wrong – while others aren’t. Formal logic won’t tell you which; only a close examination of the merits of the argument will. Take, for example, a slippery slope argument on the Supreme Court’s annoyingly expansive view of the Commerce Clause:
If Congress can tell individual farmers how much wheat they can grow on their own land to feed their own families, it can regulate just about anything.
This is a classic slippery slope argument, based on the notion that A (Congress being allowed to regulate individual farmers growing wheat for personal use) will necessarily lead to B (Congress being able to do whatever it wants). It is a very reasonable inference to draw, and with 20-20 hindsight, it also appears to be mostly correct. Yet, as a logical proof, it fails. Certainly, it was possible that the Supreme Court could have ruled for the plaintiffs in Gonzales v. Raich (affirming Congress’s power under the Commerce Clause to prohibit medical marijuana) without overturning Wickard v. Filburn (affirming Congress’s power under the Commerce Clause to limit individual farmers’ wheat production). It would have involved some fancy footwork, mind you, but that doesn’t mean it couldn’t have happened. The Supremes could have distinguished the cases any number of ways, had they been so inclined. For example, they could have noted that the regulation at issue in Wickard was a fundamentally economic in nature, that it exempted small-scale private farms, and its that its principal purpose, if not its full impact on Mr. Filburn himself, was to regulate the interstate trade of wheat, and not purely intrastate, non-economic activities analogous to growing and smoking one’s own personal supply of medical marijuana. Or it could have finally pulled its institutional head out of its rectum and looked to the obvious intent of the law, noted that it was not to regulate interstate commerce but to use interstate commerce as a pretext for regulating something else, and invalidated the statute on that basis. Or they could do as they’ve done for racially restrictive covenants in Shelley v. Kraemer or abortion in Roe, and limited Wickard to its subject, which was wheat, not weed. Or they could have gone totally loopy and discovered a constitutional “right” to smoke weed, lodged right there in the emanations and penumbras of the Abortion Amendment. Or they could have done just about anything else.
None of these alternative scenarios were ever particularly likely, of course; indeed, anyone who predicted that the court would have done them might well have been dismissed as a crank. Rather, when Raich reached the court, nearly all serious legal scholars did precisely what bored logicians argue that they shouldn’t have done: assumed that the slope was very slippery, and confine their debate to which direction we’ll slip in. Either the Court would continue down the slope that Wickard started us down, or it would reverse course and slip down the other direction, taking Wickard itself along with it (either immediatley, or in a future decision). The notion that the Court would treat the two cases separately – even though everyone knew it technically could – was generally dismissed out of hand.
The real problem, I think, is not with the concept of a slippery slope argument, so much as the lame scenarios to which it is often applied. I for one oppose gun registration because I think it is wasteful, bureaucratic and annoying, but I do not share the belief common among my fellow gunnies that it will lead to large-scale confiscation, and certainly do not share the view that any checks at all should be prohibited because they could lead to registration. To argue that such a law will lead to broad-based confiscation (as opposed to merely acknowledging the theoretical possibility that it could) requires a Rube Goldberg of increasingly dubious inferences. The most intuitive path I can think of is this:
- Allowing government to run background checks will give it information that could be, and sometimes is, used to create a national registry.
- Most people who favor gun registration also support other laws that burden gun ownership slightly more – e.g., waiting periods or “gun of the month” laws.
- Most people who favor waiting periods or gun of the month laws, also favor other stupid, pointless exercises that burden gun owners even more – e.g., prohibitions on “assault” weapons.
- Most people who favor all of the above restrictions also support a law that burdens gun ownership in a way that really matters: they support restrictions on law-abiding citizens’ ability to carry concealed weapons for their personal self-defense.
- Most people who favor all of these restrictions also support, or at least cannot bring themselves to openly oppose, laws that bar ordinary citizens from owning a gun at all.
- Therefore, requiring instant background checks will lead to the disarming of the American people.
Or, if you prefer, here’s the abortion version of this reasoning:
- Most people who support laws treating fetal homicide as murder, think late-term, unborn babies are people.
- Most people who think late-term, unborn babies are people, also think it should be illegal for their mothers to hire
hit mendoctors to kill them.
- Most people who think late-term or partial-birth abortion should be illegal, also think mid-term abortions should be illegal.
- Most people who think mid-term abortions should be illegal, also think most first-term abortions should be illegal.
- Most people who think most first-term abortions should be illegal, think all abortions should be illegal.
- Therefore, laws criminalizing fetal homicide as murder will lead to a ban on all abortions.
Each of these individual arguments, taken by itself, is probably true. However, the conclusions are almost certainly not, because at some point, a majority of a majority of a majority of a majority becomes a minority, the issues become more complex than advocates on both sides of the issue let on, and the slope ceases to be slippery. And therein lies the real objection to these arguments: it’s not an objection to slippery slopes per se, but to the fact that a particular slope is being portrayed as though it were more slippery than it really is.
Similar arguments can be made about other logical fallacies. If, for example, a person attempts to poison the well with a false accusation, or with an insult unrelated to the credibility of the statment he is attempting to discredit, then his opponent should rightfully call him on it, not because he’s poisoning the well, but because he’s doing so improperly.