I've been going through Peikoff's Introduction to Logic Course (the videos for which are available on YouTube. The handout with the exercises is freely available on the Ayn Rand Institute's website for the course. I'll be posting some exercises I've done in the course soon, but my purpose for this post is to criticize Peikoff's advocacy of mistaken methods of dealing with ideas. In doing so, I will be drawing very heavily on ideas and arguments from Elliot Temple's Yes or No Philosophy course, along with my background knowledge of other relevant thinkers like Karl Popper. This post draws in particular on the "Support Contradicts Logic" and "Burden of Proof" videos in Yes or No Philosophy. However, the analysis is my own, as are any errors. I'm going to do at least a couple of posts on this topic.
Argumentum ad Ignorantiam & Arbitrary Assertions
I'll start by criticizing some statements that begin around 1:22:50 in the "Informal Fallacies, Part 1" video. Below is a partial transcript (my own) of Peikoff's introduction to Argumentum ad ignorantiam, with my comments interspersed:
Before I explain what the nature of this fallacy is, I want to give an example at the outset. The best way to do this is point to someone, for instance you, and I will make a statement, I'll say that in the last two weeks, you committed a particularly gory murder. Now I put this forth as an arbitrary assertion. Remember what we mean by arbitrary? Something backed with no evidence, no proof.
"Backed" indicates that Peikoff is thinking of the idea of support. Peikoff also brings up the idea of having no evidence.
When does evidence "back" an idea? What is meant by "backed" in this context? Does it mean that evidence doesn't contradict an idea? Because if so, then lots of things can be said to be evidence. For example, if you accuse someone of murder and the person accused exists and is alive, that's "evidence" for your murder theory (and it's actually relevant evidence, and not just some fact that doesn't contradict, but is basically irrelevant to, your theory).
People generally mean something more specific by evidence than "doesn't contradict my theory". For example, they might consider it to be incriminating evidence if the accused was near the murder at the time it happened. And they might consider it exculpatory evidence if the accused was far away at the time of the murder. But, obviously, someone near a murder scene at the time could have been innocently having tea with a neighbor, and someone far away could have hired a hitman. So even the stuff that seems like "good" evidence is compatible with a variety of theories. Peikoff claims that you can have an assertion backed with no evidence, but he doesn't explain what he means by this claim, and interpreting evidence as meaning "doesn't contract a theory" means that our evidence "backs" a bunch of theories. I think there's no good, formal, rigorous definition of what Peikoff means by evidence backing a theory that would survive criticism. People have an intuitive sense of when they think evidence backs a theory, but an intuitive sense of something isn't sufficient for careful philosophy discussions.
Simply a caprice, a wanton bolt from the blue based on nothing. Now remember that's what we mean by arbitrary. Now I make this charge about this lady, I say: you've committed a very gory murder in the last two weeks. I don't give any argument in favor, but I say prove that you did not do it. Disprove this claim on my part. Now the fact is, as I'll show you in a moment, she cannot do it. No one can disprove such a charge, as long as I put it forth arbitrarily and am I free to continue to make arbitrary assertions – so long as I am free arbitrary, baseless, gratuitous assertions, it is impossible to prove that my original claim is was false.
I think that it is possible to criticize people putting forth what Peikoff considers arbitrary assertions. But the criticism isn't based on lack of supporting evidence (which cannot be coherently defined in a useful way). Instead, one can point out flaws with certain claims, and more importantly, with certain patterns or classes of claims. I'll provide more detail after letting Peikoff develop his example more.
For instance, I say she committed the murder. She says, oh I can refute that. For the last two weeks, that's when you say the murder was committed, I had witnesses with me every single night and I can produce the witnesses. To which I come back and say, well witnesses can be bribed, I claim you bribed those witnesses. Prove that you didn't.You would say to me alright, I'll do it from another point of view. Where's the corpse? How can you claim I committed a murder if there's no corpse? I say you dissolved the corpse in acid. Obviously it doesn't exist. Prove that you didn't. You say to me well look, I'll give you the most powerful argument. The murder took place in Australia and I was in a hospital in Manhattan in a plaster cast for the whole two weeks. Now is that a refutation or not? And I say doesn't prove a thing. Obviously what you did is you hired a consummate actor standing [?], made her up like you, got her in a hospital in a plaster cast, went off to Australia, committed the murder, rushed back, changed places under cover of secrecy, so obviously that doesn't disprove my claim. So prove that you didn't do it.
This pattern of claims is vulnerable to various criticisms. Some points you might raise with the accuser are:
- Rather than conducting a thorough investigation and coming to a conclusion based on reasonable methods, they seem determined to say you're guilty. Their "method" consists of addressing counterarguments in an ad hoc way. This isn't a reasonable, truth-seeking method, but a biased method. And it's not how e.g. criminal prosecutions work. In a criminal prosecution, the prosecutors have to lay out some specific, detailed, positive factual claims about what happened, with sufficient specificity as to each element of a specific crime. By the ad hoc method, one could "prove" that anyone was guilty of murder. But most people don't commit murder. So the ad hoc method is bad and would be likely to produce unjust results if taken seriously as a method of establishing guilt for crimes. Rebuttal?
- A similar method could be applied to the accuser to illustrate the problems with the method and to negate their claims. E.g. you could claim that they are accusing you only to deflect from their own crimes. Or you could claim that an extraterrestrial brain slug is causing them to hallucinate that you committed the crime. And in each case of their attempted rebuttal, you could make ad hoc changes and ask for them to prove that you're wrong. The fact that you could construct a countervailing ad hoc system of claims to negate their claims shows that systems of ad hoc claims are not a reasonable truth-seeking method, since they don't allow you to settle issues and get to the truth. Rebuttal?
- They are making strong claims without considering the hugely problematic issues those claims raise (and without pre-emptively addressing them). Some specific issues:
- Did they even consider whether you have the resources to bribe a bunch of witnesses? And even if you did, maintaining their silence would raise similar issues to maintaining the silence in a large conspiracy (and the difficulties involved in doing that are a good criticism of many common conspiracy theories).
- Did the circumstances of the murder take place under conditions where you would even have had the time to dissolve the corpse in acid?
- Did they consider that an actor in the hospital might be subject to medical tests which would reveal the ruse?
- Given that they seem determined to find you guilty regardless of the facts presented, why is their claim of your guilt and demand for you to disprove your guilt a worthy topic of discussion? They don't actually seem interested in a rational discussion on the topic, but instead only want to validate their pet theory about your guilt. Why should you participate in a sham discussion with someone who's already apparently made up their mind?
- Can they provide a high level summary of why they've decided that you're guilty and how they came to that determination in a way based on criteria that applies only to you and not to other people? In other words, can they provide some high level argument or summary about why they're not being biased?
These arguments raise issues/claims/points at a higher level of discussion than just arguing the specific ad hoc factual claims that the accuser is bringing up. And this isn't even a comprehensive list or anything – it's just some stuff I thought up fairly quickly. I think there's a lot more detail here in these criticisms that gives the discussion a lot more places to go than just stringing a bunch of adjectives together in front of "assertions" (as Peikoff does when talks about "arbitrary, baseless, gratuitous assertions") and calling it a day. There is a lot more analytical detail about problems with the method the accuser is using to go into than just calling it "arbitrary", and the points raised above address pre-empt whole classes of potential arguments.
For example, the argument that the method of throwing out ad hoc claims could "prove" anyone guilty of anything is absolutely devastating for one using the ad hoc method. That argument cannot be addressed by throwing out another ad hoc claim and saying that maybe the actor you hired was a genetic twin. The argument isn't operating on that level, but is instead pre-empting a whole class of claims from being relevant to the discussion until and unless it is addressed. While, by introducing that argument, you haven't "proven a negative" in the sense of definitively establishing the falsehood of the claims of the accuser, you've done something arguably better: you've made his claims regarding your guilt irrelevant unless and until he addresses your argument.
Now it should be obvious to you that you cannot prove that she didn't do it if I am free to make arbitrary claims. It is impossible. Well, if I say from that you can't disprove, you do not know how to disprove this, therefore it is true, it is true, therefore you did commit the murder, that is argumentum ad ignorantiam.
Peikoff thinks that if people are allowed to make arbitrary assertions, then they can bog you down with endless BS claims. This misses the fact that a criticism can negate/refute/pre-empt from relevance an entire class of factual claims until it is addressed, as we saw above. He constructs his definition of a fallacy on his premises, but I think the way the fallacy is defined is misleading at best. If someone says "You can't definitely disprove X, therefore you must accept it", then that is mistaken, but it is mistaken because the idea of definitely proving or disproving stuff is misconceived, and so any demand for someone to do that is mistaken. However, if someone says "If you can't refute X, you should tentatively accept it", that is a different proposition. If an idea survives every criticism you know of, including criticisms such as the above that talk about bad methods and bias and apply to entire classes of claims, what reason is there to not accept it? I think the "Argumentum ad Ignorantiam" could be taken as rejecting "If you can't refute X, you should tentatively accept it" as invalid, and thus I think it is problematic as a fallacy.