Five dumb arguments smart people make

When smart people make dumb arguments they tend to fall into one of a few categories. I’ve documented five of the most common bad arguments I see at websites where otherwise intelligent geeks, math nerds and skeptics hang out and discuss things. Chances are you’ve encountered at least one of these arguments, maybe you’ve even used one of them yourself.

#1: Occam’s razor

In simple terms, the idea of Occam’s razor is that, whenever possible, simple models are to be preferred. Note that Occam’s razor tells you absolutely nothing about whether a model or a theory is good or bad, useful or worthless. It’s is a rule of thumb. And while not necessarily a bad one, in practice it tends to act like intellect retardant to put out active minds who question existing (often simplistic), beliefs and scientific constructs.

Let me make this very clear: Occam’s razon doesn’t prove anything. In particular, and despite how it is commonly used, it doesn’t show that the more likely, or “simple”, explanation is the correct one. Unlikely, complicated things happen all the time. If you don’t believe this, go flip a coin a thousand times. I guarantee you that the exact result you get will only happen once in a godzillion tries. In terms of (absolute) likelihood, explaining what you just observed with probability theory and stochastic processes is way more complicated than the assertion “that’s what God wanted”, a theory which, if true, explains your effectively impossible result with 100% likelihood. I’m guessing that’s not the can of worms you hoped to open up with Occam’s razor?

#2: You’re a hypocrite.

Yes, and so are you. So what? Do I really need to explain why this agumentum-ad-hypocrisum (note, made-up Latin) is so bad? Given that accusations of hypocrisy are almost as popular as kittens in the blogosphere, I suspect I do. So here goes: showing that your opponent is a hypocrite proves nothing except that they are human. It doesn’t make their arguments wrong, and only weakens them under very limited circumstances, like when you catch a sworn bretharian sneaking a pint of Häagen Dazs to keep from starving to death. Beyond that, calling your opponent a hypocrite has less nutritional value than a peep.

#3: That’s just an anecdote. It doesn’t prove anything.


Repeat after me: Anecdotes are evidence. Nothing more, but certainly nothing less. In the context of a common event followed by another common event (I got a headache after stopping at three red lights in a row), anecdotal evidence is nearly useless. In terms of more rare events (I got kuru after eating my sister), anecdotal evidence can be extraordinarily powerful. Taken to its extreme, discounting anecdotal evidence led (presumably intelligent) academics to hold firm to fallacies like the idea that fireflies can’t flash all in unison, long after anecdotal evidence had come in from reliable observers.

#4: No known mechanism


There are lots of intelligent ways to argue that homeopathy doesn’t work. You can cite studies (although you may find that not all of them confirm your beliefs), or you can say that believers are the random subset of the people who have tried it and then afterwords felt an improvement. I’m not going to weigh in on the subject, except to note that one argument I see used fairly often is that homeopathy doesn’t work because it can’t work, and it can’t work because we don’t understand how it possibly could.

To see how bad this argument is you need to look at the assumptions behind it and view it in historical context. What people are really saying with this argument is: Our current scientific model is comprehensive and infallible. It accounts for all observations, and it has no holes or leaks. I’m going to assume that you are able to see the problem with this mindset yourself. I’ll just note that one particularly unfortunate use of this argument led doctors to reject hand washing before performing operations or delivering babies. After all, evidence that fastidious midwives had lower infection rates was purely anecdotal (see above), and there was no reason to believe cleanliness could make a difference in the pre-germ era. There was no known mechanism.

#5: Three guys with boards


I’m calling this the “three guys with boards” argument in honor of those skeptics who, whenever someone mentions crop circles, declare “it’s a hoax. Three guys with boards admitted they did it.” In fact, some guys with boards did indeed admit to creating a crop circle, and they showed us how they did it. So what’s wrong with this argument?

The problems is that you can’t discount all anomalous observations as “fakes” just because some are known to be fakes, nor does the possibility of faking an event mean that all such events must have been faked. Obviously, scientists don’t like playing games of intellectual wack-a-mole. If you research enough supposed “unexplained mysteries”, and come away convinced each time that the mystery is bogus, the tendency to dismiss other, similar claims outright is understandable. That said, “three guys with boards” is still a dumb argument, especially when you are going against claims of specific evidence that can’t be easily explained away as a hoax. For example, we have large, complicated, precisely implemented crop circles done in a short span of time and exhibiting strangely bent stalks. This evidence may fall far short of irrefutable proof of alien intervention, but it does require much more than dismissively stating that we know they are all fakes. We don’t.

More broadly, any attempt to automatically sort new observations into known categories (often categories that make us comfortable) is a bad idea. Unfortunately there are no shortcuts when it comes to evaluating data or evidence. It has to be done the hard way, one piece at a time.

Tags: , , ,

32 comments

  1. Is that Bill Murray? WTF does he have to do with Ocam’s razor?

  2. I’m not sure why you think Metallica is a band of hypocrite. Is it because of their stance against music piracy? That was a very principled thing for them to do it does not make them hypocrites.

  3. #1 Occam’s razor: a tool used to help stop you from wasting your fucking time and energy evaluating possibilities that are more likely to be retarded

    #2 Being a hypocrite certainly is a valid point to make when arguing with someone whose integrity just fell out his asshole. again, a way to save you from wasting time on the hypocrite any longer

    #3 Anecdotes are illustrations used to exemplify your assumptions. It *may* qualify loosely as evidence, but it’s here-say unless backed with, you know, ACTUAL evidence.

    #4 No known mechanism: a valid stance used to, again, help you NOT waste your fucking time pontificating on something you ‘believe’ to be true. if it can’t be recreated in a controlled experiment, it is not scientifically verifiable, and therefore, don’t waste my time with your voodoo

    #5 Three guys with boards: Every week, i put the garbage out on the street. and almost every week aliens come and steal it!! I’ve only ever seen these “garbage men” come by like, TWICE while I was home. There’s no way you can tell me Alien’s DIDN’T steal my garbage when I wasn’t looking. So fuck you – please note the sarcasm. and oh yah, occams razor says you’re an idiot

    • I thought I was the only one who had this exact reaction while reading this. It’s pretty stupid. Especially the part about anecdotes.

  4. three guys with boards admitted that they did do some. other guys with boards followed in their footsteps by doing some, thus proving that guys with boards and a head for elementary geometry could do them. while that is anecdotal evidence, it is still no less than evidence, and when we apply the rule of thumb known as Occam’s Razor it is much more reasonable to conclude that guys with boards did all of them than it is to conclude that space aliens traveled many lightyears to earth just to make abstract art, and completely forgot to enslave us all and steal our natural resources in order to fund the rather expensive journey because there is no known mechanism by which they would do such a thing.

    but maybe I’m just being hypocritical.

    • Sac
      “three guys with boards admitted that they did do some”.

      It’s interesting that skeptics are so eager to unquestioningly believe an “admission” when it conforms to their own belief system, but that other “admissions” (eg I admit I saw and conversed with a ghost and he told me who had murdered him) are just dismissed as anecdotal or liars and certainly constituting no evidence whatsoever.

      Either skeptics realise their inconsistency here, which makes them dishonest. Or they do not, which makes them hopeless at arguing.

  5. “explaining what you just observed with probability theory and stochastic processes is way more complicated than the assertion “that’s what God wanted””

    Complicated != ‘length of description’. Or, rather, your ‘simplistic’ explanation “that’s what God wanted” presupposes the existence of God, his powers, etc, and is actually a very complicated assertion compared to the probability theory (which can actually be relatively simple, all things considered).

    But never mind me. I’ve got some crop circles to go make.

  6. @fakey:

    Kolmogorov complexity := “length of description” (some qualifiers apply, as always wikipedia has the details)

  7. I should have been more descriptive in my complaint (ironic, I suppose)

    Yes, Kolmogorov complexity is defined to be the length of the description. However, invoking God in the explanation requires additional complexity which is hidden through the use of the word ‘God’ (which is by its very nature imprecise), and an accurate description/complexity of the assertion “that’s what God wanted” would include that. The ‘simple’ explanation whose description is ‘of short length’ is in fact quite ‘complex’, because the true description is not in fact ‘of short length’. And ultimately, if you’re flipping coins and trying to explain what’s happening, it’s either random chance derived from some math, or some all powerful dude in the sky who has inexplicably a.) decided to take an interest in this coin flipping experiment, b.) has decided on a particular outcome he wants, and c.) is able to create the result he wants through some unspecified power to manipulate things. The complexity comes from this additional dude, who needs all these cool powers to do something which can be explained without him. The article seems to claim that ‘God wanted it to be so’ is a simple explanation, when it in fact isn’t; it just appears that way because of the hidden complexity of God.

    I’m not saying that Occam’s Razor is always true — it is, after all, just a guideline. I’m saying that this is an incredibly, astoundingly wrong example to try and prove that point.

    • I’m stealing your argument here to use against the next ID (Intelligent Design) nitwit who tries to tell me that his/her explanation of “goddidit” satisfies Occam’s Razor much better than “it changed gradually by random chance over cosmological time scales”. ;)

  8. i’m not really sure how you apply occam’s razor to coin flips…

    hmmm… i’ve flipped this coin 1000 times and it’s come up heads quite a bit, but also quite a few tails. occam’s razor suggests that…. likely… if i flip the coin again it’s got nothing to do with aliens and it’s probably god who decides which face of the coin is face up.

    nope. doesn’t make sense. nor does a lot of your post.

    also, try proofreading sometime.

  9. Metallica is a band of hypocrites exactly because of the stance against piracy when they themselves stole many songs from other artists.
    Principled my arse.

    also, this write up is pure crap.

  10. People seem to be having trouble understanding why the use of Occam’s razor is so problematic. Basically, there are 2 ways the argument is used. One: To say that the “simplest” explanation is better, or more correct. Two: to say that the most “obvious”, or likely (within the author’s worldview) explanation is correct. The problem with “simplicity” is that you are often just kicking the can down the road, and you get into discussions like those from fakey about what is simple: it depends on your prior beliefs and knowledge of course, which in turn depend on other things.

    The problem with demanding that people draw the “obvious” or most likely conclusion from an observation is that you have to get into what these notions mean, so once again Occam hasn’t done you any favors. Even worse, this particular form of Occam is like bad Bayesian thinking. Here’s where the coin flip example comes in. If you actually had flipped that coins 1000 times and wrote down the sequence (call it 0′s and 1′s), you would end up with a series of digits that had only a 1 in 2^1000 chance of happening (presuming the stochastic process with fair coin hypotheses is correct). Under this framework, the particular result you actually got is *way* less likely than not getting that particular result. By contrast, “God” (all-powerful, all-meddling) or “fate” makes your result inevitable, probability of 1. Now you can go in and say, Well, what’s the likelihood of God or fate? And of course how you answer that depends on your framework of beliefs. Regardless, Occam has led you not to clarity on how to interpret evidence but to religious debate, as I say, a can of worms you probably didn’t want to open by throwing out Occam.

    BTW the band pic isn’t of Metallica, it’s Hypocrisy.

    • Occam doesn’t say “simplest” – rather, it advises striving for the explanation that makes the fewest assumptions. Which is far more reasonable. It doesn’t preclude the correct explanation being more complex, but in the absence of certainty it says, basically, don’t invent stuff that can’t be observed if you don’t have to – surely?

      It is often mis-used as a “slam dunk!” to dismiss an explanation though, even when an alternative description leads to the same results but with, say, a more intuitive picture of how it’s happening.

  11. As Iratian (rather gratingly) notes, all five of these arguments are really rules of thumb that help direct scientific inquiry toward productive (and publishable) results. I think they vary in their usefulness, but that’s their purpose–rules to winnow down the direction of scientific focus.

    For your part, you consistently forward the idea of embracing the complexity of the world in its fullest sense (hence your manifesto, which I’m rather fond of).

    What’s the way forward? It’s good to be aware of the limitations of our rules of thumb, but is there a better way? How much time should we really be spending investigating crop circles? Using these rules of thumb I think is a similar principle for any aspect of model-building–it’s a way to direct our energy and wrap our mind around aspects of this poorly-understood world. When we don’t really know what’s going on anyway, I think it’s reasonable to have a preference for simpler solutions, even if there’s some anecdotal evidence against it, that rely on mechanisms we understand, and that we have a good reason to believe they aren’t hoaxes.

    If more evidence comes up, we abandon the theory, but these “dumb arguments” can provide a good starting point. Statistics is a human endeavor, and these five arguments are arguments to help make the field comprehensible and doable for us humans.

    • Ethan Brown

      “As Iratian (rather gratingly) notes, all five of these arguments are really rules of thumb that help direct scientific inquiry toward productive (and publishable) results”.

      This is just silly.

      First of all we’re not talking about scientific inquiry here. Rather what we are concerned about is whether something is true or not, or whether some alleged phenomenon exists (not necessarily that the phenomenon should come under what would be considered scientific knowledge. Consciousness cannot be scientifically explained, however that certainly doesn’t entail that we don’t know it exists) .

      To say that an anecdote constitutes no evidence whatsoever is simply flat out false. If a million people independently at different times and places claim to observe a certain characteristic phenomenon, would that really constitute no evidence whatsoever for the existence of the said phenomenon? Do you have any notion whatsoever how utterly absurd this position is?

      Also if some explanation is simpler, but ignores a whole realm of evidence, where’s the more complex explanation takes into account all the evidence, then it’s utterly foolish to insist the simpler explanation is vastly more likely to be correct (and no I’m not attacking a straw man, over the years I’ve argued with many people on the net who claim the incompatible evidence is anecdotal and therefore doesn’t count as evidence against the simpler explanation!)

      And as for the no known mechanism, what should we make of the hypothesis that the brain produces consciousness when there’s no conceivable mechanism as to how it could do so? Does the “no known mechanism” thereby rule out that the brain can produce consciousness?

      Over the years people have continually used these tactics against me, especially Ockham’s razor. These tactics are used when they can’t think how to address my arguments. All discussion then comes to an effective end.

  12. @Ethan Brown

    Very well put.

    I think you could view these arguments as “rules of thumb” that can protect researchers from wasting energy perusing unpromising (or unpublishable) lines of inquiry. That’s why smart/savvy people use them; they are shortcuts that help cut through the noisy world of evidence. So long as you recognize the shortcut and are conscious of the potential costs, that’s fine.

    Unfortunately, I often see them offered up as direct arguments for why a person or theory is wrong.

    For those who haven’t seen the “manifesto” Ethan mentions it’s here: http://www.statisticsblog.com/manifesto/

    The motivation for this blog post is found in manifesto item number 1.

  13. Antryg Windrose

    #t.dan: The Razor’s Edge

    Occam’s Razor: “entities must not be multiplied beyond necessity” which means, the simplest COMPREHENSIVE explanation is probably right. Classic example: Copernican dynamics vs Ptolemaic epicycles.

    Anecdotal evidence: Eyewitnesses are too often wrong. We see what we expect to see, and don’t see the unexpected. And people lie. (Cf. hypocricy.)

    Absence of mechanism: If the mechanism can’t be found after a couple centuries of looking, maybe there’s no there there.

    Guys with boards: Cf. Occam’s Razor. Guys with boards are a simpler comprehensive explanation, than are mystical aliens with weird decorative habits.

    • Antryg Windrose
      Occam’s Razor: “entities must not be multiplied beyond necessity” which means, the simplest COMPREHENSIVE explanation is probably right. Classic example: Copernican dynamics vs Ptolemaic epicycles.

      Ian
      Yes that’s fine. But the people I argue with define the simplest explanation as any explanation consistent with their metaphysical presuppositions about reality.

      Antryg Windrose
      Anecdotal evidence: Eyewitnesses are too often wrong. We see what we expect to see, and don’t see the unexpected. And people lie. (Cf. hypocricy.)

      Ian
      The fact that eyewitnesses are often wrong doesn’t entail that they are *always* wrong. Eyewitnessess are often right too. If it were the case our observations are no more likely to give correct information as to what is out there then give false information, then we might as well spend our whole lives with our eyes closed.

      The truth of the matter is that skeptics don’t like people witnessing phenomena which they don’t believe exists It doesn’t fit into their metaphysical pre-suppositions about the nature of reality. “Eyewitnesses are often wrong” is an auxiliary hypothesis to save their worldview. If many eyewitnesses independently observe something similar, then although they might all be somehow mistaken, it nevertheless is unlikely.

      Antryg Windrose
      Absence of mechanism: If the mechanism can’t be found after a couple centuries of looking, maybe there’s no there there.

      Ian
      And maybe there is no mechanism, however that doesn’t entail the phenomenon in question doesn’t exist. Perhaps the mechanistic philosophy is wrong . .

      Antryg Windrose
      Guys with boards: Cf. Occam’s Razor. Guys with boards are a simpler comprehensive explanation, than are mystical aliens with weird decorative habits

      Ian
      So either your hypothesis is correct, or some other extreme hypothesis is correct. You really regard that there’s no other possible hypotheses?

  14. Without even speaking from a position of authority, it should be obvious your statements about anecdotes aren’t even incorrect.

    Imagine you conducted an experiment to test the efficacy of a new medicine. For simplicity’s sake, let’s say that after taking the new medicine a test subject either got better, stayed the same, or got worse. Well test subject John took the medicine and got better. What does that tell us?

    Let’s even look at the plural of anecdotes. 30 people taking the medicine got better. How much more helpful is that information? Absolutely none, and you’re off your rocker if you think it does — especially if we look at the EVIDENCE that in a 160 subject group 30 people got better, 30 people got worse, and 100 people stayed the same (for example). That’s the difference between anecdotes and evidence, and anecdotes are not evidence at all. (Well, a better explanation of it anyways.)

    • the circus:

      “That’s the difference between anecdotes and evidence”.

      You in common with so many people (and especially scientists) simply do not understand what the word evidence means.

  15. Anecdotes (reliable ones at least) are numerators. In the presence of huge denominators, such as what you have in circus’s example above, anecdotes don’t sway very much. In the presence of small denominators, such as what is in the main entry, they can hold huge sway.

    All of the arguments mentioned above are heuristic and are very useful in some circumstances (maybe deciding which hypothesis to pursue in your next R01, why not), but are not proofs. If you want a proof or watertight argument, it’ll cost you.

    Well done, Matt.

  16. Occam’s Razor is stated as “the explanation which requires the FEWEST ASSUMPTIONS OF THINGS NOT IN EVIDENCE is usually the correct one.” For simplicity’s sake, it is often abbreviated “the simplest explanation…”

    You took the dumbed-down version and went on to attack from there, with none of your arguments actually applicable to the original expression. That is not an impressive critique; it makes you look like Don Quixote.

  17. Matthew Bailey

    “Occam’s Razor” says:

    DO NOT multiply assumptions unnecessarily.

    It DOES NOT SAY:

    The simpler model is preferred.

    While the second statement is a basic summation or simplification of Occam’s Raxor, it is NOT correct as an overall generalization, though.

    This is why “Godidit” is wrong, even though it is a “simpler” explanation in terms of difficulty of statement.

    But it terms of explanatory power, “Godidit,” is more complex, as it leads to positing something unexplainable for an explicable process

  18. Matthew Bailey

    And ANOTHER thing you should point out (two things really) is:

    Well, so and so did that too?

    Or…

    Both sides do X.

    The first usually occurs in discussions of contemporary religious behavior, where someone will invitably say “Well, Christians do the same thing (or used to do the same thing).”

    This is attempting to dismiss one wrong by citing another.

    The second point is the False Equivalence, or Pot Calling the Kettle Black.

    Here, we have people attempting to dismiss something because “everyone does it,” or because they haven’t really examined the evidence.

    This is still a variation of two wrongs don’t make a right.

  19. This page is a great example of the Straw Man fallacy. Repeatedly, the author misrepresents a subject in order to attack his own twisted version of it. In other words, the author is intellectually dishonest and/or an idiot.

Leave a comment