What are some famous or notable counter-examples to Occam's razor (specifically, instances where a simple, obvious explanation is wrong and a complicated explanation is correct).

December 9, 2012 9:41 PM

What are some famous or notable counter-examples to Occam's razor (specifically, instances where a simple, obvious explanation is wrong and a complicated explanation is correct).
posted by jtothes to Grab Bag (47 answers total)

We landed on the FUCKING MOON!
posted by bensherman at 10:05 PM on December 9, 2012 [5 favorites]

Monty Hall problem.
posted by supercres at 10:09 PM on December 9, 2012 [8 favorites]

Birthday Paradox
posted by tiamat at 10:20 PM on December 9, 2012

Relativity versus Newtonian physics.
posted by Chocolate Pickle at 10:23 PM on December 9, 2012 [3 favorites]

And quantum mechanics, no? Basically all of modern physics.
posted by Justinian at 10:24 PM on December 9, 2012 [1 favorite]

Best answer: Hickam's dictum argues that Occam's razor is of limited use in medicine because patients are likely to have multiple diseases/health issues contributing to their symptoms.
posted by she's not there at 10:24 PM on December 9, 2012 [13 favorites]

The earth is still flat, right?
posted by mannequito at 10:24 PM on December 9, 2012

Response by poster: The Monty Hall problem comes very close, but I don't think it quite fits the bill. I'm having a difficult time articulating why that is.

I think it's that there is no simple, widely believed, but incorrect explanation of why switching doors increases your odds of winning (although there is a simple incorrect explanation of why it does not affect the odds).
posted by jtothes at 10:40 PM on December 9, 2012

How about "God did it" vs. "Cosmology + Biochemistry". Most biochemical explications of processes strike me as complicated messes looked at more than just superficially. For example, look at this process.
posted by benito.strauss at 10:48 PM on December 9, 2012 [6 favorites]

Keynesian economics?
posted by moorooka at 11:05 PM on December 9, 2012 [1 favorite]

Law. Laypeople often misstate the law by making it too general, e.g. "A contract has to be in writing." In fact, there are complex rules about when contracts need to be in writing. Still, the best approach is to put everything in writing, and people who try to enforce unwritten contracts have an uphill battle. Also, people often try to explain the outcome of a specific case with a single clear rule, which is usually wrong with a fact-sensitive area like family law.

The Monty Hall problem is not a counterexample to Occam's Razor, as the OP pointed out. It's a case where people expect the simpler answer to be right, but they're wrong. The OP is looking for cases where a simpler explanation might seem to explain the truth, but the explanation is wrong. Also, the truth of the Monty Hall problem actually has a pretty simple explanation: your initial choice has a 1 in 3 chance of being right, and the host's opening of a door can't affect that probability since the host always leaves your door alone. That isn't complicated. People overlook it, and they might not be convinced by it, but it isn't complicated.

I also disagree with the "God did it" answer. God's will is a complicated explanation, since his will could always change, and you need to figure out his specific will for each event that results from it. Science is simpler: it gives universal rules, which apply everywhere and always.
posted by John Cohen at 11:11 PM on December 9, 2012 [1 favorite]

The Kennedy assassination?

Okay, yes, the official line is still that there was only one shooter and it was Oswald, but there's obviously at least a little bit more going on there than that...
posted by Scattercat at 11:36 PM on December 9, 2012

The Monty Hall problem and especially the birthday paradox are totally irrelevant here, almost unintelligibly so, because Occam's Razor concerns how one chooses between hypotheses about empirical phenomena.

Anyway, to be clear, your rephrasing of the scenario you're interested in the comment above and your question in the OP aren't equivalent. Occam's Razor specifically applies to the situation in which one is choosing between competing models that describe observed phenomena. The implication here is that these competing models all describe the observed phenomena equally well* and that several models are currently in play. This rules out scenarios in which a later theory supplanted an earlier one, as with Newtonian physics and relativity, because in cases like that the later theory was spurred by observations that couldn't be explained by the previous theory. In light of the discovery that the speed of light is constant in all reference frames, special relativity is actually an extremely parsimonious theory, which is part of its beauty. "Quantum mechanics" is an ambiguous case because it's not clear what part of it is being referred to. J. J. Thompson's cathode ray tube experiment, which was the first experimental result favoring a particle conception of electricity, was performed in 1897, so at least in its early stages QM too was not an instance of a theory that was competing against equally descriptive models.

So if I understand you correctly, you're asking for situations in which a model or theory with more parameters or more assumptions was chosen over one with fewer of either. I think that has to be it because the other interpretation that people are rolling with here is trivially answerable: every theory that supplanted an earlier, simpler theory would qualify. I can't come up with any good answers off the top of my head, not really knowing much about the history of science, but I suspect that the actual instances of this are rare and obscure, because scientists favor parsimony very heavily. This qualification seemed necessary, though.

* There are complications here that I'm papering over because they're outside the scope of this argument -- what if there is disagreement about the notion that two competing theories do in fact describe observed phenomena equally well? The Kennedy assassination is a good example, actually, because believers in multiple-shooter theories think that there are facts that the single-shooter theory can't explain. I don't think this is as common in science, though.
posted by invitapriore at 11:44 PM on December 9, 2012 [17 favorites]

Pretty much all magic tricks.
posted by ryanrs at 11:44 PM on December 9, 2012 [5 favorites]

Perhaps some kind of complicated social conflict? Like, the simplest explanation for his late nights is the one he offers, i.e. that he's working overtime because of the big upcoming project, and every previous time it's turned out to be nothing, but this time he really IS cheating on you with the mail boy?

I can't volunteer any "famous" examples, but I'm assuming that there has to have been one somewhere in the annals of celebrity gossip magazines.

---

I'm finding it hard to contemplate this because Occam's Razor is really more of a guideline for testing theories; start with the simple ones and don't assume any more than you have to. It's hard to posit "counterexamples" in a meaningful way.
posted by Scattercat at 12:02 AM on December 10, 2012 [2 favorites]

How about the friendship paradox? The paradox says that on average, your friends will each have more friends than you do; your business partners will have more experience than you do; the people you meet while exercising will exercise more than you do. But! This does not necessarily mean that you are a lonely, inexperienced, pudgy loser. It is because you are more likely to make friends with someone who is very sociable; you are more likely to enter into a business relationship with someone who is entrepreneurial; you are more likely to meet people at a gym if they spend a lot of time there.

When you understand the paradox you will see that these people are not typical. You don't need to feel bad about not being like them. It is very likely that you are already as friendly, as entrepreneurial, as athletic, as the average person. Their successes are not what is expected of you; they just show that it can be done. If their examples inspire you that's great; if they don't inspire you then you are still perfectly OK.

I have felt a lot better about being pudgy since I realised this.
posted by Joe in Australia at 12:14 AM on December 10, 2012 [22 favorites]

I think maybe the answer here is to spend some more time learning about Occam's Razor and what it does and doesn't mean. You said:

a simple, obvious explanation [...] and a complicated explanation

there is no simple, widely believed, but incorrect explanation

But Occam's Razor has nothing to do with obviousness. It's entirely possible for an elegantly simple explanation to be far harder to come up with than a complex one. It also has nothing at all to do with the popularity of a given explanation.

Anyhow, since the principle of the Razor is basically "don't multiply entities [or elements of your explanation] unnecessarily", it's not actually possible to come up with a "counterexample" by looking for complicated explanations. If a more complicated explanation is necessary, then Ockham is all for it.
posted by RogerB at 12:18 AM on December 10, 2012 [6 favorites]

The Dreyfus Affair, where the simple explanation (soldier sold secrets) was wrong, and the correct explanation was an extensive government conspiracy.
http://en.wikipedia.org/wiki/Dreyfus_affair

More generally, I'd say that most frame-ups or cover-ups would meet the criteria.
posted by Balna Watya at 12:55 AM on December 10, 2012 [2 favorites]

RogerB, Ockham usually comes up or is invoked when it's not easy or convenient to empirically or feasibly resolve an issue, so it's used as a guide to select between a simpler plausible explanation and a complicated but still plausible explanation. IOW, Ockham is only applied amongst explanations, each of which could conceivably fit as a possible answer, matters of elegance, rigor..etc notwithstanding.
posted by Gyan at 12:56 AM on December 10, 2012 [1 favorite]

What rogerb said. It's about choosing the explanation that makes the fewest assumptions, not the explanation with the smallest number of parts.
posted by obiwanwasabi at 12:58 AM on December 10, 2012 [4 favorites]

Phlogiston.
posted by Obscure Reference at 1:11 AM on December 10, 2012 [1 favorite]

The sun going around the earth.
posted by Obscure Reference at 1:12 AM on December 10, 2012 [1 favorite]

Any paradox classified as a "veridical paradox" would likely be close to what you're talking about. One of my old math professors defined a paradox of this sort as "a surprise"--ie, the result you naturally expect from basic first principles is wrong and there is a complicated, convoluted, and unexpected explanation that is actually true. So those are "instances where a simple, obvious explanation is wrong and a complicated explanation is correct".
posted by flug at 1:15 AM on December 10, 2012 [1 favorite]

The wiring of the human visual system. It's quite complex and is a mishmash of changes that happened at various times in evolution.

When early scientists were dreaming up how vision should or could work, they came up wtih simple theoretical models that ended up being nothing like the reality of how it works. Pretty much all of the early simple elegant models were wrong. The visual system is a bit arbitrary in how it is wired up. For instance, the wiring of blue-yellow color perception is quite different from red-green, e.g. in how areas for color are connected to areas for depth. The wiring lacks symmetry. There is no special meaning there... blue-yellow just evolved later and is wired like an add-on to an already complex system. There are many such examples in vision.

Human vision, for being so beautiful and fundamental to our conscious experience, is not at all simple. It's sort of a kludgy bastardized monstrosity, more like your weird uncle than Heidi Klum. The wiring diagrams are not pretty. It still works nicely though!
posted by kellybird at 1:22 AM on December 10, 2012 [2 favorites]

The sun going around the earth.

Actually that's a bad example. Elliptical orbits are far simpler than Ptolemaic epicycles on top of epicycles. The Ptolemaic system of calculations was actually extremely accurate. Initially, simplicity was the only advantage of Kepler's system -- Occam's razor at work.
posted by justsomebodythatyouusedtoknow at 1:26 AM on December 10, 2012 [2 favorites]

Epistemological obstacles, as defined by Bachelard, tend to create such situations, when people try to understand complex things through simple metaphors. Numerous examples can be drawn from the history of our understanding of the body: the lungs acted like a pair of bellows firing and cooling a furnace and nature [...] relegated the stomach or bowels farther away from the site of reason and of the mind and fenced it off with the diaphragm in order not to disturb the rational part of the mind with its importunity.
posted by elgilito at 3:18 AM on December 10, 2012 [1 favorite]

The most obvious category for what you are looking for are wrongful convictions, or cases where a scapegoat takes the fall.

So: this 1994 British helicopter crash would qualify, as would the case of Sam Sheppard.
posted by MuffinMan at 5:12 AM on December 10, 2012

Infections as the cause of stomach ulcers rather than stress, spicy food or alcohol?
posted by MySockyWocky at 5:33 AM on December 10, 2012 [1 favorite]

When looking at windows on old buildings, the glass tends to be thicker at the bottom. This isn't because glass is still slightly liquid, but because the style was to put the thicker parts of imperfectly blown glass at the bottom of the pane.
posted by Phredward at 5:46 AM on December 10, 2012 [3 favorites]

Underlining what rogerb said: Occam's razor is not about obviousness or common sense. Although I've always thought there's a question-begging aspect to it, insofar as the criterion of "unnecessariness" (as in "multiplying entities unnecessarily") would seem to be wholly dependent on which explanation actually turns out to be true. Maybe someone can correct me on this.

Anyway… social science is a good place to look for explanations that are more complex than the simplest explanation that might feasibly account for the phenomenon in question. For example, look at the countless ways organizations, institutions, governments end up pursuing certain goals for reasons other than that the person at the top makes a decision then imposes their will successfully.
posted by oliverburkeman at 5:52 AM on December 10, 2012

Distance redshift.

(To explain a bit more: I once got into an argument with a fellow who insisted that Relativity was incorrect and was proposing some variant of the Tired Light theory as the true explanation. Whenever I asked what actual evidence he had for this, he repeated "Occam's Razor" like a mantra. When I finally, in frustration, pointed out that Occam's Razor is not actually a physical law or irrefutable principle of logic, but a rule-of-thumb guideline with about as much scientific weight as the "smell test", he looked like I had shot his puppy and had nothing further to say. I'm pretty sure he had thought it *was* a physical law or irrefutable principle of logic.)
posted by kyrademon at 5:59 AM on December 10, 2012 [1 favorite]

We landed on the FUCKING MOON!

Expanding that. Given the precision of the day, Newtonian Mechanics and Law of Gravity were all you need to explain the orbits of the planets, however, as our observations became more precise, it became obvious that Newton wasn't correct here -- in particular, Mercury was the most dramatically error prone planet when attempt to derive a future position from the measured orbit -- in particular, the perihelion seemed to precess without any known force acting on it, moving some 40 seconds of arc every century.

The answer was General Relativity, and the precession was caused by the effects of the Sun's mass and gravity on spacetime. The effect was later seen in other planets, but being further away, you couldn't spot it until our measurement's precision could see it. The Earth, for example, precesses about 4" per century, an order of magnitude less that Mercury.

Without GR, we would not have been able to correctly calculate the orbits needed to get the Apollo spacecraft to the moon.
posted by eriko at 9:06 AM on December 10, 2012 [2 favorites]

benito.strauss: How about "God did it" vs. "Cosmology + Biochemistry". Most biochemical explications of processes strike me as complicated messes looked at more than just superficially.

"God did it" contains a huge set of assumptions that simply aren't stated: there exists a supernaturally powerful being with a creative desire and the intelligence to build..., all the while setting up natural processes that appear to follow laws so as to misguide mathematically and scientifically literate human beings with false-lead data.

Biochemistry's explanations can fit in a single, large paper or textbook chapter.

Messy as it is, it really is the simplest explanation, unless one puts blinders on after noticing the implications of the data. If one doesn't bother to study the data, then one is guessing, not building a hypothesis.
posted by IAmBroom at 10:40 AM on December 10, 2012 [1 favorite]

I still think the biochemists deserve some sort of medal for dealing with messiness.
posted by benito.strauss at 10:50 AM on December 10, 2012 [1 favorite]

eriko:
Expanding that. Given the precision of the day, Newtonian Mechanics and Law of Gravity were all you need to explain the orbits of the planets, however, as our observations became more precise, it became obvious that Newton wasn't correct here -- in particular, Mercury was the most dramatically error prone planet when attempt to derive a future position from the measured orbit -- in particular, the perihelion seemed to precess without any known force acting on it, moving some 40 seconds of arc every century.

Is this true? I mean, the Moon is a pretty big damn target; seems unlikely that Apollo would have missed it because it was off by a few millions of an arcsecond (measure before launch; drift less than Mercury's over 1 day). So, unless some mission-critical parameter needed 10^-9 accuracy, it seems pretty unlikely.
posted by IAmBroom at 10:56 AM on December 10, 2012

Relativity versus Newtonian physics.

To explain this a little: Occam's Razor says that velocities should add. If I'm moving away from you at 20mph, and Abigail is moving away from me at 30 mph, obviously Abigail is moving away from you at 50mph. Special relativity posits (correctly) that this is wrong; she's (infinitesimally) slower than 50mph in relation to you. None of this matters (or is detectable) until you get much closer to light speed, which is why it took centuries for people to notice.
posted by dfan at 11:28 AM on December 10, 2012 [1 favorite]

rogerb et al have covered it thoroughly, but I want a whirl too: a lot of commenters are misunderstanding the nature of Occam's Razor. It's not really a 'law' law - it doesn't count as proof of anything and it doesn't rule anything out. It's just advice on which explanation is the safest bet (all things being equal). Properly understood, the idea of counterexamples to Occam's Razor is meaningless.

F'r example, if you were standing in front of a steep cliff and a paddling pool, and you wanted to know which one to jump in, I (Occam) would tell you that the best idea is probably the paddling pool. If it then turned out that the paddling pool was covering a deep pit lined with pungee sticks, and the cliff just concealed a secret ledge full of partying A-listers, I haven't been proven wrong about anything - it was still the only reasonable advice given our epistemological status at the time.

Occam is not telling you that paddling pools will never ever conceal a deadly spike pit.
posted by forgetful snow at 11:58 AM on December 10, 2012 [2 favorites]

The complete correct understanding of any phenomenon is going to be more complex than the superficial explanation.

Look at Mendelian genetics versus epigenetic events, jumping genes, plasmids, multigene traits, etc. (etc and etc)
posted by dances_with_sneetches at 12:50 PM on December 10, 2012

Is this true? I mean, the Moon is a pretty big damn target; seems unlikely that Apollo would have missed it because it was off by a few millions of an arcsecond (measure before launch; drift less than Mercury's over 1 day).

A small error at the start of a burn can result in a large position error. Two critical burns were the LM Ascent stage to rendezvous with the SM, and the burn that brought them out of lunar orbit into the descent trajectory.
posted by eriko at 1:35 PM on December 10, 2012

"I still think the biochemists deserve some sort of medal for dealing with messiness."

That is really both not as complicated or messy as it looks, at least once you start grokking the underlying principles, and a dramatic oversimplification.
posted by Blasdelb at 1:37 PM on December 10, 2012

Well, I'm coming from a math background, where we do our darnedest to make things not look messy (at the expense of creating cryptic terms and symbols). For example, here is the very cool beast knows as Stokes' Theorem.
posted by benito.strauss at 4:24 PM on December 10, 2012

Frictional static electricity fits the bill here, kind of. In that what we used to think was the cause (rubbing strips ions) has been experimentally proven to not be the cause.

But we don't know yet what the cause actually is. Just that it's more complicated.
posted by klangklangston at 4:51 PM on December 10, 2012

Also, almost all of modern math is in this category!

Proving the following:

x^n=y^n+z^n is only true for integer x,y, and z for n less than 3

required 100 pages, and that doesn't even include a few decades of preliminary papers leading up to it.
posted by kellybird at 5:57 PM on December 10, 2012

Biology, in general.
posted by Good Brain at 8:37 PM on December 10, 2012

Famous, relevant HL Mencken quote: "There is always a well-known solution to every human problem — neat, plausible, and wrong."

http://en.wikiquote.org/wiki/H._L._Mencken
posted by pocketfullofrye at 9:55 PM on December 10, 2012 [1 favorite]

eriko: Is this true? I mean, the Moon is a pretty big damn target; seems unlikely that Apollo would have missed it because it was off by a few millions of an arcsecond (measure before launch; drift less than Mercury's over 1 day).

A small error at the start of a burn can result in a large position error. Two critical burns were the LM Ascent stage to rendezvous with the SM, and the burn that brought them out of lunar orbit into the descent trajectory.

... which doesn't answer my question. Are either of those burns sensitive to 1/1,000,000,000 errors? Sensitive enough to miss being able to land successfully on the Moon? (Make it easier - to land successfully on the Sea of Tranquility?) I highly doubt it.
posted by IAmBroom at 10:21 PM on December 10, 2012

I think the counter example to occam's razor is... occam's razor. Take the argument of the moon landing/hoax. Occam's razor just suggests whichever would have made sense to you regardless. Eg a hoax is a simpler explanation than a moon mission that worked. A moon mission is a simpler explanation than an elaborate conspiracy.

Ie a control group using occam's razor and one that isn't, would I suspect show that occam's razor fails to increase the chances of someone's predictions being accurate (at least at the high end), because garbage in garbage out - you already need to be making the right weightings and observations before you can accurately even identify the simpler thing so you can apply occam's razor in a way that doesn't take it off the rails, and if you're doing that, occam's razor isn't helping.

Kind of like astrology - the best astrologer can make good predictions, but the best astrologer can't make better predictions than the best non-astrologer. Hence, you can say it helps some people get around their own mental obstacles, but at the high end, it's not imparting anything useful. Goldmann Sacks isn't interested in astrology (or occam's razor).
posted by anonymisc at 9:50 PM on December 19, 2012