It's wrong, but it still works

January 14, 2010 12:11 PM

Can you identify some scientific theories or models that were wrong, yet did a reasonable job of predicting real-world events. For example, the Ptolemaic Geocentric Model of the solar system predicted the paths of planets in the sky relatively accurately.

Another example I can think of is the Ultraviolet Catastrophe which the Rayleigh-Jeans law accurately calculated black-body radiation at lower frequencies of the electromagnetic spectrum, but failed for higher frequencies.

Those examples were essentially attempts at mathematical models of the real world. I'd be in interested in non-mathematical models as well.

This brings up philosophical questions about the relationship between a model of the real world and the real world, and if you have any thoughts on that, I wouldn't mind some discussion on that as well.

posted by ShooBoo to science & nature (21 answers total)

Newtonian gravity.
posted by mr_roboto at 12:17 PM on January 14, 2010

The Bohr model of the atom.
posted by mr_roboto at 12:19 PM on January 14, 2010 [1 favorite]

Not just gravity, but Newtonian mechanics in general.

"Fluid" models of heat, electricity, etc.
posted by hattifattener at 12:23 PM on January 14, 2010

Not merely "Newtonian gravity"; Newtonian mechanics in general.

Also, as I understand it, either quantum mechanics or general relativity or both, although it's not currently known which one (or both) is wrong. They apparently contradict each other in certain ways, so at least one of them is incorrect. I don't have enough of an understanding of them to say how they contradict each other.
posted by Flunkie at 12:23 PM on January 14, 2010

I clicked "Post", the screen came up, and I thought, "I didn't post anything about fluid models!"
posted by Flunkie at 12:25 PM on January 14, 2010

I think that electrical current was first thought to be caused by the movement of positively charged particles, whereas in reality it's (typically) caused by the movement of electrons, which are negatively charged.
posted by Flunkie at 12:31 PM on January 14, 2010

In a more general sense, lots of real world events can be predicted perfectly well if you refuse to believe in various mathematics which would be involved in their more accurate prediction, such as the existence of irrational numbers. And there have been points in the history of science when the consensus about irrational numbers was "burn the witch".

Similarly, Euclidean geometry's parallel postulate was, for much of history, thought to be a truth about geometry in general, which is now known to be untrue.
posted by Flunkie at 12:51 PM on January 14, 2010

Titus-Bode law for planetary orbits. Kepler's nested platonic solids for planetary orbits.
Phlogiston.
Dirac's infinite electron sea with "holes".
The ether.
Parity conservation.
posted by Schmucko at 1:01 PM on January 14, 2010

Phlogiston Theory is a great example of a theory that was neat, plausible, and wrong.
posted by borkencode at 1:04 PM on January 14, 2010

Can you identify some scientific theories or models that were wrong, yet did a reasonable job of predicting real-world events.

I think that it would be difficult to find a scientific theory that does not do a reasonable job of predicting real-world events, because many such theories are constructed in a post-hoc manner: people witness events, then construct a theory to "explain" them. A lot of discredited medical theories are examples of this. For example, miasmatism held that noxious fumes from decaying organic matter, foul-smelling water, and dead bodies caused disease. While it was wrong about the exact cause (its not the smells that kill, but the microbial vectors that are often associated with these pheonomena), it nevertheless led people to conduct some effective sanitary reforms.
posted by googly at 1:14 PM on January 14, 2010

The theory of luminiferous ether was how most physicists thought the universe worked in the late 1800's. It's the electro-magnetic version of Newtonian Mechanics, and consistent with Maxwell's equations (at least, if you don't do big interferometry experiments).
posted by bonehead at 1:22 PM on January 14, 2010

Lamarckian inheritance is a non-mathematical one.
posted by pete_22 at 1:49 PM on January 14, 2010

Indeed, linked from bonehead's link: wikipedia's page of superceded scientific theories.
posted by mjg123 at 1:54 PM on January 14, 2010

Strictly speaking, all our theories are wrong. (What does it mean for a theory to be "right"?) As we learn more, we make better and better (more accurate) models. But even now we can only predict things "relatively accurately".
posted by phliar at 2:20 PM on January 14, 2010

superseded, isn't it?

What does it mean for a theory to be "right"?

This is the philosohpical point that the last line is asking, and it's non-trivial. There is more to it than it's-all-just-approximations, job-done. And it it is interesting to consider the notion of an absolutely correct model. Of course, all models are open to being disproven by counterexample. But that doesn't mean that there is a counter-example to each model, does it? Is it really impossible to construct a model which is correct to an arbitrary degree of accuracy? I think that model would be "right", even though we may not be able to know it as such.
posted by mjg123 at 3:03 PM on January 14, 2010

Thermodynamics is _full_ of "close enough" approximations derived for the sake of making steam engines work, that do a good job of mirroring what we later figured out with quantum mechanics and whatnot.
posted by paanta at 4:03 PM on January 14, 2010

Is it really impossible to construct a model which is correct to an arbitrary degree of accuracy?

Probably.
posted by flabdablet at 6:57 PM on January 14, 2010

What do you mean by "wrong"? Any theory is "right" for the data it fits; no one knows what's fundamental for sure yet. In 500 years people may laugh (respectfully) at our relativity and quantum mechanics the way we laugh at spontaneous generation now: the theory fit the (limited) observed data and was useful for people of the era. Get on Wikipedia and look up the history of science for a near-limitless number of examples.
posted by sninctown at 7:10 PM on January 14, 2010

What do you mean by "wrong"?

Provably so, presumably. We can't prove evolution, plate tectonics, QCD or GR wrong at the moment, so, with that proviso, they're accepted as "correct". It's probably better to think of them as the best we can do right now. We can, however, definitely rule out the wrong models with contradictory evidence.
posted by bonehead at 7:34 PM on January 14, 2010

Asimov's short essay The Relativity of Wrong might be of interest to you. As sninctown says, there are different values of 'wrong'. For example, I would characterise Newtonian physics as incomplete rather than outright wrong, due to the fact that the more complete relativistic physics generally approximates to the Newtonian case in human-scale interactions, and Newtonian physics is still in use. This is not the case for something like phlogiston theory.
posted by Jakey at 2:22 AM on January 15, 2010

Newtonian/classical mechanics in general, as pointed out in the initial limerick of David Morin's Introduction to Classical Mechanics:

There once was a classical theory,
Of which quantum disciples were leery.
They said, “Why spend so long
On a theory that’s wrong?”
Well, it works for your everyday query!


To extend what mr_roboto said, the Bohr model works well for predicting the spectra of atoms with only one electron, and for explaining simple chemical bonding qualitatively. But it does not explain the spectra of atoms with multiple electrons well, and better models of chemical bonding depend upon the idea of atomic orbitals. Additionally, the very assumptions of the Bohr model are absurd, as a moving charge should continuously radiate energy, and so the electrons should eventually drop down into the nucleus, rather than orbiting it at a constant radius.

As George Box said, "Essentially, all models are wrong, but some are useful".
posted by James Scott-Brown at 6:58 AM on January 15, 2010 [1 favorite]