Warning: This blog is written for a rational audience that likes to have fun wrestling with unique or controversial points of view. It is written in a style that can easily be confused as advocacy for one sort of unpleasantness or another. It is not intended to change anyone's beliefs or actions. If you quote from this post or link to it, which you are welcome to do, please take responsibility for whatever happens if you mismatch the audience and the content.


Let's get this out of the way first...

In the realm of science, a theory is an idea that is so strongly supported by data and prediction that it might as well be called a fact. But in common conversation among non-scientists, "theory" means almost the opposite. To the non-scientist, calling something a theory means you don't have enough data to confirm it.

I'll be talking about the scientific definition of a theory in this post. And I have one question that I have seen asked many times (unsuccessfully) on the Internet: How often are scientific theories overturned in favor of new and better theories?

I assume Creationists are the ones usually asking the question. And if history is our guide, the comments on this blog will focus on that one area and destroy the value of this blog post. I'm hoping we can ignore evolution and creationism and climate change for one day and just ask the following question: How often does a scientific theory get discarded or replaced with a better one?

I don't think there's a good answer to my question, for lots of reasons.

For starters, I doubt anyone has been keeping a stat on overturned theories. And I don't think it's fair to compare theories from a hundred years ago to theories created today because our ability to collect confirming data today is better than it used to be. I would expect that a theory created recently would be more likely to stand than one created last century.

Still, it has always been true that the stuff we believe today looks way smarter than the dumbass things our grandparents believed. Why wouldn't that be just as true for our future great-grandkids looking back at our primitive beliefs? Some humility is always called for.

Science requires credibility to be useful. And that's a problem. The non-scientist asks "What is your success rate?" and gets no useful answer. Scientists, as it turns out, are terrible at marketing. About 90% of my exposure to science involves media reports that get correlation and causation confused. As a result of that exposure, the more I hear about science, the less credible it feels.

To make matters worse, I have a jaded Dilbert mindset about every industry. Unless science is different from all other human endeavors, 10% of scientists are honest and amazing and doing important science while the other 90% are like Dilbert's worthless co-workers. So when I hear that 98% of scientists are on the same side of an issue, I wonder how many unreliable people you have to add together to get an opinion you can trust.

I don't think I'm alone in my views. I'll bet that if you did a poll you'd find that scientists believe theories are fairly dependable and useful whereas the average non-scientist believes that everything we think we know today eventually gets disproved. Part of the problem is that scientists are looking at utility and non-scientists are looking at "truth" which is a fuzzy and overrated concept.

In every other field, your track record of success determines your credibility. Personally, I have no idea what the track record of science is. All I know are anecdotes about wonderful successes and notable mistakes. I don't even have a general sense of whether scientific theories have usually held up over time or not.

So when scientists say a particular theory is backed by the majority of scientists, how much weight should I put on that? Is that a situation in which I can depend on the scientists to be right 95% of the time or 5%? What's the track record?

Note to the Bearded Taint's Worshippers: Evolution is a scientific fact. Climate change is a scientific fact. When you quote me out of context - and you will - this is the paragraph you want to leave out to justify your confused outrage.


Scott Adams

Co-founder of CalendarTree.com

Author of a book on success. (Makes a good graduation gift, btw)


Rank Up Rank Down Votes:  +75
  • Print
  • Share


Sort By:
Apr 14, 2014
I also think that the more fundamental question you're trying to answer is, "How often do we get it totally wrong?" When the world universally accepts a theory as truth (or as close to the truth as scientists would allow themselves to be), how often does that get overturned? We universally accept that gravity comes from mass, but someday we may learn that it's created by aliens hiding at the centre of earth and other celestial bodies. That would be an amazing discovery, and it would immediately render all physics textbooks outdated.

That's the question I would like answered - how often have we discovered we had it totally wrong? All the other theories that are known only to a few select scientists or that is still super theoretical no one really cares about.
Apr 14, 2014
Evolution is a scientific fact. Climate change is a scientific fact. That cinnamon rolls are ruined by microwaves is a scientific fact.

My wife and I were at a B&B near Yosemite recently and on Sat morning they served an excellent cinnamon roll topped with white frosting. It was fresh, and had had time to cool after being baked, so it was flaky, fluffy and perfect. The next morning I mentioned to the hostess with a mock pout that I didn't see any cinnamon rolls. With a wink, she said there was one left from yesterday with my name on it. In anticipation, I poured myself a cup of black coffee, and contentedly looked out at the beautiful scenery.

A minute late she served the roll saying "I heated it up for you!" and left. My jaw dropped. Steam was rising from the defensless little roll. Its white frosting top looked to have been replaced by a thin layer of mucous.

My brow furrowed, my eyes narrowed. I stuck my index finger down into the now-rubbery center swirl. Just as I had feared. It was hot. My blood pressure shot up. If I had been Bruce Banner, I would have transformed and flipped the table over with a roar. Knowing it was the last surviving roll from the day before made it worse, as if the last cinnamon tree in the world had succombed to a virus.

It took a few moments to recover, but I remembered seeing a diner up the road surrounded by pickup trucks, so we went there and had some servicable apple pie. But the mental trauma never left me for days, as you can probably tell.

Apr 14, 2014
Really liked this post, but disagree on one point:

"I don't think it's fair to compare theories from a hundred years ago to theories created today because our ability to collect confirming data today is better than it used to be."

You might very well have said this 100 years ago about the theories from 200 years ago. And someone may very well say this 100 years from now about today's theories. Our ability to collect confirming data will always be better than previously, and it will absolutely have an impact on our ability to create, evaluate, confirm, or disprove theories.

So I disagree: it's perfectly fair to compare old theories with today's.
0 Rank Up Rank Down
Apr 14, 2014
Scott, can I suggest adding a delta factor to the measurement so we look at not only how often theories change, but how different the new theory is from what it replaced. If we apply an arbitrary percentage difference to each we would get an adjusted total that also reflects how wrong the previous theories were.

An example using the controversial but already stated theory of evolution, one preceding theory it supplanted was the !$%*!$%*!$%*! of the species which had the concept of mutation of one species leading to a new one. Evolution added a concept of natural selection to existing work. For this example the delta factor might be anywhere from .3 to .7, but the concept is that it is less than one.

An example from a search on "new theory of" gives this example http://www.sciencedaily.com/releases/2013/10/131010205325.htm
Where an existing theory of individual synaptic plasticity is expanded with a theory of structural plasticity in the brain. While this theory replaces previous thinking it adds maybe 10% to an area of study that was already acknowledged to be incomplete.
Apr 14, 2014
On a side note, how many god have been replaced by better gods?
Apr 14, 2014
That's an easy question: Theories are proven "wrong" 100 percent of the time. Real theories that allow us to predict become wrong when we find ways to measure more accurately and discover deviations from the predictions that then lead to better theories.

Since our observations can never be completely accurate, we can always get better (i.e. we will never find "the answer.") And i think that older theories would stand longer than the newer ones that replace them, since technology and our ability to measure outcomes increases at a non-linear rate.
+13 Rank Up Rank Down
Apr 14, 2014
Part of the problem is that there isn't a consistent standard of proof for a theory between disciplines. Physics, astronomy, and mathematics have remarkably high (5 sigma ) standards of proof for a hypothesis before it ever gets to theory, while other disciplines are more lax, where even "consensus" is good enough to call something a theory.

It really is a disservice to physics to refer to something from biology or earth science as a "theory" with the same tone that you refer to relativity or gravity.
Apr 14, 2014
I often work with expert witnesses: engineers, doctors, accident re-constructionists, and many others. I've found that no matter what your position is in a case, you can find an expert willing to support it and call the opposing side's expert a charlatan. Scientists are supporting an agenda with their research, and it is rarely solely for "the good of humanity." A couple of books I read recently: "Age of Wonder" by Richard Homes and "The Snake Charmer" by Jamie James illustrate well the infighting among scientific communities, the pursuit of personal glory, how science can be hostage to its funding sources, how politics affects science, etc. The first book covers the 18th century and the second ends 2001, and it is interesting to see how little the humans behind the science have changed.
Apr 14, 2014
Science has not inherent track record, and cannot, because its purpose it to create the knowledge necessary to move to an ever-closer explanatory model of reality. Just about all scientific theories are thrown out in favor of new ones eventually, but even that's not quite right - Einstein didn't overturn Newton, he merely generalized Newton's theory. Unifying gravity and quantum mechanics will involve a further generalization, of which what we think of as gravity and quantum mechanics naturally fall out. It's a process of moving ever closer to a goal, a goal that we may never arrive at. There can no data to track such progress, even in principle.
Apr 14, 2014
In the book, "The Structure of Scientific Revolutions", Thomas Kuhn shows how changes in scientific thought often evolve over a generation of a gradual paradigm shift. Usually not in a sudden switch. More & more anomalies are found that need to be fitted into a theory or explained away. Phlogiston explained a host of physical phenomena, until it could no longer. Scientists cling to their old beliefs, and invariably do not follow the "science", if it conflicts with their ingrained view, just like non-scientists.

An example is the person who posited that a bacterium causes stomach ulcers, not stress. He was universally derided for this hypothesis, until he proved it on himself.
Apr 14, 2014
There was a story I heard recently on NPR that related an anecdote about people at a county fair trying to guess the weight of an ox. The true weight of the ox was, say, 1200 lbs. Hundreds of people would guess and no one was terribly close, with individual guesses ranging from 300 lbs to 3000 lbs. However, if you took the averages of all guesses, the mean was 1195 lbs, way closer than any individual guess.

Tying this story back to your post, I think what makes science different is peer review (obviously the mechanism is different than in the ox example, but the underlying point is strength in numbers). Most other industries don't have peer review to the extent that science has, and in fact, much of the information is kept hidden from the public due to IP considerations. Peer review is a pretty solid way of vetting the work of flawed individuals; it's difficult to duplicate results based on erroneous data or flawed methodology. Hundreds of people/experiments usually aren't going to arrive at an identical, but incorrect, answer consistently.

Also, as others have mentioned, tweaks to existing theories (such as what may be necessary to the standard model due to the discovery of exotic hadrons) shouldn't be classified as incorrect; there wasn't data previously to support such a claim. Scientists will tell you anything that goes against current theories is the most exciting part of science, and proof that the scientific method is working, although this may just be a cop out on their part to avoid being wrong. Then there are times when science is blatantly wrong, and unless it's a rare case of scientific fraud, the error stems from some fundamental misconception about an aspect of a study, usually buoyed by universal shortcomings in human reasoning (confirmation bias, etc.). It's in these cases where you can really say a theory is incorrect. I'm not sure of any scientific theories that follow this example, but recently it has been suggested that mammograms should not be administered every year, which goes against previous peer-reviewed and approved recommendations.
+16 Rank Up Rank Down
Apr 14, 2014
Science gets too much scrutiny, probably because scientists have admitted they were wrong so many times. What about… say… humanities. We live in a world where 98% of English literature experts will agree that a given classic is a great, entertaining work of fiction (and you basically have to agree with this to become an "English literature expert") and people buy into this, even though all of us who read said work find it to be a boring piece of crap. I think you should write a post about that. Why do we trust these people to tell us what is "good" and what isn't?

Sorry if this is only vaguely relevant.
Apr 14, 2014
Ludwig's Speculation:
New theories replace old theories when the old guard defenders of the status quo die off, unless an elite scientist is the one with the new theory.

Peer review supports this distortion, because the reviewers have to be able to get papers accepted, too. Simply Machiavellian power politics. This is why the journal of last resort (Nature) exists, and this is why Einstein had to publish the special theory of relativity there.
Apr 14, 2014
Science seems to have a few different facets. I think the core motive to do research and represent it fairly is reliable under normal !$%*!$%*!$%*!$%*!$%*! scientists are people, and they often have to spin their research as generating some benefit to the public so as to justify their research. I believe scientists are often extremely dishonest about this, and it is my impression that scientist are pretty shameless about this financial motive, because they perceive that research is an inherently noble cause and Philistines need to be manipulated for the greater good. I have seen numerous interviews where a scientist almost brags about the preposterous scenarios they have to concoct to "win" research dollars.

Also, scientists have personal beliefs and sometimes they're tempted to alter and misrepresent their research to distribute their beliefs. I think that's an extremely small percentage of scientists and science research due partly to the hypothetical peer review process. But I think passionate scientists are inherently less trustworthy, and particular people with strong religious, political, economic, and social beliefs.

Then, you have the science that scares people, which the media loves to flail about as irresponsibly as possible, because it's attention-getting.

Where you have a combination of economic, activism, and media motives to misrepresent scientific research, I seem to see a much higher level of scientific fraud and imo the field deserves a fair bit of skepticism.
Apr 14, 2014
I think the "track record" of Science cannot be measured by "how right the theory was" but rather by what useful applications the theory yielded. Generally speaking, science itself is only useful to the extent that it can be applied to improve our daily lives. I assume advanced discoveries in physics were quite useful in inventing the airplane, or launching satellites for example. The fact that those things could be successfully done proves that the science behind them was essentially accurate.

I would say the same thing about medical science. Curing someone with leeches or by bleeding was "right" only to the extent that those things actually did improve someone's health. Granted we're looking back with hindsight here, but it would appear that they didn't. Curing someone with modern medicine seems to be right because most modern medicine does seem to produce tangible benefits.

It also depends on how broadly you define "science." Is engineering a science? If so, I would suggest humans mastered quite a bit of that pretty early. The pyramids are still standing, after all. The science behind building massive structures hasn't been overturned, even it has been steadily improved upon.

Just a few thoughts. Also climate change is a giant conspiracy perpetuated by socialists who are attempting to bring about the New World Order via fraud and deception. (If Scott can ruin an otherwise intelligent post by trolling people at the end, so can I!)
Apr 14, 2014
I think you may be asking the wrong question.

The strength of a theory should not be judged on the percentage of scientists who agree with or accept it. The strength of theory should be judged on its ability to explain and predict experimental outcomes and to some extent broader outcomes.

There are quite a few theories that had unanimous support that have since been discredited. Conversely, Einsteins theory of relativity was initially met with a lot of skepticism by other scientists, and it was only 50 years of observations and experimentation that brought it to the current level of acceptance. Simlilarly, Ptolemy's earth-centered universe was met with universal acceptance, while Copernicus' sun centered theory was ignored for 100 years until Galileo was able to test it with telescope observations.

I am not going to go down the roads you said were off limits. But can you think of a widely agreed upon theory nowadays that has had a very poor track record of predicting the future? If so, I would ask why we put the "anti-science" label on those who question the theory, yet give all kinds of respect to those who would like to settle science by a show of hands.
Apr 14, 2014
What we called science today used to be called 'natural philosophy', because it was considered one field in the pursuit of knowledge and understanding, or 'science'.

In modern times, with the help of successful scientists in the field of natural philosopy, that school came to dominate what we understand as science. The now so-called 'soft science' disciplines like philosophy, meta-physics, psychology, sociology, history, etc. Are scarcely considered 'science' at all, while mathematics is often considered a tool for natural science, rather than a science in itself. (It doesn't help that pure mathematicians often look insane to the rest of us).

In most cases, scientific theories based on good observational data are never overturned, but they are often superseded by newer theories that are more generally applicable. Newton's theories concerning light, motion, and gravity are still applicable for most things we interact with, but Einstein's theories of relativity were found to be more generally applicable to the universe. It isn't that general relativity overturned Newton's theories - Einstein's equations reduce to Newton's for most things you can directly observe yourself - but they cover many things outside Newton's experience.

Other theories - those based on indirect observation and extrapolation - are less secure. The standard model of quantum physics depends on events that don't normally occur in observable ways. So it is a little like trying to figure out what an animal looks like by studying its poop. You can learn a lot from it, but there is a chance you could be way off.
Apr 14, 2014
[Based on what data? -- Scott]

Based on what I know of history. How big is the universe? Science has been changing the answer to that a number of times. How will the world end? Ditto. How to build a better computer? Science has been pretty good at giving us the answer to that one. How to get into space? Track records not so good on this one (our early efforts didn't work out so well), but overall its been good at telling us how to do stuff better.
+23 Rank Up Rank Down
Apr 14, 2014
And you've got to take into account the cases where a theory wasn't total rubbish, just slightly not right.

Take something less controversial than the average "science controversy": Gravity.

The theory of gravity, F = G*(m1/m2)/r^2, isn't right for very small values of m1, m2, and r. I've seen conjecture that it's wrong for large values as well. The value of G has been changed repeatedly.

That said, for everyday "how fast is this rock going to fall when I drop it" math, it's perfectly fine. For engineering purposes, it's a fine theory and the math always works. For astrophysics or particle physics, it's not good enough. So is the theory wrong? How do you want to count that one?

0 Rank Up Rank Down
Apr 14, 2014
The most thorough examination of the nature of scientific theories is the Structure of Scientific Revolutions by Thomas Kuhn.


The Wikipedia article is not a model of clarity, but the book itself is definitely worth a read.
Get the new Dilbert app!
Old Dilbert Blog