Built on Facts

What is science?

Whew! ScienceBlogs is back up. Mostly. There’s been some kinks, but so far things are mostly working. No LaTeX yet, but I’m agitating for it vigorously. Especially for the comments. The ability to work with equations is something that no self-respecting science blog should be without. I’m hopeful that the process will be simplified soon, and that you can write your own equations as well.

While things are still getting squared away internally, let’s tackle a simple question. What is science?

It’s a question that has come up several times when the various ScienceBloggers chat in our internal discussion forum, mainly with regard to the difference between science and pseudoscience. In one recent instance, I waded right in with my own opinion, which is that science is simply what you do when you test your ideas with experiment.

This would not do, said some people. What do you mean by test? What about sciences where repeatable experiment isn’t really possible, such as climatology, cosmology, or population ecology? What about theoretical ideas that aren’t yet testable, like string theory? What, if any, is the ontological status of falsifiability in all of this? Must the tests be numerical? Is it even possible to define science in an airtight way?

We went back and forth on those and other questions. Progress was limited, not least of the reasons why is that I’m only modestly familiar with the philosophy of science. I’m also strongly of the “shut up and calculate” persuasion and can’t escape thinking the whole thing is a waste of time. Still, it’s an interesting way to waste time. After bantering back and forth here’s what I concluded at least for myself. I don’t know if I convinced anyone else.

Science is the testing of ideas.

That’s all. Every technicality I can think of is avoided so long as the person doing the science is honest. Create fair and objective tests, try not to fool yourself or anyone else, don’t be wedded to your hypothesis, basic things like that. Be dishonest and I doubt there’s a definition in the world that some sufficiently clever pseudoscientist can’t wriggle out of. Test your ideas and be honest about it. That’s about it.


  1. #1 Rhett
    January 12, 2009

    I like to say science is the building of models. What is a model? A model can be an equation, a computer program, a physical model (like globe) or a conceptual model.

    If the model is useful and agrees with observations – that is good. If there comes an observation that disagrees with the model, the model will need to be changed.

    Take a model for gravity – Newton’s Law of gravity. Is this useful? Oh, yes. Is it always true, no. So this model does not agree with all the data, but that is ok.

    I had a video posted somewhere that gave a good example of models in science, but it was on an older blog. I will try to find it.

  2. #2 Coriolis
    January 12, 2009

    Maybe because I’m also a physics grad student but I think the simple definition you propose is perfectly fine. Science is when you make up some model, then compare it to the experimental evidence and reject it if it doesn’t fit. That’s it.

    Not having easily repeatable experiments makes it harder but it doesn’t change the definition of science. Even in physics there are plenty of things you can’t experiment with – i.e. you can’t move a planet around to see how the gravitational force affects it; but that didn’t stop Newton from figuring out gravity (although admittedly there are many other easily repeatable experiments to verify gravity). Alot of astrophysics also has similar issues, it being a bit hard to make a supernova in your lab and all.

    And theoretical ideas that aren’t yet testable are simply hypothesis without confirmation. So long as those proposing them are working towards having them verified experimentally someday, all’s well. Whether string theorists actually do that is another question that I don’t know the answer to.

  3. #3 Bob Sykes
    January 12, 2009

    “Science is the testing of ideas.”

    Ambiguous to say the least. Do thought experiments count? What about arguments in the humanities? Can you test ideas with ideas?

    We should stick to numerical measurement and calculation, and admit that string theory, cosmology, climatology and the social and mental sciences are not science.

  4. #4 Steven
    January 12, 2009

    I think your definition should be slightly expanded to be “Science is the testing of ideas through experiment and observation in a systematic way”

    Some fields aren’t lucky enough to be able to set up nicely controlled experiments, but can still test their ideas based on observations of whatever system they are studying. It just makes drawing conclusions more difficult.

    I don’t think its as simple as labeling the social and metal sciences “not science”. I myself am wary of them, however it is certainly possible for scientists in these fields to do things in controlled ways that allow them to draw strong conclusions, however it is my experience that some do not.

    As long as one follows the scientific method in their investigations it should be considered science, regardless of the field.

  5. #5 Max Fagin
    January 12, 2009

    “Science is the testing of ideas.”

    I agree, with the small addendum that the ideas must be CONFIRMED by their testing. If the idea is discredited by testing and experimentation, then it shouldn’t be considered a part of science anymore.

    Example: Astrology is an idea (The planets are indicative of terrestrial affairs), and it is tested all the time. That would make Astrology scientific by the above definition.

    But if we include the addendum that the ideas must be confirmed by testing, then astrology is put in the proper non-scientific category.

  6. #6 Taxorgian
    January 12, 2009

    I would say something slightly different from Max Fagin above: there’s nothing at all with mentioning Lamarck or phlogiston theory or heat fluid (or astrology) as failed theories. They are good examples of how science is able to dismiss demonstrably erroneous ideas, which is a critical component to the scientific enterprise.

  7. #7 andyb
    January 12, 2009

    I am a physicist too.

    I think climatology (and possibly the other disciplines mentioned) does have testable hypotheses. I would say that if string theory never produces a testable hypothesis, then it is art for art’s sake.

  8. #8 JK
    January 12, 2009

    You could test your idea by honestly asking, “does the bible agree?”

  9. #9 Peter
    January 12, 2009

    “I can’t define it but I know it when I see it!”

  10. #10 Nemo
    January 12, 2009

    Science is the testing of ideas.

    Isn’t coming up with the ideas at least as important as testing them?

    As others have mentioned, your definition is probably too broad. I predicted the sunrise would be beautiful this morning. It was beautiful. Was that science?

    Or: I prayed for happiness. I became happy. Was that science?

    I do not think you can define “science” without the concepts of “precise statements”, “making predictions”, and “external world” (as in what you get from your senses, not just your internal mental states). Your definition only includes the “predictions” part, and that only implicitly.

  11. #11 Matt Springer
    January 12, 2009

    I like the various proposed refinements; the problem is all of them lead into defining the words in the refinements and pretty soon it’s turtles all the way down. You can always fool yourself and finagle your way out of definitions if you try hard enough, because I’ve never yet seen a crackpot give up because he’s presented with a clever enough word game.

    Some notes:
    “Isn’t coming up with the ideas at least as important as testing them?”

    Sure, the ideas have to come from somewhere. It’s not explicit in my wording, but it’s certainly required to have an idea to test in the first place.

    “Or: I prayed for happiness. I became happy. Was that science?”

    Sure. Doesn’t mean it was necessarily supernatural, but the mental effects of prayer are an active area of research. An n=1 sample size isn’t good, but every sample has to start somewhere.

    “I agree, with the small addendum that the ideas must be CONFIRMED by their testing.”

    Yes indeed. A test is intrinsically something that can be passed or failed, and having failed the test an idea ought to be abandoned unless further testing shows the first test to be faulty.

  12. #12 Paul Johnson
    January 12, 2009

    damn the others. i agree with you, even if your definition is blatantly broad. essentially all of psuedoscience falls within poor testing anyways

  13. #13 Aune
    January 13, 2009

    How about:

    “Science is the search for naturalistic explanations of observable data by testable hypotheses”

    The word “naturalistic” is optional, it is there to disqualify hypoteses of the kind “god did it” or “it just magicaly happend”… But those should be disqualified anyway by the requirement for testability of the hypothesis.

  14. #14 Aune
    January 13, 2009

    Also, ID and other crackpottery would be disqualified by the reqirement to explain the data.

    Most crackpots actually use the method of denial in the face of data rather than explanation. All theories that are testable AND pass the tests agains the data should after all be considered valid.

  15. #15 Ian
    January 13, 2009

    ‘You could test your idea by honestly asking, “does the bible agree?”‘

    And if the Bible disagrees, we know that our idea is scientific?!

  16. #16 Vangel
    January 13, 2009

    Arguments in the humanities do not fit the definition of science so we do not concern ourselves with distractions. The statement, “Science is the testing of ideas,” is more than adequate.

  17. #17 Tom
    January 13, 2009

    I’d say rigorous testing of ideas, to close a loophole, but yeah, that’s basically it.

    Do thought experiments count?

    All you can do is test the self-consistency of your model. That’s useful but incomplete. Until you compare it with nature, you’re not done.

  18. #18 Paul Johnson
    January 13, 2009

    Also dont forget that the placebo effect to some might be proof that crackpot medicine works. Just about any ridiculous treatment gives positive results to some extent. Having the data support your hypothesis is vague

  19. #19 yogi-one
    January 14, 2009

    I…can’t escape thinking the whole thing is a waste of time. Still, it’s an interesting way to waste time.”

    And that is exactly why it’s perfect material for launching a huge blogosphere discussion. I think you’re onto something here…

  20. #20 Aune
    January 14, 2009

    Well it seems that there is some form of consensus about what science is. To do science is “to test ideas”.

    The rest of the discution is mainly about the nature of tests. What qualifies as a good test? Obviously “does it feel right?” is a bad test but “does it contradict the data?” is a good one. This is an interesting diskution indeed.

    Another interesting discution would be “what constitutes a valid idea?”. Obvously testable is a requirement, if you cant test it that means it does not to any predictions and is thus not a scientific idea.

  21. #21 Umlud
    January 14, 2009

    Wouldn’t people also think that science is the collection of information created using the scientific method? This is the “standing on the shoulders of giants” part of science.

    I would also generalize your statement more by saying that science is a way of describing the mechanisms of the physical world. I like your final clarifying paragraph, though.

  22. #22 Neuroskeptic
    January 14, 2009

    Funny you should mention that…
    I just wrote about a study in which the authors basically took 70,000+ variables and looked for the one which was affected by a certain drug. They found one which was very reliably altered by that drug, and they found that it was also affected by other drugs of the same type, implying that it could be very useful for discovering new drugs.

    But is that science? There’s no hypothesis. There’s just a huge amount of data being searched for something interesting. On my blog I said that it’s not science.

  23. #23 Jim Thomerson
    January 14, 2009

    Have you read Karl Popper’s “Conjectures and Refutations”? It is available in paperback and I understood much of what he had to say.

    Hypotheses are not confirmed; they are, at best, supported.

    I like to think of scientific theories as not right nor wrong, but rather as more or less general in application. I’m pretty sure my house was built on a flat earth, for example. I suppose having no application would equal wrong.

  24. #24 Bob Sykes
    January 14, 2009

    With respect to Vangel and a few others, my introduction of testing of ideas in the humanities as a counter example is serious. What was Socrates doing if not testing ideas? His tests were verbal, not numerical, but rational. So, if numerical tests are excluded as too restrictive, then the humanities merge with the sciences. Unfortunately,if we go along with Popper we have to exclude from the sciences some disciplines now included. As an applied biologist and believer in natural selection, I find this painful.

  25. #25 Jeremy
    January 15, 2009

    I always like to think of science as a technique to remove bias from observation. Even something as simple as going to grocery store on a full stomach is a form of science because it acknowledges the bias and empty stomach can bestow upon innocent shoppers. We do science all the time we though it is admittedly as less formal version.

  26. #26 Stephanus Rensburg
    January 16, 2009

    Science is a process that at the very least attempts to provide a well reasoned description of an observation.
    The “at the very least” leaves the definition open ended , it could by many more things but must be this at the least.

  27. #27 abb3w
    January 16, 2009

    Even at the most central, I’d add one weasel word: Science is the competitive testing of ideas.

    More formally: Science refers to the process of gathering evidence, forming conjectures about the evidence, developing a formal hypothesis which indicates how the current evidence may be described under the conjecture, competitive testing of all candidate hypotheses under a formal criterion for probable correctness, plus the body of hypotheses testing best thereby and which thereafter are referred to as “Theories”.

    The method of Testing used by Science is dependent on the philosophical assumptions that propositional logic is valid for formal inference, that joint affirmation on the Zermelo-Fraenkel Axioms is self-consistent, and that Reality is Relateable (with at most RE-comlexity) to Evidence.

    The rest? Just details of the math.

    Nemo: Isn’t coming up with the ideas at least as important as testing them?

    Not so much, in that methodologically it doesn’t matter how you get your ideas, only that you have them. It does not matter if you get the idea via drinking excessive coffee, watching an apple fall by moonlight, take LSD and see snakes chasing each others’ tails, or even read answers inscribed on tablets of gold delivered by Seraphim choir. (As an anthropological practice, the last might require co-authorship for publication, but that’s incidental to the methodological requirements.)

  28. #28 Eppendork
    January 19, 2009

    “But is that science? There’s no hypothesis. There’s just a huge amount of data being searched for something interesting. On my blog I said that it’s not science.”

    Neuroskeptic – how do you feel about metagenomics which generate a lot of data or pangenomes in Microbiology? I mean all of these studies could be said not to have a traditional hypothesis – in that they are go see type science. I believe their work is valid science. Is this considered “pseudoscience”?

  29. #29 Andres Villarreal
    June 23, 2009

    I am very sorry to disagree with this definition. What you or I call testing and what a dowser calls testing have nothing in common, and yet all of us think we test our hypothesis honestly.

    The James Randi Educational Foundation (JREF) regularly tests dowsers who honestly think their discipline has passed every necessary test and should not be considered pseudo-scientific.

    Science includes work on every step of the knowledge gathering process: it includes the speculation on possible new hypothesis, thought experiments to seed new and future areas of investigation, definition of hypothesis, designing of experiments and observations, performing experiments and observations, analysis of results, even the teaching of the hypothesis and the methods of science.

    But the main point of dissent is: what is a good-enough test? And the answer cannot be expressed in a few words. The farther down the previous list we are, the harder the tests get. But people will always be involved in deciding what is a good enough standard for that area of knowledge.

New comments have been disabled.