According to Lawrence Krauss, science cannot not exactly say 'always' or 'never'. We all know that's for sure because most of science theories relies on inductive arguments.

But then he adds: 'What science can tell is what is not true'. I don't seem to understand what he's saying.

As far as I know, science have the same power of making pressumably safe 'always' or 'never' affirmations, as 'xxxxx is untrue'.

Reference: https://www.youtube.com/watch?v=uSwJuOPG4FI#t=5329

Edit: Yes, he says 'Wrong' rather than 'true'. But if something is wrong in science it's because it's not true.

  • Done... better this way.
    – user13664
    Feb 11, 2015 at 20:14

3 Answers 3


Krauss was being a bit sloppy. It's easy to see that science can tell us what is true iff science can tell us what is not true. (You can always rephrase "Is X true?" with "Is not-X untrue?"). Putting this verbal quibble aside, it seems that Krauss was making the (not original) point that it's much easier to falsify claims with "always" or "never" than to verify them. To falsify (or show to be untrue) the claim "All ravens are black" you only need to observe one non-black raven. However, to prove that "All ravens are black" one would have to observe every raven, and then one would somehow have to scientifically verify that there are no ravens that you haven't seen (which looks to be impossible). Karl Popper famously advocated the thought that science is about falsifying hypotheses rather than verifying them. As a last point, it turns out that falsifying a claim scientifically is often not as easy as it appears, as argued by in the Duhem-Quine thesis.


According to Lawrence Krauss, science cannot not exactly say 'always' or 'never'. We all know that's for sure because most of science theories relies on inductive arguments.

If this is all a description of Krauss's position, and it is accurate, then he is wrong.

Induction is an alleged process that starts with observations, derives a theory from them and then proves that theory or makes it more probable by doing more observations. None of the alleged steps in this process can be carried out in reality.

First, you can't start by observing stuff since some things you could observe will be irrelevant. For example, there may be a particular number of atoms in the pen sitting on my desk, but it's not worth observing how many since it has no relevance to anything important.

Second, there is no way that any theory can be derived from observations. For any given set of observations, multiple theories are compatible with it and so no particular theory follows from them. For example, both general relativity and Newton's theory of gravity accurately predicted many of the observations made before the 19th century, but they are not the same theory.

Third, any particular theory is either true or false and we have no way even in principle of showing a theory is true for two reasons. (1) Many theories make predictions about a whole class of situations, e.g. - theories about the behaviour of some particular type of star, say. If you misunderstand those stars, then something you find works for one star or 100 or 1 million might not work for others. (2) All observations use explanations. These include explanations of how the observation itself works, e.g. - explanations of the optics of a telescope. Another important issue is the link between the observation and the theory. Under what conditions does the theory predict X and are those conditions realised in your experiment? As such, whether your interpretation of the observation is correct is dependent on other ideas that may be true or false. If you want to show the theory true you would have to prove all the theories you rely on to understand the observation, and then you would have to prove all the theories those rely on and so on. This process would never end and so you would never create any knowledge.

In reality, scientists create knowledge as discussed by Karl Popper. The scientist notices problems, proposes ideas about how to solve those problems, criticises the proposals until only one is left and it has no remaining problems (the solution), and then moves on to a new problem. The criticism can include experimental testing, but is not limited to such testing, and a theory may not be tested if it is inconsistent or has some other flaw. All knowledge is guesswork that has survived criticism and nothing else. If a theory has survived criticism, it might be true. If it has not been refuted, then the only guess available at that time is that it is false. Any theory, including a theory about what has been observed, can be reconsidered if you can some up with an alternative that has some implication that could be criticised independently of the problem is was intended to solve. See "Logic of Scientific Discovery" by Popper, Chapters 1-5, "Realism and the Aim of Science" by Popper, especially Chapter I. See also "The Fabric of Reality" by David Deutsch Chapters 1,3,7 and "The Beginning of Infinity" by David Deutsch, chapters 1,2,4,13.

But then he adds: 'What science can tell is what is not true'. I don't seem to understand what he's saying.

This is inconsistent with the position given in the first pragraph, but it is the least inaccurate statement he has made. The useful content of a scientific theory is in what it forbids. If you come up with a plan for a bridge, physics may be able to tell you whether it will stand up under the relevant conditions. But physics will not design the bridge for you.


Krauss is rehashing Karl Popper's ideas about empirical falsification. Popper wanted to build science on only deductive reasoning because of the problems associated with inductive reasoning. Popper recognized an asymmetry between verification and falsification; one can falsify a theory with a single counterexample or observation, but one cannot verify a statement in that manner. He felt that science couldn't grant that a theory was probable or correct, only that it hadn't been falsified.

That is the gist of Krauss' remark "what science can tell is what is not true"; science can reject theories that are not true (with deductive reasoning), but if a theory isn't falsified science cannot say whether it is probable or a "good theory" (whatever that might mean).

Krauss says that science cannot say "always" or "never" presumably because that would be a type of inductive reasoning. With such reasoning, the conclusions don't strictly follow from the premises. E.g. if I see 100 white sheep, I can never deductively conclude that all sheep are white or that no sheep are black.

Popper's ideas influenced generations of physicists and scientists, were heavily discussed and, in some quarters, criticized. It's worth remembering that Popper isn't the final word on the philosophy of science.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .