Science and rationalism—a brief epistemological exploration
Is rationalism scientific? Yes. Is science rationalistic? Depends.
Within scientific disciplines, I believe that computer science is more rationalistic than others due to its deductibility (can be proved mathematically).
How about other fields?
Physics? Chemistry? Biology? They rely mostly on empirical results to support their arguments and theories. They can observe. They can experiment. Some of them claim they can prove… but to what extent can they be so confident? Can we really bridge empiricism to rationalism?
Verificationism. Sure, scientific theory should be able to be supported by empirical evidence. But lack of contradicting evidence doesn’t necessarily mean that the theory is true. It just means that the theory isn’t yet made false, even if a hypothesis can be empirically tested and the study has been replicated again and again. Falsifiability is like a time bomb. You don’t know the conditions in which a said theory doesn’t apply. There may be unknown unknowns, like Newton didn’t know his theory didn’t apply in the outer space.
Moreover, some fields can not be experimented, and in some cases, observed. Examples: astronomy, natural history. This is more of a speculation—yet does not receive as much scepticism as social science. Big Bang theory, how dinosaurs were extincted, etc.. cannot be replicated or confirmed given current technology. Scientists in those areas are playing on “what ifs”… trying to explain possible causes without really knowing how cause-effect relationships may have been different in prehistorical times. I don’t find them very different from, say, political analysts trying to explain why Kennedy was assassinated.
I myself am a fallibist. But I won’t go as far as supporting the Münchhausen Trilemma. To science: I have reasonable doubts, but I believe that reasonable (blind) faith is necessary for practicality/pragmatism.
Relativism. A proposition is only true relative to a particular perspective. Like the story of blind men and the elephant. How can scientists be sure that they see the whole ‘truth’, if ‘truth’ is definable at all? Maybe the elephant is too large? Maybe blind men cook the results in order to get their opinions published? Maybe blind men lack a good common measurement (e.g. eyes of the same quality) to give the elephant a ‘fair’ assessment?
Maybe it’s not blind men and the elephant—it’s Plato’s Allegory of the Cave (in modern days, it’s called Matrix the movie)?
The point here is that to scientists need to use judgment in measuring and interpreting the results, and this process relies on the limits and sharpness of their senses, intellect, measurement equipment as well as their experience. Why is light year used to measure distances in space? Why is IQ used to measure intelligence? There are limitations from both cognitive and methodological points of view.
Subjectivism. Whatever methods scientists use to gain confidence in theory from empirical evidence, they “participate” in measuring the results, rather than “observing them objectively”. When you use a ruler to measure the length of something, are you sure you have good eyes? Your visual ability remains constant the whole time? The ruler doesn’t contract or expand while measuring the object? This is particularly present in measuring behaviour of waves and particles at atomic and subatomic scales.
Taboo “rational”, “rationalistic” and “rationalism” or provide very precise definitions.
You seem to mean by rationalism something like “access to absolute truth.” That’s a problem because even mathematics doesn’t have that sort of access. There’s always a chance that a “theorem” has an error in it. Even 2+2=4 might be wrong. So all truth claims are probabilistic, and shouldn’t have a probability of 1 assigned to them. I strongly suggest you read the sequences.
Of course, tabooing this can be useful, to sort out the various things it means in practice.
Sounds like someone needs to delve deeper into how we use statistics to make inferences form the evidence.
I can’t think of a good resource that would tell you only what you seem to want to know—wikipedia and some of the more statistically-minded articles around here might be a good bet.
For a more in-depth treatment, I highly recommend E.T. Jaynes’ Probability Theory. It will take you a long time to read, if you’re anything like me, but no worries, it’s worth it.
Jeez, this is not about statistics. It’s about philosophy. With assumptions about distribution, randomness and other stuffs aside, you still have type I and type II errors. The fact that you see 1,000 out of 1,000 elephants are grey doesn’t mean that all elephants are grey, or elephants are naturally grey.
You are assuming too much, maybe? I also recommend that you read Karl Popper and related articles about falsificationism, verficationism and philosophy of science in general.
No. Statistics and epistemology are deeply related. That’s why Manfred is recommending Jaynes. Or if you prefer, read some of the essays here on Bayesianism.
Philosophy of science has progressed quite a bit since Popper (Jaynes, Ramsey, Quine, Lakatos, and Kuhn would all be relevant individual authors if we’re playing the throw out names game.) I’m not as sympathetic to the strong Bayesian position as many people on LW, but it is a position that is a) most commonly accepted here and b) does a strikingly good job of resolving many of the major phil sci problems (for example it makes the raven paradox almost trivial instead of being a deep issue.) If you are going to take a position other than Bayesianism here, you are going to need to understand Bayesianism and articulate in detail what you think is wrong with it and why your viewpoint is better. And neo-Popperism isn’t going to satisfy people.
Despite the name, “computer science” is not a scientific discipline, but an engineering one. There’s some science involved, but that tends to be empirical testing of what you’ve just built, benchmarking and so on—that is, engineering. The field is an application of mathematics, which is of course pure philosophy rather than a “science” as such
“Computer science” is sometimes called “theoretical computer science”. It is a branch of mathematics that includes such things as computability and algorithmics.
You seem to be talking about computer engineering, which is a quite different discipline.
This argument over definitions comes up frequently, but it doesn’t seem to correspond to any useful distinction in reality. Most of the students who enter computer science programs (including me) end up mainly studying and doing programming (computer engineering), not theoretical computer science. The only reason to maintain a distinction is status—some people think theoretical computer science is higher-status than computer engineering. I don’t, so the whole thing looks really silly to me.
There are three different things in reality that we’re talking about:
Theoretical computer science (P vs. NP)
Computer engineering (hardware; related to electrical engineering)
So-called software engineering (programming)
These are very different things and it’s useful to have different names for them. David said “computer science” but was clearly talking about 2 or 3, not 1, so I corrected him.
There are study programs in some universities that are called CS but really teach programming. They are misnamed. There are many other CS programs that really do teach CS, and very little programming except for the minimal skills needed in algorithmics courses. The CS program I’m in is one (at Hebrew University of Jerusalem) - there’s almost no programming outside electives, and even those are a joke, it’s not a place anyone would go to to study programming.
CS and programming really are different things. Programming is a useful part of CS just it is of any math, physics, or engineering discipline. It’s not a matter of status; its just like not saying “physics” when you mean “engineering”.
(not disagreeing, just thinking aloud)
Some things that fall under Computer Science like Artificial Intelligence seem to be more about theory than learning to programming—whether that counts as “Science” is a question of semantics, but when I was learning about pattern recognition, data mining, natural language processing etc. (and reading/writing research papers on those), it feels a bit of a stretch to say that I was learing programming or software engineering (which would be more about thing like database systems, programming languages, design patterns, compilers, operating systems and the like).
Are “Computer Science” courses more engineering than science? Possibly, but the same could be said of any university course on say Materials Science.
CS courses are CS courses. CS is a (applied math) science but is different from other sciences—just as natural sciences (physics+chem+bio) are different from other sciences, and social sciences are different from other sciences...
CS is not engineering, just like physics is not engineering. CS is not programming. Engineering and programming are useful tools—in CS and in other disciplines too—and so CS students often learn their basics, but they are in no way part of CS.
Where is the “science”? There appears to be engineering and mathematics.
Mathematics is a science—indeed, it is called the Queen of Sciences.
It’s not an empirical science, but that’s not the only kind. (Although today there is a beast known as Empirical Computer Science...)