Is the "scientific method" well-defined?

>doesn't require faith based arguments
>"it's self evident"
err...

>err
self evident isn't the right word
just evident

Computer science is a subset of math, you failure.

>Computer science is a subset of math, you failure.
Other way around, you brainlet.

Math is a subset of computer science?

>Math is a subset of computer science?
What else would 'other way around' imply?

Alright, user, thanks.

existentialtype.wordpress.com/2012/08/11/extensionality-intensionality-and-brouwers-dictum/

I'd looked briefly at type theory but something like this hadn't occurred to me. Interesting concept, thanks for that.

You're in luck, there's a whole field dedicated to this.
en.wikipedia.org/wiki/Philosophy_of_science

I studied Physics. It might be other branches of science take different view. However...

>There are a number of huge epistemological issues with the scientific method, which are not usually discussed.
I am disturbed to hear that.

>>if two theories fit the data, which theory should we favor?
Chose the simplest (Occam's Razor)

>>what if one has more parameters, but its equations are more elegant?
Still chose the simplest but be open about two approaches giving same results. Explore more. That is how we proceeded from Newton's classical physics to, say, relativity.

>>what if one fits the data slightly better, but the processes in the theory are not defined in terms of constructs relevant to the field, e.g. fitting a deep neural network to data vs. a more simple but less accurate theory
Not sure what you are saying here. Do you propose to replace a hypothesis with a neural network??

>>when do you decide to reject data as faulty, vs rejecting the theory?
The map must follow the terrain, not the other way round. Beware of rejecting data. At least two Nobel Prices have been awarded where the laureates demonstrated such mistakes had been made.

>>how do you maintain a theoretical superstructure where the theory is so specialized that only a handful of people understand each deeper aspect of it?
That is a major problem. You need to put a lot of energy into moth science and also to train the new generation researcher. And that is hard given the pathetic salaries offered.

>And that's just scratching the surface. Anyone working in science knows that most researchers don't really think about these questions, and its reflected in many atrociously planned experiments.
When I did my PhD we were told in no uncertain terms that we very much HAD to keep these things in mind, otherwise we would be close to fraud.

>Chose the simplest (Occam's Razor)

This is a basic heuristic, but not something that gives a good reason to simply throw away alternate possible explanations. Its also debatable what the 'simplest' even means and how it would affect the likelyhood of a given result being the ultimately better explanation.

>Chose the simplest (Occam's Razor)
what the fuck kind of popsci approach is that?

>Appealing to Occam's razor
You're kind of making his point.

>This is a basic heuristic, but not something that gives a good reason to simply throw away alternate possible explanations.
Said every crackpot theoretician.

>popsci
Please use argumentation and not meme words.

>if two theories fit the data, which theory should we favor?
The simplest to implement, utilize and teach.
>what if one has more parameters, but its equations are more elegant?
Elaborate on what you mean by "elegant", because I don't give a shit about elegance, I care about functionality.
>what if one fits the data slightly better, but the processes in the theory are not defined in terms of constructs relevant to the field, e.g. fitting a deep neural network to data vs. a more simple but less accurate theory
Then you utilize the more accurate theory. This was poorly worded and explained.
>when do you decide to reject data as faulty, vs rejecting the theory?
When there's no plausible explanation for the data to be faulty, yet it still does not fit the theory despite repeated attempts with multiple data sets.
>how do you maintain a theoretical superstructure where the theory is so specialized that only a handful of people understand each deeper aspect of it?
Refine the theory (or try to) until it becomes simple enough that a high schooler can understand it.

t. wagecuck

the scientific method is basically just "test stuff, and observe it, and repeat". it's essentially just saying "empirical observation is the only thing that matters in science".

Is it that simple? Well yes it is, the rest is just a formality to hammer home the empirical elements. The scientific method is actually extremely intuitive to anyone with any semblance of intelligence

"simplest" means "most parsimonious"
>you fookin idiots

tfw when you will never be taught by feynman

monoskop.org/images/7/7e/Feyerabend_Paul_Against_Method.pdf

>monoskop.org/images/7/7e/Feyerabend_Paul_Against_Method.pdf
what's the tl;dr?

There's no unified method, nor should we strive for one. Or so I'm told, I haven't finished.

>if two theories fit the data, which theory should we favor?
Highly unlikely that two theories will have the same robustness, generality, and predictability. It will almost always be obvious which is superior. In any case, if you have two that are exactly the same in all but the core theory itself, than you find which fits into the web of other theories better, or you pick one at random, or you retain both and use them differently, or depending on future developments. But they will not be the same in all but the core theory itself, even then, there will be differences even if only in practicality. So you retain both and use them for different applications. You can describe what classical physics describes, using modern/quantum physics but it's impractical and useless, unless you want more than what classical physics can provide.

>what if one has more parameters, but its equations are more elegant?
Empty question but see above.

>what if one fits the data slightly better, but the processes in the theory are not defined in terms of constructs relevant to the field, e.g. fitting a deep neural network to data vs. a more simple but less accurate theory
Use both, extend the field. There is no reason to not do so, but generally if it's not "defined in terms of the constructs relevant to the field" then it's likely to be too simple and undeveloped to be of much use. You require that culmination once delving into the highly abstract, otherwise, it will, in most cases, just not be a superior theory when compared to one that is defined in those terms. Because of this, it's really not an issue.

>when do you decide to reject data as faulty, vs rejecting the theory?
They're both the same. It's easy to delude if you skew the collection and processing of data, or how it functions in a theory, toward some end. Understand you are working with limited data and a limited model. And always follow the data, not warp it to fit the model. But yeah, how data is skewed and applied should be heavily scrutinised to see if it misrepresents reality.

>how do you maintain a theoretical superstructure where the theory is so specialized that only a handful of people understand each deeper aspect of it?
Systematisation, decentralisation, and automation. T.bh, pre-information age standards of doing this should be dismissed. Even scientific papers, should instead be a "map". But yeah, understanding how the computer you're using right now works requires a lot of different fields and subfields. Basically no one will have a proper grasp on every aspect, it's like this in science too. Automation helps and removal of the "individual" helps.

>Even scientific papers, should instead be a "map".
How do you propose to do this in practice?

I can agree that there is too little re-use of scientific papers and patents which is a larger publishing source than many realise but going from the problem to a solution is not obvious to me. I can agree that it would be enormously beneficial.

Induction is a matter of process of elimination. If you define a set of every possible model of some given data and make them countable, such as ordering them from the shortest model on up, then through process of elimination you'll reach a model that's consistent. As new data comes in you eliminate models that no longer work. If there is a model for the data it will eventually be reached.

In the end, people don't have the time or capacity to go through a thorough process of elimination so instead an approximation is used. With the scientific method we come up with a model, test it for consistency, and if it works it works. If there's more than one model that works then Occam's Razor is a rough approximation of the process of elimination, not to mention having the efficiency benefit of being the shorter of models.

I'd recommend all of you faggots with nice and shiny set in stone ideas about philosophy of science to read Larry Laudan's Science and Relativism if you have any interest in thinking about your activity.
It does a really nice job of raising problems with simplistic views of science.

>faggots
Why the homophobia?

Why do people doing "normal science" think they're some great theoretician or that they have a monopoly on truth? Just go back to looking at test tubes.

>Occam's Razor
literally a meme

It works. What else matters?

>It works.
In what sense?

>It works.
Where? I don't believe I've ever read a publication that went "well I guess we'll just apply Occam's Razor here and carry on lads".

I'm not secure about my sexuality.

Mathematics is birthed from axioms.
As such, science is birthed from observation of natural law.
Therefore, scientific methods first well defined element is "Observation."

>Therefore, scientific methods first well defined element is "Observation."
What's its definition?

Have you guys lived under a rock the last decades?? The rest of the world has experienced scientific progress and greater understanding of how nature works. Occam's Razor is part of that process.

Occam's razor works in what sense?

>Occam's Razor is part of that process.
In what way? I've never seen OR referenced anywhere but popsci material.

A function of your ocular organs.

>A function of your ocular organs.
Blind people can't be scientists?

Broader definitions also apply.
A function of your sensory organs. So long as you can envision basic geometrical forms. dot. line. etc

>Chose the simplest (Occam's Razor)

lmao physics is a joke

>Chose the simplest (Occam's Razor)
>When I did my PhD
Dead giveaway you're lying about having a PhD, either that or some universities are giving them away like candy now.

My supervisor and I used to joke about the razor at the pub back in the day

>For retards like me double blind randomized controlled trial = science.
I know it's not your fault but I hear that a lot from medical people or people who get their info from medical people and it's pretty frustrating.

The scientific method is an umbrella of methods, and while double blinded RCTs are certainly an extremely powerful tool they're often inapplicable or overkill.

>Chose the simplest (Occam's Razor)
the simplest answer is always "God did it"

>It works. What else matters?
I know this is probably bait, but there are a lot of freshmen and freshwomen on Veeky Forums that actually think like this.

That user has it right. Feyerabend made some mistakes with history of science in that book (sometimes bordering on seemingly willfull misrepresentation), but the core of his argument is still an important one - that if you try to formalize a definition of the scientific method you end up making some really important historical inferences invalid while technically allowing some batshit insane inferences as Good Science.

The important note is that he's not trying to say we should toss the scientific method out or that it's useless - just that as a tool it's slightly flawed and we need to be critical of our tools and their shortcomings.

It's a lot like Kuhn's argument that all science has some element of human fallibility and emotion in it somewhere, and while we'll never be rid of it we need to be aware of its influence.

Candy may be closer to the truth than I'd like to believe.

Another user posted awhile back about starting their PhD and being given a lecture on the importance of academic integrity and warned that reporting data that may be unreliable, even accidentally, is the same as plagiarism / intentional false reporting.

And here I was when I started they got the new grad students a pizza and told us to try and come in for 8 hours a day even though no one is keeping track of the hours you put in....

>either that or some universities are giving them away like candy now.
Not the top tier unis, but some of the mid level ones have some disturbing standards for admission judging by the students I've encountered from them.

Most scientists and PhD students I know have no problem accepting Feyerabend and Kuhn's argument, possibly because as soon as you start doing research you understand it has little to do with the simplistic "scientific method" you were told about in middle school and high school. It seems like it's always non-scientists who shout cries of indignation at relativist philosophers of science.

Do you even know what Occam's Razor is? Unless this is really bad bait you should have known there is a second clause.

The most successful rejection of Occam's Razor must be string theory. No ... wait...

>Ockham acknowledges three sources for such grounds (three sources of positive knowledge). As he says in Sent. I, dist. 30, q. 1: “For nothing ought to be posited without a reason given, unless it is self-evident (literally, known through itself) or known by experience or proved by the authority of Sacred Scripture.”

>Occam's Razor
There's a great story about Marcel Schein and Hans Bethe in aip.org/history-programs/niels-bohr-library/oral-histories/5020-4
tl;dr:
>“between your and my explanations is that yours is wrong and all of mine are right. Your single explanation is wrong, and all of my multiple explanations are right.”

That's filled with bullshit inspired by popsci and scifi than real science.

Call the news! Occams Razor has been disproven by anecdotal evidence!

It only takes one counter example to show a thesis wrong, yes

All I'm saying is that you can't just blindly apply it. It's an heuristic, nothing more.