There are two very peculiar things about the scientific method: first, how late in the development of civilization it became mainstream, and second, that is there is no generally accepted definition of what it actually is, certainly not within the philosophical community.
Hints of the scientific method date back to the astronomy of ancient Babylon (c. 1000BCE), to the philosophy of Thales (624BCE – 546BCE) of Miletus in ancient Greece, and to the experimentation of Frederick II (1194 – 1250) and Roger Bacon (c. 1214–1294) in Medieval Europe. But it was only when Galileo (1564 – 1642) turned his telescope on the heavens in 1609 that it “took”. It was only then that the scientific method was finally on the road to becoming a dominant part of everyday culture. When Kepler (1571 – 1639), and especially Newton (1643 – 1727), consolidated Galileo’s work, there was no turning back. As they say, the rest is history.
There have been various ideas put forth in the past for what science is: induction, verification, falsification, and various other ‘tions’. There have also been monstrosities like methodological naturalism; dogma masquerading as method. But all these have their critics and justly so. In the end, the current consensus in the philosophical community—to the extent there is a consensus—is that the scientific method, as a unified concept, does not exist. Strange as it may seem, there is this general idea that there is no such thing as the scientific method but that different fields of science use different unrelated methods.
The problem is that the scientific method is not what people, especially the philosophical community, expected. The philosophical community has concentrated on things like knowledge, explanations, truth, facts, naturalism, realism, and other such abstruse metaphysical concepts. Yet, they have missed the obvious—that science is something simpler, much simpler, namely model building[1] . This view of science allows us to understand the scientific method in a simple, unified manner valid across the whole spectrum of scientific endeavours and to see the shortcomings of other views of science. This model-building approach also allows us to minimize the metaphysics required. Unfortunately, it can never be completely eliminated.
Model building is not enough to specify the scientific method. You need two additional concepts: observations and parsimony. The models of science are constrained by observation, and judge by their ability to make correct predictions about future observations. Like a model boat, scientific models cannot be proved right or wrong—what sense does it make to claim a model boat is right? But we can certainly say which of two model boats is a more accurate representation of the original. Similarly with scientific models: we can say which of two models is more accurate at making correct predictions for observations. We do not have induction, verification, or falsification, but rather comparison. As Sir Karl Popper (1902 – 1994) pointed out, we have replaced certainty with progress: models are becoming more accurate over time.
Now, observations by themselves are not able to uniquely determine a model. An infinite set of models make the same set of predictions, the same way an infinite number of mathematical curves may be drawn through any finite set of points. But, once it is accepted that science is about model building and making predictions for observables, it becomes clear that adding frills—that don’t change the predictions—is counter productive. Thus we use parsimony or simplicity to make our observationally constrained models unique. It is the combination of simplicity and observations that fully constrain scientific models.
Models do more than allow one to make predictions; they provide structure and meaning to the observations. This is the point missed by the logical positivists who wanted to go straight from the observations to the meaning. Thomas Kuhn (1922 – 1996) pointed out the folly of this with his idea of paradigms: the structures need to give order to any field of endevour. Thus we have the essence of the scientific method: Observational constrained model building, with the meaning in the model.
This is the first post to Quantum Diaries since I have been given a personal blog here. In this set of posts, I will be fleshing out these ideas based on the metaphor of science as model building. I have already put a number of posts on TRIUMF’s Quantum Diaries blog and they have been moved to my new area. I would like to thank Quantum Diaries and TRIUMF for giving me this platform for my views on the nature of science and my distorted sense of humour. Also thanks to J. Gagné for editing the posts and turning my mishmash into something readable.
[1] Either that or they are still annoyed they earned less as philosophy graduate students than the science graduate students
Tags: philosphy of science
Did you notice, that your models (SUSY, Higgs, etc.) any more don’t work? Probably, time of the logic analysis has come: http://arxiv.org/abs/1101.4507
I feel that this is a very lucid explanation of the scientific method. Even though I work in software development, I feel that I use the same three components to produce good software. Unfortunately, my models cannot be specified as succinctly as those in physics, where Newton only had three laws of motion, but I have 60 requirements just to develop 10 web pages.
Indeed, I am always striving to simplify the specification of my software when there are unnecessary “frills” as you called them. In software development, those frills are often misinterpreted and cause people to write useless software. I suppose similar issues could cause physicists to explore problems that are merely imagined, rather than being physical.
As for observations, I can only hope that my clients have a stable concept of what they need. When it is clear that they don’t, I have to call it a null observation, meaning nothing.
Yes, that’s a good point, science is very much about making models of the real world and confronting them with data. I’ve written two related blog posts that are maybe of interest to you: Models and Theories and What is a scientific prediction. Wish you fun with the blogging
)
I tend to disagree with your distinction between theories and models. To my mind all we have are models. Even the best theories are just models for how the universe works.
You’re misrepresenting the philosophers. Many philosophers if science – particularly Bas van Fraasen (one of the most famous ones) – have treated science in terms of model building.
I was probably a bit to harsh on philosophers, but there seems to be no consensus among philosophers or indeed among scientists themselves about what the scientific method actually is. That is very strange. In the early 20th century, there was the received view based on logical positivism. This view has fallen out of favor. A comment in one of the anthologies I was reading said it was recognized that logical positivism was not valid but there was no agreement on what its shortcomings were. I have also had the model building metaphor attacked by physicists who thought their theories were more than mere models.
I am not surprise that some philosophers see science as model building or something similar. There are equally others that don’t. For example, Churchland criticizes van Fraasen’s ideas. And on it goes. I had a paper I submitted to a philosophical journal strongly criticized because I did not pay enough homage to the role of explanation (it had other problems as well). While van Fraasen recognizes some aspects of the model building metaphor (particularly his emphasis on phenomenological adequacy), it is not clear he recognizes all the implications.
But, I do not want to give the idea that the model building metaphor is my invention or unique to me. It has been pieced together from too many sources to list but Poincare, Kuhn and Popper played a large role.
Byron, my take on it is that, in science, we assume that reality is causal, that the fundamental symmetries we observe correctly model real symmetries and that the events we observe are accurately modeled by our observations. The equations aren’t the reality. The conservation symmetries are the reality. Every physical law can be described in the classic form of, “For every action there is an equal and opposite reaction.” I don’t think it makes sense to distinguish between observable (or observed) reality and “real” reality.
Of course we have real theories. Perhaps the most famous “modern” example is quantum electrodynamics. It is not a model of electromagnetism. It is exact. And it is tested experimentally to better than one part per trillion for the electron spin factor.
Quantum electrodynamics is very accurate but not exact. It is a low energy approximation to the standard model of particle physics. Note the word model. By low energy, I mean energies much less than the masses of the W and Z bosons. The standard model is, in turn, believed to be a low energy approximation to a grand unified model or something similar. And that, in turn, a low energy approximation to a model that includes quantum gravity (perhaps string theory). Models all the way down and probably none of them exact. The only one that can be exact is the last one and there is noway to show it is exact and not simply a very good approximation to the next one we have not yet discovered.
To Jim Rohlf,
Sorry – but why can’t a model be exact? Suppose I build a small but otherwise perfect replica of the Statue of Liberty. It’s still proper to call it a model. Anyway, I don’t think anyone here denied we have theories. The claim was simply that theories are models…and then that exact theories are exact
models…
Marcus,
Let us be careful not to hijack Byron’s thread! A model becomes theory when it has sufficient significant predictive power to be verified by experiment. I think Byron’s point is that a theory like QED is doomed to fail at some very short distance scale, not yet probed. Sure it will. We are eventually going to see electroweak unification physics. I would still call it a theory, not a model for the reason I stated. Let me cite another example. Murray Gell-Mann and George Zweig invented a model of quarks (aces) to explain hadrons. This model ultimately grew into the theory of QCD. David Gross in his nobel lecture said:
“Thus, QCD provides the first example of a complete theory, with no adjustable parameters and with no indication within the theory of a distance scale at which it must break down. Indeed, were it not for the electroweak interactions and gravity, we might be satisfied with QCD as it stands. It is the best example we possess of a perfect, complete theory.”
You know they said the same thing about Newtonian mechanics: the perfect complete theory. Some even claimed it was known independent of experience. At least that has not been done with QCD yet.
Scientific Method is, to me, the process of finding the most simple and complete description(s) of observable reality.
This inherently includes an ongoing attempt to disprove all assumptions and falsify discoveries. Every student of science, upon learning a new assumption (like the speed of light is the max. poss. velocity) should challenge the assumption. Regularly challenge assumptions, but keep track of the resilient ones. They must be re-validated whenever new means become available, but until invalidated they are our best tools.
Scientific Method is grounded in assumptions regarding observations; specifically, that observations map, causally, to events. If we do not assume causality, that the effect, an observation, is causally linked to the event observed, then we get lost in solipsism.
• For all observers, reality is entirely defined by observation.
• Appearance is Reality.
• Observation is the final arbiter of reality.