http://www.guardian.co.uk/science/blog/2009/mar/03/science-definition-council-francis-bacon

Edit: leaving aside politics and religion as much as possible, this looked like something interesting to discuss.

Original image removed by the Sheriff.

It's a brief, concise definition of what "science" is. However, it doesn't necessarily have anything to say about how science is conducted in the real world.

For example, there's a lot of scientific shenanigans that go on all the time in the pharmaceutical industry surrounding how they go about getting a drug approved for sale. The process may adhere closely to this definition of science even though the actual result--say, a drug that doesn't actually work or a drug that actually causes a lot of harm to people--shouldn't really be possible since everyone was being "scientific" about deciding whether to approve the drug, right? But it happens.

Or scientists are human beings, just like anyone else. There are biases, assumptions, and motivations that can color their judgement and influence their actions just like in any other endeavor even though they are "good" scientists. E.g. what about the hidden problem of topics that scientists avoid researching because they are controversial and could kill their careers or which are not "hot" topics, like say string theory in physics?

Anyway, a definition is nice and all, but it takes a lot more discernment into the whole process and world of "science" to really be critical consumers of it. And just as important as knowing what information to throw out as being being "not scientific" is also noticing what is missing or not being studied and figuring out why that is.

I suppose a more complete definition of "science" would include the word's use as various parts of speech: noun, verb, etcetera.

For me science is defined by the scientific method, which is more specific than that Science council suggestion. They say "a systematic methodology based on evidence", but that doesn't go far enough. You can think you are basing your conclusions on evidence, when you are actually suffering from confirmation bias. The key to the scientific method is not the confirmation of hypotheses but their elimination. You develop two or more hypotheses which differ in some significant way, and you use that difference to design an experiment which distinguishes between them.

Do go on.

An online search for "scientific method" will turn up plenty of articles. Most of them talk about forming a single hypothesis, rather than multiple hypotheses, but I think they are taking for granted the "null hypothesis" which is always present. Ie that your main hypothesis is wrong. They agree that in the scientific method, experiments are more about falsifying hypothesis than about confirming them.

It's fairly easy to imagine and find evidence that supports your hypothesis. It's harder to imagine and find evidence that refutes it, and the result is correspondingly more powerful when you do that.

While I agree with your statement in general, the reality is that you can't conduct science this way in every field. Physics, microbiology, sure. But what about humans? You can't cage humans and control every little bit about their environments except for the single thing being studied. Clinical trials are good, but there are many things which can't be practically studied this way because of the cost, the follow up time involved, or even the ethics.

Therefore, we must rely on scientific methods that fall short of the gold standard of the "experiment" (or in the case of people, randomized, double-blind, placebo-controlled clinical trials). So, the definition of science needs to be flexible enough to include these other methods and I think the phrase "systematic methodology based on evidence" is appropriate.

Besides, science also includes the process of building up observations, knowledge, and understanding that leads to more formal experiments. That process may be systematic and involve evidence but hasn't yet reached the stage of any sort of experiment. Cosmology and theoretical physics is full of concepts that we haven't been able to formally test yet, but I think most folks would still include that labor under the umbrella of "science".

Telltale signs of what is not science, taken from Massimo Pigliucci's book Nonsense on Stilts, citing John Casti's book Paradigm Lost:
1. anarchronistic thinking

2. glorification of mysteries

3. appeals to myths

4. cavalier approach to evidence

5. appeal to irrefutable hypotheses

6. emphasis on probably spurious similarities

7. explanation by scenario or story-telling

8. literary rather than empirically-based interpretations

9. extreme resistance to revising one's position

10. a tendency to shift the burden of proof [prove me wrong]

11. sympathy for a theory just because it is new or daring

Been getting educated a bit. Contemplating the so-called historical sciences, whose data lies almost exclusively in the past, like archeology, paleontology, human evolutionary biology, astronomy, etcetera, in which the connventional "experiment" model does not work. These contrast with the so-called hard sciences whose nature permits such experiments, like physics, chemistry, etcetera.

Then there are the "soft" sciences like human psychology and biology which, by their nature, allow experiments but whose data can only be evaluated statistically. In other words, those sciences that can predict with great accuracy things about large groups of people, death and disease rates for example, but cannot predict with any accuracy what will befall any one individual.

Yep, being taken to school and reminded there are some scary smart people out there!

Just a quibble, but I don't think we should accept a blanket statement that biology or medicine can't predict what will happen to an individual. We do understand the mechanics about enough processes to know what will happen to certain people, such as those people with various genetic defects, such as for Huntington's disease.

Actually, the fact is that biology/medicine encompasses elements of "hard" science and "soft" science depending on the particular field or the question, so it's not so easy to give it a single label. Look at Watson and Crick, the discoverers of the structure of DNA. Using x-ray crystallography to tease out its structure is pretty "hard science".

I disagree that biology or medicine can predict what will happen to an individual. The problem lies in the fact that people, or biological systems in general, are not a uniform media. You can predict odds based on a population meeting X, Y, and Z criteria, but that's the limit of the science. The anatomy/chemistry/psychology of a single person is unique to that person, and a sample size of 1 is near useless without a complete understanding of the mechanisms (which we don't have).

The problem of variability also extends into pure hard sciences like metallurgy. A certain material, with a certain composition, unit cell structure, grain size and orientation, heat treat, forming process, etc. will still have variability in its properties. You can see the differences in A-basis (99% values with 95% confidence are greater then stated value), B-basis (90% values with 95% confidence are greater then stated value), and S-basis (industry specific) strengths for a particular material.

Somebody with medical credentials should step in to lead us through this maze. However, I seem to have accumulated information that the progress of any one person's disease is highly variable and unpredictable.



First, let recall that it was Rosalind Franklin who was the x-ray crystallographer, not Watson or Crick. It was only the no-postumous-Nobel rule that deprived her of that recognition for her DNA work. She went on to a distinguished career perhaps capped by her on Nobel prize for other contributions to science.

Second, you are right that there are aspects of science, hard, soft, or otherwise that permeate or invade their theoretical opposites.

One thing that tends to distinguish science is that it progresses over time, that is, it changes and tends to become more reliable and useful - even if it must do so by utterly disproving earlier things the same science postulated as true.

If you don't have falsifiable theories, you don't have science. "Social science" isn't a science. "Economic science" isn't a science. As a rule of thumb, if a discipline has "science" in its name, it's not a science.

Not everything has to be a science.

Because they make predictions that can be disproved. They can be wrong.

I am currently reading a bit about the development of "science" and find it is an evolving concept. What things are included as "sciences" is therefore somewhat in flux as well.

There are some areas of study not readily amenable to experiment because they are, at least for now, too complex or ethically out of bounds. Those things tend to be studied instead by many detailed observations generating data which are checked for statistical reliability. If deemed reliable, then theories can be propounded and old and new data can be examined to see if they are consistent with the theories.

Perhaps your definition of "experiment" is broader than mine, but to me, an experiment is a specific kind of activity involving tightly controlled conditions where only very specific variables can be adjusted. However, as I mentioned, you generally can't do such experiments on people. The core of research on human health involve other kinds of study designs (look at the whole field of epidemiology), and the analysis of those results also includes hypothesis testing. My point is not about the necessity of having a falsifiable hypothesis, but on how to go about obtaining the evidence to disprove these hypotheses.

So, either we have a definition of science that includes other methods of investigation besides a strict experimental design, or we concede that much of human medical knowledge is not based on "science," which perhaps is a view you subscribe to. (Although unorthodox, a case can certainly be made for that view on a number of points but that's another discussion...)

I agree that hypothesis testing is at the core of the scientific method, and we both concede that scientists can be wrong. There are certainly many people who complain about how researchers flip-flop about various topics when it comes to human health, such as various dietary guidelines. Eggs are good. No, eggs are bad. No, wait, eggs are good. Butter? Margarine? Soy, good? Soy, bad? Hormone replacement therapy is good. No, hormone replacement is harmful. Mammograms every year starting at 40. No, mammograms not necessary until 50, and then only every other year.

To the dismay of average folks, that happens quite a lot with human-related research and one reason is because we can't conduct experiments on people like we can with lab rats. But that's just the meandering path that science takes on its quest for Truth.