scientists build to understand, engineers understand to build
Engineering is the use of scientific principles to design and build machines, structures, and other items, including bridges, tunnels, roads, vehicles, and buildings. The discipline of engineering encompasses a broad range of more specialized fields of engineering, each with a more specific emphasis on particular areas of applied mathematics, applied science, and types of application.
performance optimization/tuning is a type of debugging, i.e. a squaring of your mental model of the behaviour of a program with the reality of its behavior
as such, the scientific method and engineering, the application of science, applies
There must be something shared by all the sciences that makes them scientific, and it would be this somethign that is missing from the unscientific or trhe pseudoscientific.
What is common to the sciences is the basic structure of how they study, and the standards they use to judge acceptable results.
This is the scientific method.
We will presume that there is a shared shared method for all the sciences, and that it is the method that makes them scientific.
Objectivity Furthermore, we will presume that the criteria for being scientific are objective criteria. That is, it is not a matter of personal judgement as to what the scientific method is or what qualifies as scientific. These are impersonal, objective standards for that it is to be scientific, and there is an objectively accurate description of the scientific method.
It is the forum in which fundamental concepts and claims that are taken for granted in other disciplines and life on the street are questioned and clarified.
It may be about relations between people, where we rely on concepts like justice, morality, rights, and even what it is to be a person.
Or it could be about relations between people and nature, as in the concepts of knowledge, truth, observation, and evidence.
Or it could involve relations within nature itself, such as cause and effect, space and time, and the laws of nature.
The business of philosophy is to make sure we understand these important concepts and that the things we assume about them are true.
In some cases we find that we don't really understand or that there is no basis for our assumptions.
Philosophy of science does this with fundamental concepts of science.
What is "science"? What do all those things we call science have in common? What makes science scientific?
Usually the answer to this invokes a reference to the scientific method, and that requires some manner of empirical testing.
The details about testing, and making sure it leads to good reason to believe the results, must be examined.
More basic ideas emerge, like evidence, experiment, prediction, hypothesis, theory, law, and so on.
Science and philosophy have some important things in common.
Different sciences, like chemistry, geology, and biology, share a basic method.
They differ in what sort of thing, what aspect of nature, they study.
And then, details in applying the method will differ according to the demands of the topic.
Different aspects of philosophy have a basic method in common, a shared form of analysis. They differ in what they study.
Our focus will be on just one of the many topics addressed by philosophy of science. It is worth explicitly mentioning one of the topics to be skipoped, scientici realism.
The question of realism asks whether the best scientific results show theories to be true or, no less respectably, simply the most practical guides for dealing with nature.
The difference between the methodological work to be done here and the challenging issue of realism is a difference between describing and evaluating science.
Abstract Scientific method is not very different than what everyone does on a daily basis in coming to know about the world. Respect for evidence and reason are basic common sense and basic scientific method. Disregarding scientific standards and results in selected aspects of life amounts to disregarding common sense.
What distinguishes science from the other things people do or study?
It's not about what science studies, it's about how it studies things.
The basic ingredients of the scientific method are familiar to most of us: some mix of observation, evidence, testing, and logic is required.
Characteristics
In everyday life as in science, there there is neither supporting evidence, nor logical-mathematical proof, there is no knowledge.
And where there is inconsistency in the evidence, or inconsistency between evidence and theory, the responsible thing to do is withhold judgement either way.
Difference between science and common sense
The key difference between science and life on the streets is that the scientific process is more deliberate and explicit in following the steps and standards of the method.
In science the procedures must be articulated and described.
The scientific process is purposefully slowed down in the interest of control and transparency.
Science is thus more deliberate and dedicated thatn non-science in following the method.
Visibility
It is also more public and open to independent review.
In science, one is routinely expected to not just have good reasons in support of one's claims but to actually produce those reasons.
Science is a communal activity in which ideas, procedures, and standards are sharted and compared among people.
All of thee features of science, the slow, deliberate, explicit, public application of reasong from evidence, make the process clear and plainly visible.
What you learn about the scientific method will clarify the details of what we expect of common sense. Reciprocally, an intuitive appreciation of common sense will be helpful in understanding scientific method. There should be no reason to suspend the standards of one in the context of the other.
Refusing the authority og evidence and logic, either in the form of believing without evidence of believing in spite of contratry evidence, is not just turning away from science; it is turning away from good sense.
Naturalistic fallacy
It is important to restrict this view of scientific method to matters of fact.
It does not apply to matters of value.
Science, and the empirical side of common sense, can help us figure out the way the world is, not the way it ought to be.
Scientific method works for description, not evaluation.
Sometimes people ignore this distinction between fact adn value, and conclude that things really ought to be a certain way, simply because in fact they are that way. This is a logical fallacy.
This reasoning from the facts to an evaluation happens frequently enough that the philosophers and logicians give it a special name, the naturalistic fallacy.
Remember this law when discussing about laws in science, laws of nature.
Laws, like the law of gravity or the laws of thermodynamics, are more than simply generalizations of what does happen in nature; they are about what must happen in nature.
There is a kind of necessity associated with laws of nature.
If we can't imply "ought" from "is", can we imply "must" from "is"?
Inverse naturalistic fallacy
There is also a kind of reversed version of the naturalistic fallacy that should be recognized and avoided.
This would be drawing conclusions about the way things are, based on the way you think they should be, inferring "is" from "ought".
Wishful thinking commits this fallacy.
Uncertainty and degrees of certainty
There is another important limitation on scientific method and its companion common sense; neither results in perfect, indubitable certainty.
The available evidence and our own abilities to reason are limited.
This means there will always be things about nature that we just don't know, and maybe even can't know.
Sometimes, the responsible thing is to withhold judgement.
Limited evidence and fallible human reason also result in lingering uncertainty even in what we do know.
We deal with uncertainty like this all the time in life, and we are sensitive to degrees of uncertainty. So too in science.
Usually science deals in entities or processes that cannot be directly observed, so evidence is indirect and theoretical conclusions are never 100% certain.
But this does not mean that theories are pure guesswork. There is an important spectrum of good reason between guesswork and certainty.
The challenge of the scientific method is to locate a particular theory on that spectrum, and assign our beliefs about nature accordingly.
As evidence is collected, a theory can change position on the spectrum.
Another fallacy
Another fallacy is considering whatever is not 100% certain is pure guesswork fallacy.
It's true, we can't know for sure that x, but we have overwhelming evidence in support of x, so the responsible thing to do in almost all cases is to believe x.
If there are degrees of proof and a spectrum between guesswork and certainty, then there must be important details and nuances of scientific method that influence the degrees and locate theories on the spectrum.
Understanding these degrees of proof is the most important and challenging aspet of understanding scientific method, so it is to the relevant details that we turn next.
Abstract All scientific knowledge must be based on observation. This is the basis of scientific method, but there is some ambiguity in how close a link is required between observation and theory. The method cannot be simply a process of generalizing knowledge from observations, since some, at least tentative, knowledge is prerequisite for making scientific observations.
Theory is not truth or mean (un)believable
The term "theory" does alot of work in most descriptions and evaluations of science, yet equivocation and ambiguity are common.
Sometimes "theory" is used in a pejorative way, as ain invitation to doubt or even believe the opposite, as in "that's just a theory".
But other times "theory" is an honor, implying a coherent network of ideas that succesfully explain some otherwise mysterious aspect of nature.
A clear, unambiguous meaning of the term "theory" emerges from a survey of examples of cientific theories and seeing what it is they all have in common, what it is that makes them theoretical.
It is not going to have anything to do with how well-tested or well-confirmed an idea is, or how likely it is to be true.
Theories occupy all positions on the spectrum from near-certainty to pretty-speculative, so the term "theoretical" cannot distinguis an didea as being believable or not.
Theory definition
What do theories have in common?
They all describe objects or events that are not directly observable.
This is the core of the concept of theory.
Theory != Unreal
A theory describes aspects of nature that are beyond (or beneath) what we can observe, aspects that can be used to explain what we observe.
A theory is trye if it desribes unobservable things that really exist, and describes them accurately.
Otherwise it is false.
This shows the mistake in contrasting "theory" and "fact".
Fact
A fact is an actual state of affairs in nature, and a theory, or any statement for that matter, is true if it matches a fact.
Some theories are true, some are false, and the scientific method is what directs us in deciding which are which.
To say of some idea "that's a theory not a fact" is a confusion of categories.
Facts are, theories describe.
And a theory can describe facts.
Hypothesis
Unlike "theory", this term does refer to the amount of good reason to believe, that is, to the location on the spectrum between certainty and speculation.
A hypothesis is a theory that has little testing and is consequently located near the speculation-end of the spectrum.
It's a theory for which the connection to fact is unknown or unclear, but usually there is some tentative reason to believe this link will be made.
There is reason to think that evidence and support from other theories will allow the hypothesis to move up the spectrum to the well-supported side.
Being hypothetical is a matter of degree, and the term "hypothetical" wears off gradually.
It is the work of scientific method to determine the appropriate degree of being hypothetical.
Law
Theories differ in terms of their generality.
The most general theories, including the theory of gravity, are laws.
In other words, laws are theories of a particular kind, the ones that identify whole categories of things and describe their relations in the most general terms.
Being a law has nothing to do with being well-tested or generally accepted by the community of scientists.
A theory is a law because of what it describes, not because of any circumstances of confirmation.
And a theory is or is not a law from the beginning, even when it is first proposed, when it is a hypothesis.
The status of law is not earned, it is inherent to the content of the claim.
So neither "theoretical" nor "law" is about being true or false, or about being well-tested or speculative.
To describe a statement as a theory, or just a theory, does not imply any implausability or weakness.
All scientific knowledge must be based on observation. It must have empirical foundations.
Convervative extreme
Most conservative is the view that one should endorse claims only when one can directly observe what they are about and observe that ther correspond to the facts.
In other words, "based on" means explicitly linked to.
On this interpretation, theories, since they are about things that cannot be directly checked by observations, can only be regarded as useful models, but never as true descriptions.
This is not to say they are false, only that we cannot know which are true and which are false, because in this interpretation knowledge requires direct observation.
But this is just one extreme in the possible interpretations of "based on observations".
In our everyday lives, and on most accounts of science, we allow some manner of careful inference that goes beyond the observations.
The key is in specifying what kind of inference is reliable for making this extension.
Induction
Induction is a prime candidate, and making it the cornerstone of scientific method is the next most conservative account beyond allowing only claims about direct observations.
Staying close to the empirical roots would suggest allowing inference that starts with observation and results in theory, not the other way around.
It would require a one-way flow of information, from nature to us, from outside-in.
The process should be no more than a generalization of what is observed.
This is the essence of induction.
Limits of Induction
Some people, Isaac Newton among them, have claimed that inductive generalization from observation to theory is all there is to scientific method.
Anything else, such as speculative hypothesis, would be an irresponsible dirst step on a slippery slope to make-believe and mysticism.
But inductive generalization can't be the whole story about scientific method, and it is instructive to see the specific reasons why.
Pure induction, with only observations as premises, could never imply a statement about something unobserved or unobservable.
How did concepts such as atoms, germs, or curved space-time occur?
It could not have been simply by generalizing on what has been observed, since none of these things has been observed.
If the goal of science was simply to catalog empirical generalizations liek all metals conduct electricity, the pure induction might suffice.
Science routinely does more than that.
If offers explanations for how and why metals conduct electricity.
This is the value of science, getting beyond the merely observable, and pure induction does not suffice as the way to do this.
There is a second, more fundamental, reason why induction cannot be the whole story, or even the most important part of the story, about the scientific method.
Pure induction pressuposes pure observation, an uncontaminated flow of information from outside-in.
This is simply impossible.
In life as in science, perception is influence by ideas.
Scientific observations are influenced by scientific theories, so the order of events cannot be strictly observation then theory.
It is an important insight into scientific method to show not only that theory influences ibservation, but exactly how this influence comes about.
Here are four reasons why there canbe no theory-neutral observations in science.
1 Observation selection
It is impossible to observe everything.
Selecting what to observe and what to ignore cannot be haphazard.
Science isn't simply a catalog of observations; it's inferences fro relevant observations.
Some basic idea of what is relevant to what is needed in selecting what observations to make and record.
2 Detail selection
It is impossible to note and describe every detail of the observations that are made.
Selections have to be made, this time regarding the relevant aspects of the observations.
And again, the selections are not random or haphazard, but informed by some existing understanding of the situation.
Only some background knowledge allow scientists to safely ignore some details and attand to others.
3 Scientific observation
A third reson for some theoretical imprint on the information from the observation is the requirement that scientific observations be careful and reliable.
Observing conditions must be proper.
Relevant conditions must be controlled.
If machines are used, they must be working properly, etc.
Accounting for reliability will call on a theoretical understanding of how the machines work, which conditions are relevant, and what amounts to proper conditions.
Scientific observation, unlike casual observation out on the street, is accountable, it is always open to challenges to its accuracy.
Meeting the challenge calls on background knowledge.
4 Theoretical language
Scientific observations must eventually be rendered in theoretical language, and this certainly presupposed a theory in place.
Evidence
To summarize the ways in which theory influences scientific observation we can say that in science one needs evidence and not merely sensations.
Evidence must be meaningful and reliable.
It must be a credible indication of something.
Of course the theories used to make the best and the most of the evidence are themselves subject to revision.
But at least some tentative theories must be invoked in the course of scientific observation.
This means that the inductive route from observation to theory cannot be all there is to science.
There will have to be a flow of information back-and-forth, from theories to observations and from observations to theories.
Induction plays an important role in the discovery of a new idea, a hypothesis, but there is work still bo the done in testing the hupothesis.
Sometimes it is suggested that it is the testing stage that is the essence of scientific method.
It doesn't really matter how a hypothesis is discovered or how someone got the idea.
The important role of empirical evidence may come after the idea is proposed rather than before.
Empirical testing of a scientific hypothesis is always indirect.
A hypothesis is tested by making predictions and seeing if the predictions come true.
A look at the logic of this shows that a true prediction cannot prove a hypothesis.
Nor can a false prediuction disprove a hypothesis.
So empirical testing is always indecisive, and scientific method must involve more than just evidence and logic.
If a statement is about something that is itself observable, then the empirical testing can be direct.
But science is most interesting and mose useful to us when it is describing unobservable things.
This is where we will find explanations, and not mere summaries, of what happens in nature.
Theories, i.e. claims about unobservable things, are not amenable to direct empirical testing.
These claims are nonetheless accountable to empirical testing that is indirect.
The nature of this indirect evidence, and the logical relation between evidence and theory, are the crux of scientific method.
Statements about unobservable things can be tested by their observable implications.
To test the truth of a statement X, we reason that "if X is true, then we will observe Y".
Y is an observable implication from X, and it is by observing Y that X is indirectly confirmed.
If we look for Y but don't see it, then X is indirectly disconfirmed (falsified).
Y is the eviden for (or against) X.
Example: Theory of Evolution
When the theory of evolution was first proposed, it had to be empirically tested.
In this context, the theory of evolution was the hypothesis.
Since the theory describes events that cannot be observed, since they happened long ago in the past, the testing had to be indirect.
If the theory is true, then the fossils we observe today should fall into a pattern with older fossils.
We test the hypothesis by looking at fossils.
Example: General Theory of Relativity
When Einstein proposed the theory it had to be empirically tested.
The theory says that gravity is caused by the curvature of space and time, and since neither space nor time can be observed, Einstein had to figure out some observable consequences of their being curved.
He predicted that if space and time are curved, then light rays passing by a massive object will actually bend.
This bending should be observable, and we can test the hypothesis by this implication.
Since we are looking for a general pattern in the logic of indirect empirical testing, it will help to symbolize Einstein's argument.
Let H stand for the hypothesis in this, or any other, case of empirical testing.
Let p stand for the implication, i.e. the prediction.
In the particular case of testing the general theory of relativity:
H = Space and time are curved
p = Light rays will bend when they pass near the sun
Then Einstein's reasoning is in the form of an if-then statement:
If H is true, then p will be true.
or "if H then p"
Deduction
This kind of statement, "if H then p", is the central premise of indirect empirical testing.
Since it is a case of deducing the preduction p from the hypothesis H, any test that involves an if-then statement like this is called hypothetical-deductive testing.
The complete test requires observing whether the prediction p is true or not.
Hypothetico-Deductive (H-D) testing
In Einstein's case, it is a case of H-D confirmation.
The argument can be put in an abbreviated form:
if H then p (this premise is from Einstein's reasoning)
p (this premise is empirical, from observing stars)
-----
H
But look closely at the form os this argument.
It is not a valid argument.
It commits the fallacy of affirming the consequent.
The conclusion that the hypothesis is true does not follow from the observation that the prediction is true.
(Malaria example: if malaria then fever, but fever does not imply malaria)
So the succesful prediction of the bending of starlight did not prove that the hypothesis is true, any more than having a fever proves you have malaria.
The hypothesis could be false, as a false hypothesis can make a true prediction.
Furthermore, there is sirely some other hypothesis that makes the same prediction.
The moral of the story is this: a single true prediction does not confirm a hypothesis.
Disconfirmation
Suppose the prediction had come out to be false.
Suppose we had empirically discovered that p is false, that is, not p
This would be a case of hypothetico-deductive disconfirmation.
The argument still has the H-D premise.
The difference is the empirical premise.
if H then p (the H-D premise)
not p (the empirical premise)
-----
not H
Now the conclusion is that the hypothesis is false.
This argument is valid.
There is no way that the premises coud be true and the conclusion false.
The conclusion in this form of argument follows with absolute certainty.
Falsification
It seems as if disconfirmation of a hypothesis can be done with a single test, if the prediction is false.
Disproof of a hypothesis, falsification, appears to be decisive in a way that proof is not.
This apparent disparity between falsification and confirmation has led many people to claim that the essence of scientific method is falsification.
But this is wrong.
Disconfirmation seems so easy and so definitive only because we have ignored many of the important details in the example.
In particular, we have ifnored the theoretical details of how the prediction was deduced in the first place, and the practical details of how the experiment was done.
Filling these in shows that disconfirmation of a hypothesis is no more decisive than confirmation.
Example: Nuclear Fusion
Stars output enormous amounts of energy.
Where does this energy come from?
One theory is nuclear fusion.
The theory states that nuclei of hydrogen are being fused together to form nuclei of helium at the core of the sun where the pressure and temperature are high enough to squeeze the nuclei together.
This process is unobservable, because it occurs deep in the star and it invoves subatomic particles.
So the theory must be tested indirectly.
We need experts to predict what observable implications to look for.
One prediction involves an elementary particle called a neutrino.
Neutrinos are created by nuclear fusion, and neutrinos are so light and electrically neutral what they will escape from the center of the sun and fly to earth.
The prediction then is that we will detect neutrinos here on earth.
The prediction is very precise as to exactly how many neutrinos we will detect, given the amount of energy generated in the sun.
H = nuclear fusion is taking place at the core of the sun
p = a specified number of neutrinos will be detected on the earth
For decades no neutrinos were found in the right amounts.
The mismatch between the presumed correct theory and the reliable but disconfirming, evidence was vecing.
The missing neutrinos were recently found, but the problem had a life of several decades.
During this time were scientists unreasonable?
Should the not-p observation have forced them to conclude not-H? No.
When we add more of the scientific details, the logic of the H-D argument becomes much more complicated and the conclusion less certain.
We will find that in all cases of indirect empirical testing, a false prediction does not necessarily falsify the hypothesis.
Experimental conditions & repeatability
There are two kinds of detail we need to consider, theoretical details of how the prediction was made, and practical details of how the experiment was done.
p = if we put a huge tank of cleaning fluid deep under ground and use the appropriate radioactivity detector, then we will get a specified number of clicks on the detector.
The if-part of this statement is a list of experimental conditions.
C1 = do the experiment deep underground
C2 = use a sufficient amount of cleaning fluid
C3 = use the appropriate radioactivity detector
C4 = have the detector warmed up
These specify how the experiment is to be set up.
The then-part is the final expectation, the predicted outcome.
The experimental conditions are effectively the recipe for doing the experiment.
It is the precision and explicit presentation of experimental conditions that are the key to repeatability in scientific conditions.
We need a detailed record of what was done, in case we want to do it again, for ourselves.
This is also an essential component in judging the credibility of the data.
It specifies just what it takes to do the experiment properly.
Expectation
The then-part of the prediction is what to expect when the experimental conditions have been properly done.
The predicted result of the experiment is the expectation.
E = there will be a specified number of clicks on the radioactivity detector
p = if C1 and C2 and C3 and ... then E
In other words, if all the conditions are done right, then we will get the expected results.
All scientific experiments are like this, where the final result is carefully controlled by the epxerimental conditions.
From now one, whenever we use the symbol p to stand for the prediction, we are it as an abbreviation for the if-then statement involving the experimental conditions and the expectation.
Theoretical considerations
It takes an expert to come up with good predictions.
All that background knowledge is theoretical, in the sense that it is about things that cannot be observed.
The individual statements that are drawn from background knowledge and used in deducing the prediction are often called auxiliary theories.
Experimental considerations
Now we need to see how the experimental conditions and the expectation affect the logic of H-D testing.