How scientific empiricism works
In my judgement, empiricism is the view that knowledge comes, ultimately, only from observation of the real world. This does not mean that we must each repeat every experiment done by every scientist in history in order to know about the scientific description of reality; of course not. We can read non-fiction books, attend school and college, and go to other lectures and classes where information is presented; and we can, nowadays, watch video documentary material from various sources, and read website content, all of which we must regard as potential indirect sources of empirical information.
However all books, lessons, talks, documentaries and other sources of scientific information must be regarded with caution, with an attitude that I shall call skepticism, although that term must not be taken to carry all the baggage attached to it by some discussions of the term. (See My disagreement with Wikipedia about empiricism.)
Evidence and skepticism
The degree of skepticism which we should use must vary with the kind of document or other source of information which we are considering. If we read something in a science journal, we should be able to consider it more reliable than if we read it in a posting to Facebook or Twitter — just to give an example of extremes of credibility. In the same way, we may give more credibility to a BBC science documentary than we can give to a YouTube clip posted by a random person and made by an unknown person or persons.
So much, for now, for sources of reports of observations. This brings us to the distinction between empirical observation and other uses of the term “evidence”. Empirical evidence means observation of the real world. The word “evidence” is used in other contexts such as descriptions of the proceedings in a law court; however, what a person says (whether in a law court or elsewhere) is, empirically, only evidence either that the person believes what they say, or that he (or she) is telling a lie but wants us to believe it. What they say is not empirical evidence of anything except of the fact that they said it.
In the same way, the fact that an ancient document (or a copy of a supposedly ancient document) contains a statement does not constitute evidence of what the statement says. From a scientific perspective, it is only evidence that somebody at the time of the original writing of the document had some reason to write what wrote. As before, the reason might be that they believed it to be true or that they were telling a lie but wanted readers to believe it. Or they might have been telling a story simply to entertain their readers.
The development of scientific theory
So how does a scientific theory develop? It begins in one of several ways. One way is that an observant person notices a pattern and wonders whether this is a sign of an unknown law of nature. Of course, it might be a sign of a law of nature already known to other scientists; the first thing the observant person must do if they decide to investigate the pattern is to check whether this pattern is a consequence of a known law of nature. If it is, the only question is whether the new observation could be of interest as supporting evidence for the relevant branch of science. (If it is, the person might report what they noticed to scientists who might be interested. Otherwise, they just forget it and move on.)
However, if the pattern which the observant person notices is not predicted by any known law of nature, this may be an opportunity for new research; and if the person is a scientist, and if the observation is in an area connected with their work, this might be an opportunity to do some research on the observation. So let us assume that a scientist has heard of an observation of a pattern which is not to predicted by any relevant branch of science. The next step is to decide whether the pattern suggests some new law of nature.
Another way in which is scientific theory may develop is when there is a need for technology which poses questions for which science does not yet have answers. When this happens, scientists may be asked to examine behaviour of systems in use in order to develop new systems with more capabilities.
What then happens is that scientists must formulate a description which is new by containing at least one new proposition which can be tested by carefully chosen observation or experiment. Here by “observation” we mean observing nature with, where suitable, instruments and recording devices, but without setting up the situation to be observed. This is how scientists study natural phenomena.
By “experiment” we mean cases where, in order to study nature, the scientists need to set up a system to exhibit the behaviour in which they are interested. It should be clear that this means that kind of work which is done in places that are called laboratories.
What the scientists must do is establish precisely which observations to make and which measurements to take, so that they can examine those results and make a judgment about whether these results confirm the truth of each proposition which they are trying to confirm or disprove.
The evaluation of propositions
It is very important to understand that scientific work can only disprove any given proposition. What science does is to try to disprove statements from existing or new scientific theory. When all attempts to prove that a proposition is wrong have failed, the proposition is considered worthy of including in the standard scientific model of the world.
Empirical science does not claim to prove that positive, generalizing scientific statements are true! It attempts to disprove such propositions, and concludes that a specific proposition is probably true when all attempts to disprove it by experiment or field observation fail.
According to the amount of work done in a given area of science, and the completeness of the set of observations made while attempting to disapprove some given statement, scientists can develop a high degree of confidence that the statement is true. One way to express how confident be in any given statement is to use a number with a value between 0 and 1.
Sometimes scientists issue public statements in which they express how confident they are in results using such a number, and usually this is in the form of a percentage. One such case was the announcement about the Higgs boson, where statements mentioned the confidence level of 98 per cent in the conclusions being announced.
I say more about this way of expressing how confident we are in a proposition in Proposition evaluation.
More about empirical truth
The important thing to remember here is the difference between empirical truth and mathematical or logical truth. In mathematics and in pure logic, the result of of piece of work is typically a proof of a proposition. As I hope I have made clear here, this is never what happens in empirical science. Many religious people and others with delusions about reality have no understanding of this important distinction.
One final important point to emphasize on this page on this topic: notice (and never forget) the limiting adjectives governing the class of scientific statements which empirical science does not claim to prove directly: positive, generalizing statements.
An example of such a statement is: “All swans are white”. This is a positive statement, and a generalizing statement. Centuries ago, people in Europe might have supposed that this statement was true: it was before the discovery of Australia. Of course, when the black swan was discovered there, the European generalization hypothesized here was found to be wrong. (Yes, I know that there is also a South American black-necked swan that has a white body but a black neck.)
One way in which empirical science can formulate positive generalizations so that they can be considered true when uttered is to specify limitations to what is known. Continuing the present example, if a 17th century scientist had published the statement “All currently known species of swan are white”, that statement would have been true at the time, whereas the generalization without the specification limiting the statement to species known at that time was already false, if anybody had uttered (let alone published) it. Only, anybody claiming that was being too careless in their generalization.
It is worth contrasting positive generalizations with negative ones. These are often perfectly safe. For example, because we now know about the black swan, it is perfectly safe to say “Not all swans are white”, or “Some species of swan are not white”. The only way that these statements could be made false would be if the black swan (and those others) became extinct; but even then, only the formula (about swans being white) specifying that the statement only refers to known species is safe. As long as it is conceivable that some species of wildlife of any variety might as yet be unknown to zoology, specifically (here) to ornithology, positive generalizations not explicitly limited to current knowledge will be unsafe.
However, in this discussion remember also that the kind of statement made by physicists about the laws of nature is necessarily different from the kind of statement made by zoologists, because there is an underlying assumption that the laws of physics apply throughout the universe. Even so, there is a caveat that is understood to apply, by anyone knowledgeable enough to read new research: namely, as far as we know the laws of physics apply throughout the universe. For now, it is assumed that unusual effects are caused by phenomena only found in nature (as opposed to nuclear research labs such as the tunnels at CERN around the large hadron collider) away from the Earth. For example, black holes ... but no, I am not implying that black holes actually occur at CERN.
Note, in conclusion, that there is no semantic difference between empirical truth and mathematical or logical truth. In any context, any given statement, any proposition stated clearly enough so that its meaning is not a linguistic or semantic problem, is either true or false, as stated. The nature of truth does not change. What changes is the degree of possibility of being able to assign with any degree of confidence an evaluation of being true or false to each stated proposition. The discussion of Proposition evaluation is about the circumstances that prevail in different fields of study which limit our ability to assert whether or not any particular proposition is true, and (whenever we cannot) what we can say about our degree of confidence that, if we ever did reach certainty about a given wording, the value would be truth or falsehood.
That concludes this description of how scientific empiricism works.