Philosophy of Science

Karl Popper thought that the fundamental feature of a scientific theory is that it should be falsifiable. to call a theory falsifiable is not to say that it is false.

Popper criticized Freudians and Marxists for explaining away any data that appeared to conflict with their theories, rather than accepting that the theories had been refuted.
In general, scientists do not just abandon their theories whenever they conflict with the observational data. Usually they look for ways of eliminating the conflict without having to give up their theory.
A simple criterion for demarcating science form pseudo-science is unlikely to be found.

To be falsifiable is no doubt true for any scientific theory, but the more important is what to do facing the conflicting facts. 

Deductive reasoning is a much safer activity than inductive reasoning. when we reason deductively, we can be certain that if we start with true premisses, we will end up with a true conclusion. But the same does not hold for inductive reasoning.
So for Karl Popper, if a scientist is only interested in demonstrating that a given theory is false, she may be able to accomplish her goal without the use of inductive inferences.

Without inductive there's no way to generate new knowledge.

Hume's "uniformity of nature" dilemma
Inductive inference is used to take us from the examined to unexamined, and to provide the best way of accounting for the available data.
A popular way to choose the best answer for the competing hypothesis is the simplest and the most parsimonious one.

That's why it is important to 大胆假设, for almost there's no way to justifiy one hypothesis is better than other alternatives. 

Probability could be a subjective estimation, or the objective measure of the strength of evidence in its favor.

The logic interpretatino of probability itself is a product of inductive inference, for example statistics is to generalize the observation on a sample to the whole population.

Hempel's Covering law model
Scientific explanation consists on 1) the premisses should entail the conclusion; 2) the premisses should all be true; 3) the premisses should include at least one general law.
General laws+particular facts=explanation of particular phenomenon
Hemple implies that every scientific explanation is potentiall a prediction.
Nevertheless Hemple's model has the symmetric problem (that means you can exchange the position of facts and phemoneon to be explained, and then use the phenomenon to explain the facts)

A good explanation should also contain information that is relevant to the phenomenon's occurrence.

Natural sicence or more specifically physics is not superior than other discplines, for the objects studies by other disciplines are multiply realized at the physical level (so diverse that it is impossible to define it in a single physics term).

Anti-realism is sometimes called "instrumentalism" for scientific theories are regarded as instruments for helping us predict observational phenomena, rather than as attempts to describe the underlying nature of reality.
The limits to scientific knowledge are set by our powers of observation.
Realism proponents use the empirical success of theories that posit unobservable entitites to argument for scientific realism even on unobserved world, but there existed in the scientific history the counter examples.
Unobservable vs. unobserved (the effects of techonolgy development)

The positivists argue that it makes no difference how a hypothesis is arrive at initially. what matters if how it is tested once it is already there, for it is this that makes science a rational activity.
Again, 大胆假设, 小心求证。
The disputes between rival theories could be solved in a perfectly objective way, by comparing the theories directly with the neutral observational facts, which all parties could accept.

Kuhn's contribution
Normal science is a paradigm, a constellation of shared assumptions, beliefs, and values tha tunite a scientific community and allow normal science to take place. The job of the normal scientist is to try to eliminate these minor puzzles while making as few changes as possible to the paradigm.
Adopting a new paradigm involves a certain act of faith on the part of scientist. He allowed that a scientist could have good reasons for abandoning an old paradigm for a new one, but reasons alone could never rationally compel a paradigm shift.
Concepts cannot be explained independently of the theories in which they are embedded, and the data are "theory ladenness", no such data that are neutral and can be accepted by all.