The definition or nature of "science" has been debated in history for many decades. According to the current theory, science plays the same role for a collective intelligence as the beliefs for an individual intelligence, that is, to summarize the common experience of the members into a general, compact, and structured form, which can be used to predict the future, and guide the system to take promising and efficient actions when achieving its goals.
Therefore, it is the following major features that distinguish science from other types of knowledge and belief:
Science is often said to be a collection of "objective truth" or "accurate descriptions of the world". Though such opinions usually work well in everyday life, they fail to explain the development of science, and to clarify the difference between science and other types of knowledge. As discussed in the previous chapters, the knowledge of an intelligent system is nothing but summary of the system's experience, and therefore is partially determined by the system's cognitive capability. What distinguishes scientific knowledge from personal knowledge is that the former is shared, or socialized, among multiple systems, not restricted to a single system anymore. In some branches of science, this request takes the form of asking for repeatable experiments and rejecting one-time observations. Here science is "objective" in the sense that it is "unbiased from personal opinion", not that it is "belonging to the object" and independent of any observer.
As explained in Chapter 5, not all kinds of knowledge are expressible in a communication language. Inexpressible knowledge includes sensorimotor experience, procedural skills, intra-conceptual and inter-conceptual priority rankings, etc. Though these types of knowledge have high importance to an individual system, they cannot be considered as scientific. The fact that some of them can be fully expressed in the internal representation language of the system is irrelevant — they are still not socialized, after all. Only socialized knowledge can be communicated from system to system, and passed by education from generation to generation.
As a special case, it is easier to establish shared outer experience than inner experience, since internal events of a system are only observable to the system itself, their descriptions are inevitably from a first-person point of view. On the contrary, the external events happen in the shared environment, so can be described from a third-person point of view. It explains why a "scientific explanation of consciousness" is still different from "the experience of consciousness" itself.
Furthermore, even knowledge fully expressed in a communication language is not considered as scientific, if it is not accepted by other members of the community. A new opinion in science is always initially promoted by an individual, who has strong reason to believe it according to personal experience. Such an opinion is considered as a hypothesis, and will be considered by the other members. If it is supported by, or at least consistent with, their own experience, it will be accepted. Please note that when we say "science includes confirmed/justified hypotheses only", the confirmation and justification means the hypotheses with other people's experience, or "the world as people perceived", not directly with the world itself, or "a description of the world as it is". Consequently, "scientific knowledge" is historical, since a belief may be considered as part of science at a certain time, but not at another time in the past or in the future.
As many other criteria, whether a piece of knowledge is "socialized" is a matter of degree, so it makes sense to say one opinion is "more scientific" than another because the former is supported by more people — since this means the knowledge is supported by more evidence, collected from multiple system's experience, the knowledge indeed has a higher truth-value (see the definition given in Section 3.3).
This request is not implied by the above request for the knowledge to be socially confirmed. On the contrary, if a theory just want to be consistent with people's experience, a simple way is to make the statements in the theory very general and flexible, so as to be consistent with whatever observation there can be. For example, it is always correct to predict the price of a stock as "it will fail whenever it is too high", but without further specification on how high is "too high", such a prediction is almost useless to a system, though it is indeed "true".
This problem is especially common in philosophy, being the most general theory among theories in various domains. Many metaphysical opinions are so far away from experience that they become completely useless, though it is hard to say that they are "wrong". Triggered by this problem, positivism require scientific knowledge to be verified by sensory experience, while people like Karl Popper demanded it to be falsified. Though these ideas are intuitively attractive, they cannot really be used to draw a boundary for scientific knowledge, since they are too restrictive. As summary of experience, scientific knowledge tends to be general, so its truth-value can neither be verified nor falsified by a single testing case. The reasonable factor in this ideas, however, is the request for the knowledge to be instructive to the system's behavior, that is, it should make concrete predictions about future situations, and take the responsibility — based on new experience, a previous prediction may be either right or wrong, then the truth-value of the relevant knowledge should be adjusted accordingly, even though only for a little if the current evaluation is based on a large amount of evidence.
This is where science differs from religion. In every religion there are some core beliefs, which are considered as "the truth", and cannot be challenged no matter what new experience is obtained. To make these beliefs consistent with all possible observations, the beliefs have to be tautological, and therefore not instructive. Many people think science is preferred over religion because it has been confirmed or it can explain observations, though religion actually does better on those aspects by avoiding to make accurate prediction and to give concrete instruction. Consequently, though religion may play other role in a community, it is not a knowledge type that provide unambiguous guidance for the system when dealing with the environment. On the contrary, a scientific theory must make concrete prediction about future events, and be ready for revision if the predictions are different from future observations. Therefore, it cannot consists of general statements that explain everything but predict nothing.
As usual, whether a theory is instructive is a matter of degree, and can only be addressed when comparing with other theories. Philosophical theories are always less instructive than theories in natural or social science, given their generality, though it is not a reason to deny their role in guiding people's behaviors, usually on a larger scale than the concrete sciences.
A scientific theory is not an arbitrary collection of widely accepted instructions. Instead, the knowledge in a theory is harmonize, generalized, abstracted, condensed, and structured. It summarizes a large amount of concrete experience, and express it in a form that is highly efficient for communication, education, and application.
A scientific theory is the result of deliberate mental works, and it does not directly come from experience. For given experience, the possible theories one can build is usually not unique, though not all of them are equally good in terms of systemization.
This topic will be further explored in the next section.