Context : OnScience
We can be interested in demarcating science from other types of knowledge gathering project.
KarlPopper and others have been concerned to demarcate science as a rational and legitimate activity from mere pseudo-science, those projects and theories which masquerade as science, but are in fact based on irrational denial of contradictory evidence. But not all non-science is pseudo-science.
I'll use metaphysics as a general catch all for all those forms of legitimate enquiry that aren't answerable to empirical testing, but which are rational in that they are answerable to rational criticism. These are broadly those projects which can claim to investigate analytic truths by investigating the logical consistency of our sets of theories.
Here I depart from the standard Popperian line on science. But use a distinction due to DavidHull
I take it that science is the study of universal classes and particular things qua members of classes, whereas history is the study of historical particulars and their interactions without references to their classes.
For example, Philip, is a particular. If you study Philip, qua massive body, falling down the stairs, you are doing science. But the statement "Philip lives in Brasilia" which relates two particulars, without reference to any types, is a historical claim, not a scientfic one.
See also DefendingAdaptionism
There are a couple of implications of this definition.
- Some things, you might think of as science, turn out to be questions of history - or the term I prefer : natural history. These include questions about biological species or lineages, which are historical particulars. It also includes questions about cosmology, such as "what happened in the first minutes of our universe" or "how did the Earth form". Of course, these historical questions can still be informed by scientific theories of planet formation.
- Historicism for Popper, is a reprehensible, irrational way of thinking. On the other hand, I think that we can make a case for it's equivalent rationality. Just as the scientist conjectures the existence of a kind, and that a particular is a member of it, so a historicist could conjecture the existence of a trend and that several moments are points within it. (See also OnHistoricism)
- It doesn't seem to me that there's anything in the logical structure of trends and moments which makes it less rational than kinds and members.
: (Note that this is not a particularly useful defence of some of the projects Popper criticizes as historicist. I think, some sort of Marxism, for example, is better defended by being recast as a properly scientific theory about the dynamics of economies which is open to falsification. In practice, Marxism should perhaps be seen as an orienting attitude which can inspire many different scientific hypotheses, many of which will be properly shot down. See TestingMarxism)
Update : I wrote a version of this answer on QuoraQuestionSite today : http://www.quora.com/What-questions-can-science-not-answer/answer/Phil-Jones
From Conspiracy Theories
I started describing ConspiracyTheories as the "religion" of NetoCracy. Or rather, playing the role for NetoCracy that religion plays for Feudalism. Check the ConspiracyTheories page for a better explanation. Here it's worth noting that, by the discussion above, ConspiracyTheories are really a kind of "history". They're hypotheses about the causation and information flow between a number of individual.
From Art / Design / Technology
Engineering and technology often apply knowledge gained scientifically. And engineers need to know a lot of the same stuff as scientists. But the output of design and technology are things whereas the output of science is knowledge
See ArtAndScience, DesignTheory, ComputerScience
From Patterns / Sciences of Form
This is, for me, the most interesting case. Sciences of form study patterns of interaction and influence in the abstract. For example, the study of ReactionDiffusion models in computer simulation which can be applied to chemical reactions, populations of bacteria and star formation.
These are interesting in several ways ...
It can be argued that studying abstract computer simulations are closer to a kind of mathematics (metaphysics) than science. As I argue in ALifeAsScience, computer models may be a form of properly scientific investigation because they study real examples of members of classes which are defined by their functional properties.
More difficult for my Popperian purism is the question of falsifications. Patterns are highly statistical theories, they are SoftLaws. There are plenty of exceptions.
Another question is what is the relation between these kind of patterns and the design patterns of ChristopherAlexander. What kind of knowledge is captured by a pattern. What kind of project are we engaged in when looking for patterns? Are we doing science? Or metaphysics? Or history? Or something else?
(See also PatternLanguage, TheScientificStatusOfPatterns)
- review of IsabelleStengers's book : http://www.shaviro.com/Blog/?p=401
Discussion with ScottMartins : http://pedantry.fistfulofeuros.net/archives/000776.html