Hello Folks! Apologies for the delay in posting a meet, which was induced chiefly by difficulty finding a location that was both spacious enough to accommodate our group and accessible by BART. I finally found one, in the very shadow of the 24th Street and Mission BART station, and talked to the owners of the said locale, the Silverstone Cafe, who would be happy to have us. Our meetings will be held in the courtyard in the back, weather permitting, or inside else-wise.
The key to finding a location was looking at the environs of several BART stations. Likewise, the key to empirical investigation is doubtless looking at several hypotheses that are available in the sense that they seem reasonable and are testable. But can we say anything in general about what makes them 'look reasonable', or give us the related information of which among reasonable hypotheses are most reasonable so as to decide which tests to conduct first? One of the very few guidelines that have been consistently proposed as universally applying to the choice of hypotheses is known as ontological simplicity, or Occam's Razor, which states in its original version that “entities should not be multiplied beyond necessity.” A more modern formulation might run that simple hypotheses are to be preferred to complex hypotheses, provided that we are willing to make a set of background or auxiliary assumptions that together with the simple hypotheses are capable of explaining what we observe. But what are the possible justifications of such a principle, and do these justifications lead us to believe that we should apply the Razor everywhere or instead tell us where to apply it?
Clearly, we cannot declare that all hypotheses we happen to consider simple are more likely to be true than all hypotheses we happen to consider complicated. Even if ontological simplicity simply is more likely in nature than ontological complexity, simplicity in our language does not equate to simplicity simplicitur. The elements of everyday language are not the elements of the ultimate constitution of reality. For instance, in everyday language Aristotle's hypothesis that sublunar and supra-lunar objects simply are different may be parsimonious (apply Occam's razor). They would seem simply different to us too if we didn't already know Galilean-Newtonian science, which uses a language whose building blocks are different from those of the language prior to that science. In retrospect, that there be only one kind of object in the universe with respect to gravity seems like the essence of simplicity, but the very fact that it didn't occur to people for a very long time suggests that what we consider simple is not just a given but a function of how leading theories of our day teach us to think.
What explanations we consider parsimonious with regards to a specific subject is of course also a function of our background information concerning that subject. When he learns that a certain county in early American New Jersey was Federalist, though surrounded by and economical similar to Democratic-Republican counties, a historian of the period actually ignores a number of extremely simple explanations that the evidence doesn't contradict explicitly but that don't fit the pattern he has learned generally governs early American politics. It is not likely, for instance, that the inhabitants of this particular county have an ideology favoring centralized government that overrules their economic concerns if that has been found to be the case nowhere else in New Jersey or the early Middle Atlantic US. It may be the verbally simplest hypothesis, but it may also simply be a waste of time to put forth the considerable effort to test the hypothesis against the record because background information makes it unlikely. This it does largely by requiring very low-probability events within the domain researched that would serve to make this county simply different. Like considerations forms the basis of a view of Occam's razor as applying only given a particular definition of simplicity formulated relative to background knowledge. An example of this family of views can be found in the Eliot Sober piece whose link I attach at the end of this mini-essay.
Both of these qualifications of simplicity simplicitur suggest that ontological simplicity may not be the best principle to apply chronologically first in investigating a completely mysterious natural phenomenon, because having an idea what it really means in such an instance requires a lot of prior work. It may be a great principle to apply pedagogically, that is when reconstructing why a theory is true for novice students, but that does not make it the key, or the first key, to actually producing human knowledge for the first time. So one perhaps cynical answer to the question of when Occam's razor is a good principle to apply is 'when you are already sure that what seems simple really is simple'; or more precisely, when one has a good theoretical framework which there are reasons to be believe maps words unto things as they actually occur.
However, there is a more direct criticism, rather than a qualification, of the razor that involves situations where we already think we have a theoretical framework. That theoretical framework may seem to parsimoniously make good prediction and invoke only entities we have good reason to believe are out there, but be simply wrong on account of other considerations we were unaware of. Much here turns on the imprecision of the term 'good prediction'; there may be good predictions, but new entities or principles of greater complexity may make yet better predictions. The deepest question concerning the razor is perhaps 'when is simplicity real, rather than just convenient?', which relates it to large and intractable metaphysical issues.
At http://plato.stanford.edu/entries/simplicity/, a number of different kinds of general justifications of Occam's Razor and certain other issues and criticisms can be found.
At http://rationallyspeaking.blogspot.com/2011/05/razoring-ockhams-razor.html may be found a summary of a Sober-like view.
At http://philosophy.wisc.edu/sober/Let's%20Razor%20Ockham's%20Razor.pdf is the Sober piece itself.
I have also put my personal view of the Razor in the 'Files' section of this website.