Thanks for your thoughtful reply! I suppose it is rather appropriate that when I wrote my thesis on Fundamental Theory that I included photographs of headstones from the graveyard where Eddington is buried. At any rate, I'm not sure I entirely agree regarding Fundamental Theory, but that's for another discussion over a drink sometime.
I still think I disagree about reductionism, though. As big a fan as I am of dictionaries, I find they do not always capture the subtleties in the actual usage of certain words. So nothing in the definition you posted is necessarily wrong, but I think the interpretation of that definition is almost too literal.
Consider a car, for example. I don't think anyone would disagree with the suggestion that a car can be easily understood via reductionist methods (in fact, Robert Pirsig demonstrates this, albeit with a motorcycle, as a beautiful demonstration of a reductionist scientific method in his Zen and the Art of Motorcycle Maintenance). But are there facets of a car that are meaningless outside of the whole? Absolutely! We could understand how every individual part works *and* how they work together to move the car, but the *purpose* of the car is entirely holistic.
Now consider a mechanic. Can a mechanic fix a car if he/she has a wholly holistic understanding of it, i.e. only knows its purpose? No because if the car fails the mechanic must still understand how the individual parts work in order to figure out how to fix the car! Yes, he/she needs to know how each of the parts is connected, but fixing the problem means isolating it.
As George pointed out in his essay, in many situations there are holistic issues that often enact a "top-down causation" effect where the "whole" somehow enacts restrictions on the parts. But that's really true of anything. For the car example, an accelerator peddle will go completely to the floor if the cable attaching it to the engine snaps (this actually happened to me). The limitations in the motion of the accelerator pedal are driven by a whole host of issues, many of which are "holistic" or even unrelated to the actual mechanisms of the car itself (e.g. laws may constrain design). Despite all of this, I can't see how a reductionist method could possibly be avoided here nor do I think reductionism itself excludes certain holistic notions such as "purpose" (incidentally, in order to determine the purpose of a car, if one is ignorant of such things, amounts to obtaining more information which is, in itself, a reductionist thing to do - the "whole" doesn't proclaim itself a car).
So the argument goes that there are certain phenomena that are simply either too complex or too abstract to be understood in the same way as a car. But this begs the question, how do we *know* these phenomena are too complex or abstract? Are we simply assuming they are since our usual reductionist methods haven't worked (yet)? If so, then we are a priori assuming there's a problem with reductionism. But this is a logically unprovable assumption. As Carl Sagan once said, "[y]our inability to invalidate my hypothesis is not at all the same thing as proving it true."
Or, what happens if we give up when we are just on the verge of understanding but don't realize it? Worse yet, what happens if this anti-reductionist movement takes on a life of its own and drives reductionism into obscurity? Because the argument that certain phenomena are too complex to be understood in this reductionist manner is the exact argument that is used by proponents of intelligent design and creationism. I find that a bit frightening.
At any rate, regarding shape dynamics, I agree that there absolutely is a huge conceptual difference between it and Newtonian dynamics. I understand precisely where you are coming from on this. However, I completely disagree with the idea that you can quantify the difference between two nearly identical wholes without appealing to something external to both (this is exactly what Eddington tried to do with Fundamental Theory). I mean, certainly at some point we get into some kind of recursiveness (even language is recursive since it is used to define itself). But with things like shapes, how does one quantify a difference without some reference? At some point one needs to define something which I say is reductionist. You may say this is still holistic, but at this point the argument has become one of semantics because the fact is that we need to define properties (which are inherently *not* holistic) by which we can compare the shapes. (I have more to say on this point, but I'll send you an e-mail about it.)
Now, in regard to my point about reductionism as a methodology versus reductionism as a formal structure to the universe, let me explain the latter by comparison to the former. While it might be possible that the universe contains structures whose function or purpose or nature simply cannot be understood by reductionist methods, this is *not* the same thing as saying that those structures' function or purpose or nature actually *is* independent of the behavior of its constituent parts. It may simply mean that there are limits to our *knowledge* of the universe. Part of this comes from the fact that we are *part* of the universe we are attempting to describe and thus naturally we will run into some problem of recursiveness. But taking a more holistic approach won't necessarily rid us of this problem.
So, reductionism doesn't obviate the need for holism, but fundamentally reductionism is still at the core of the scientific enterprise and must remain so if it is to remain science and not succumb to a lot of hokus pokus.