Douglass North, On One Foot

Arnold, there is a problem with your bumper sticker. There is not agreement among institutionalists about the definition of institution. I prefer A. Greif's definition in his recent book (I suggest you to compare this definition with anyone given by North in his "Understanding the process of economic change"; they are quite different).
You may remember that old bumper sticker associated with Friedman, "Money matters". It had the same problem; in the 1970s I liked to define money as M? so I could accommodate the evolving definition of money.
Your three-point answer to Eric poses the same problem. What do you mean by "institutions are what lock in economic patterns"?

My last post was deleted, I hope not on purpose (please take my criticisms as a compliment).

Anyway, I think "institutions matter" is too vague, as it is the bailiwick of sociologists and poli scientists to no great effect. Why not, 'institutions matter, and must be focused on general welfare, not their members'. Thus bad institutions are anti-productive because they merely wish to monopolize their services and maximize the welfare of existing members (eg, teachers, auto workers, the United Nations). Good institutions are built by outsiders for outsiders, like good constitutions, courts, stock exchanges.

This is completely armchair, but maybe we're overestimating the value of having a comprehensive or simple philosophy when it comes to economics, or at least economic history.
In some fields (math, some types of cognitive science) having a simple, general theory is desirable because there is a bounded range of phenomena that you are trying to model. Defining such a range in economics is difficult, because many of the systems (read: institutions, however defined) that are of interest to economists have their own unique parameters that may not be shared with other systems. My point: human decision making is context sensitive, and the range of possible contexts is effectively unbounded.

This is why, for example, a history of economic activity that is patchwork (a collection of unrelated theories to describe a collection of unrelated phenomena) may be more illuminating than one that trys to group a bunch of things under the umbrella of a few axioms.

I say all this because I study linguistics, so when I try to build a model of language I know specifically what the output of the model should be (because deducing all the variables that you need to incorporate is a doable task). Economics, on the other hand, doesn't seem to be this way, since sometimes it is nearly impossible to identify all the variables you need to incorporate into a predictive model. There will never be an e=mc^2 of economics (however definied) because its not a fundamental system; that is, its amalgamation of a nonfinite number of possible subsystems. I hope you get what I mean.