This is an index for an upcoming series of posts on the various dichotomies used to apply abstraction to modeling and problem solving. The list below will gradually be replaced with links.

• Precision / Accuracy
• First Principles / Empirical
• Syntax / Semantics
• Endogeny / Exogeny
• Qualitative / Quantitative
• Subjective / Objective
• Transient / Steady-State; Kinetics / Thermodynamics
• Terminal / Instrumental; Ends / Means
• Induction / Deduction

Each pair of terms is hopefully balanced, being specific to the same degree as the other. The different pairs are not exactly orthogonal, but are hopefully orthogonal enough for the collection as a whole to serve as a typology.

The MIT Alum club in New York organizes an event called finance brunch, which forwarded me this article last weekend:

http://www.truthsetsusfree.com/ModernMoneyMechanics.pdf

There is a blow-by-blow account inside on the mechanics of the balance sheet when a bank makes loans and takes deposits.

The basic idea is this: ΔAssets = ΔLiabilities is always true because it is an accounting identity. However, liquid assets = liquid liabilities is not always true.

When a bank lends out money, it merely creates a number in its ledger. The (asset~liability) pair is (loan~checking account). When the borrower or a depositor withdraws currency, however, it is (reserves~checking account) that is deducted. The reserve requirement is the required ratio of reserves to checking accounts, basically a requirement that banks always be able to pay cash for a certain fraction of its checking account balances.

In a digital computer, wires have one of two states, 0 or 1. This is not natural, however - many possible voltages can be put on the wires, but only a small subset of these possible voltages ever occur inside a digital computer, being those voltage values which can be unambiguously interpreted as either 0 or 1. This is a design goal - the components of a digital computer work in such a way as to maintain the interpretability of the wire voltages, and also to maintain the  interpretability of the transistor physics as logic gates.

This idealization of the computer's voltage states is used to standardize the components and the component interactions. That standard behavior then forms the basis for logical induction, such that we can design circuits which as a whole have too many possible states to exhaustively test, yet can be known to be correct by virtue of the correct interactions between its parts. The whole can be known to be correct because the parts are correct. There are many devices that can be interpretable as logic gates - different implementations of the same idea. Implementations include domino tiles, water channels, game of life, wooden blocks. The reason we choose silicon is because silicon devices maintain interpretability even at high frequency, and can be made very small.

---

Economists have theories about how economics processes work, and many of these theories start with assumptions like "no transaction cost" and "no externality". These don't have to be taken as passive observations about reality. Rather, they should be taken as conditions worth bringing about, because of their utility in market processes.

One case which interests me greatly is the government intervention used to create the neutrality of money. Brad Delong describes Keyne's General Theory as trying to make Say's Law true.