The following is a line of thought I had.
I'm not particularly confident in any of it.
The popularity of the democratic and republican parties in the US are very nearly in perfect balance.
There is approximately a 50/50 split, swaying only slightly back and forth.
This is a remarkable fact.
Why is it the case?
I don't know.
But it seems that there must exist some homeostatic process which maintains this balance.
The idea that it happens by simple coincidence is obviously absurd.
The conditions under which this process operates are unclear to me, but I will try to analyze it regardless.
I will call it "electostasis".
What might one infer from this?
Whatever this process is, it is clearly dominating the political landscape.
Major imbalances are suppressed, returning the population to some "equilibrium" of symmetric binary division.
(To call such a polarized state an equilibrium might sound absurd, but it follows from the postulating of electostasis.)
Presidential elections are majority votes (or rather, let's pretend they are).
A perfectly balanced population does not produce a majority.
What a majority vote measures, then, is the sign (as in +/-) of the offset from equilibrium.
But offsets are unstable, since they will (apparently) tend to be reduced.
My conclusion is that the result of the election is a measure of whatever chaotic background noise happened to rise above the homeostatic floor at the time of the election,
the random fluctuations which had at that moment not yet been squashed down.
My assertion that this background noise is effectively random is not something I've justified.
Of course, I'm sure one can (retroactively) trace some lines of causality to make the result appear predictable, and I'm sure there exists cases where it really is predictable.
Currently, or usually, however, society seems to be much too complex for such a thing to be reliably predicted.
The number of factors is overwhelming.
Thus, we are forced to consider it as "background noise", as it is effectively random, compared to the apparent reliability of electostasis.
The effect of electostasis on the election, then, is to reduce its predictability, or in other words, to increase its complexity.
Maybe my choice of name was not very good, as "electostasis" might imply stable election results are maintained, which is the opposite of my claim...
Call it "electocomplexification" if you'd like.