Rule 32 of 36 · Chapter V — Limits & Uncertainty
Entropy sets the arrow of time
Why this rule exists
Isolated systems evolve toward more probable configurations simply because there are overwhelmingly more ways to be disordered than ordered. Entropy counts those ways, so it almost never decreases, and this statistical asymmetry — not the underlying mechanics, which run the same forwards and backwards — is what distinguishes past from future. The arrow of time is a matter of counting.
In practice
Use the second law as a filter: any proposed process that lowers the total entropy of an isolated system is forbidden. To find the equilibrium state, maximize entropy subject to constraints. When a subsystem's entropy drops, look for a larger increase elsewhere — a refrigerator cools its interior only by dumping more entropy outside.
When it doesn't apply
The law is statistical, not absolute: small systems can fluctuate to lower entropy briefly. It applies to closed systems taken as a whole; open systems can order locally by exporting entropy. It says nothing about rate, only direction.