Show someone footage of a glass falling and shattering, and they instantly know which way time is running. Play the same footage backward — shards leaping up off the floor and reassembling into an intact glass — and it reads as obviously wrong, even though nothing in the basic laws of motion governing each individual shard's collision technically forbids it. The reason lies in a single quantity: entropy, and the second law of thermodynamics, which says the total entropy of an isolated system never decreases over time.
Why "never decreases" comes from counting, not force
Entropy is often loosely described as "disorder," which is a decent shorthand for a more precise, statistical idea: entropy measures how many different microscopic arrangements of a system correspond to the same macroscopic description. An intact glass has relatively few arrangements of its atoms that count as "intact glass." Shattered glass scattered across a floor has an astronomically larger number of arrangements that all count as "shattered." Nothing forces the shards toward disorder — it's simply that, among all the ways things could randomly evolve, the "more arrangements" outcome is vastly, overwhelmingly more likely to be the one that actually happens.
The arrow of time is really a statistical bet, not a law of motion
The fundamental laws governing individual particle collisions are, famously, time-symmetric — run them backward and they're still valid physics. What makes some processes visibly one-directional isn't the underlying physics of any single collision; it's that starting from an ordered state, like an intact glass, and reaching a similarly ordered state again by chance, after it's shattered, is so staggeringly improbable that it essentially never happens within the age of the universe, even though it isn't technically forbidden.
What we're still unsure about
The statistical explanation accounts for why entropy tends to increase going forward from any given starting point, but it doesn't by itself explain why the universe apparently started in an extremely low-entropy state in the first place — a state that, on the same statistical logic, should itself have been staggeringly improbable. Physicists and cosmologists still don't have a fully settled explanation for why the early universe had such low entropy, which is one of the genuinely open problems connecting thermodynamics to cosmology, not a solved footnote to the second law.
This sits inside Second Law & Entropy, one of seven topics in Thermodynamics, one of five domains in Physics, one of seventeen subjects the app can quiz you on.