Week 4: How does a crowd have properties its members don't?
You cannot predict where one gas molecule will be a second from now. Give a physicist a litre of the gas and they will tell you its pressure to several decimal places.
17 August 2026
Watch one molecule bouncing around a box and try to say where it will be a second from now. Classical mechanics says this is possible in principle: know its starting position and velocity exactly, know the outcome of every collision it is about to have, and its path is fixed, no dice involved anywhere. But try to actually do it. It collides, changes direction, speeds up, slows down, and the smallest error in your knowledge of its starting position gets amplified every time it hits something. Pinning down its future would take a precision in your starting knowledge, and a precision in tracking every collision, that no real measurement can supply. One molecule’s future is, in practice, unpredictable — not because the physics is undetermined, but because following it is beyond reach.
Now fill the box with a litre of the same gas and ask a different question: what is the pressure on the walls?
That question has an answer, and a good instrument will give it to you with startling precision. Measure it again a minute later and it has barely moved. Measure it in an identical box across the room and you get the same number. Nothing about any individual molecule got easier to predict. There are more of them now, colliding more often, in directions still just as hard to call in advance.
This looks backwards. Uncertainty about the parts should make you more uncertain about the whole, not less. Instead the opposite happens. As the numbers climb into the billions and beyond, individual unpredictability turns into collective reliability. Pressure settles down. Temperature settles down. The wild swings of any one molecule get buried under the combined effect of everything else in the box doing its own unpredictable thing at the same time, and what survives on the far side is a small set of numbers that barely move.
That is not an accident of gases. The same statistical idea that appears in the law of large numbers is at work. Flip one coin and you cannot say whether it lands heads. Flip a billion and the fraction of heads will sit extremely close to one half, every time, even though no individual flip became more predictable. Pressure and temperature are doing something similar for a box of gas, except the underlying trials are not coin flips — they are the positions, speeds, and collisions of an enormous number of molecules, each one following ordinary mechanics.
So a crowd’s properties are not just a scaled-up version of one member’s properties, seen from a distance. They are a different kind of object. Pressure is not “what one molecule does, times a very large number.” It is a statement about the shape of an entire distribution, and that shape can be far steadier than anything happening inside it.
Last week’s question asked how a property like temperature could exist for a crowd of molecules and not for any one of them. This week sharpens that into something more useful: statistics is not a way of coping with the fact that you cannot track every molecule. It is doing real physical work. Large numbers do not just hide our ignorance. They actively produce stable, reproducible behaviour out of pieces that are, individually, anything but stable or reproducible.
Once you accept that, a strange experiment becomes available. Film the box of gas from the moment someone releases a burst of it in one corner. Watch it spread evenly through the whole volume, the way gas always does. Now run the film backwards.
Every collision in that reversed film still obeys the same laws of motion the forward film obeyed. Nothing about the underlying mechanics forbids it. And yet the reversed film looks obviously wrong. Gas does not spontaneously gather itself back into one corner. You would call it fake the instant you saw it.
But if the microscopic rules don’t care which way the film runs, where does that obvious wrongness come from?
What we explained#
A crowd can have properties, like pressure, that are far more predictable than the behaviour of any single member. Statistics is not merely a tool for dealing with the physicist’s ignorance about individual molecules; it is doing real physical work, turning unpredictable individual motion into stable, reproducible collective behaviour once the numbers get large enough.
What became stranger#
Why does time run one way?