Video: why flipping a coin is completely deterministic


Originally published at:


That bell curve generator gizmo with the tiny balls is really neat!


Hey Google, Ask the Action Lab to bleep out action words.


Now you’re just posting Google ads. I feel so alone.


Meh. I think it needed a bit more google.


He says something toward the end like “Scientists have found we’re 93% predictable in almost every aspect of our life.” Wtf does that even mean?

Also, fuck google assistant.


This would seem to imply that entire universe of events can be divided into random (“unpredictable”) events and chaotic events (unpredictable just because we can’t have the information to predict them) and maybe predictable events? But if you make that division based on information then you invite in crazy ol’ Ms Quantum (mechanics) who will quickly tell you that information shall never be available (at some level) so all events are chaotic according to this definition. Why even a coin cannot be “completely deterministic” because that atom on the end of the Queen’s (or Lincoln’s) nose is has many different energies until sometime after someone ‘calls it’.


He’s a little off on his explanation of chaos. You cannot predict the future behavior of the double pendulum based on its initial condition, it will be different each time if it is a true chaotic system. If you are constantly gathering all the information about its movement, you could predict what it’s going to do next, but not what it’s going to be doing 30 seconds later. This phenomenon is why we can’t predict the weather very far into the future.


No, this is wrong. The “chaos” is not a stochastic property, rather it is that the deterministic system is ergodic. What “ergodic” (roughly) means is that if you consider all those initial conditions (position, velocity) which are nearby the one you are at and let them evolve, they will spread out and eventually at least one of them will get close to every other configuration.

In practice, you do not really know the exact initial conditions. You probably know them up to some tolerance, so it’s like your true initial condition is something close to – but not exactly equal to – the one you write down. So really, if you repeat the experiment a number of times, you are sampling from a distribution of initial conditions which are close to your reported value. Since the system is ergodic, those points can end up just about anywhere – without any randomness in their time evolution.


Nine comments in and nobody has a Rosencrantz and Guildenstern Are Dead joke?


I’m not sure. Our best model of reality (as far as I understand it) has space and time as discreet, but still assumes probabilities as continuous. What if the dice the universe rolls to determine the position of a particle only have 10^34 sides? Would even one of those make the position of a single random event affect the outcome of the coin flip? Hell, maybe the universe uses a pseudo-random sequence.


I wish they (this video, and anyone trying to teach about probability or uncertainty) made clearer that there are many different sources of uncertainty, and when we talk about unpredictability and probability we usually don’t distinguish carefully (logical? epistemic? indexical? Knightian?). The big differences are perspective and knowledge. Oh, and a large helping of caring more about whether the deterministic/random/chaotic distinction is useful than about whether it is true in some vaguely-defined absolute sense. If you insist on reasoning from the perspective of physical determinism, it’s really hard to even frame what the concept of “could” could possibly mean.

Laplace’s Demon, sitting outside space and time, can honestly say that the universe’s entropy is zero, and never increases. I, sitting here inside the universe, had better get used to the laws of thermodynamics being important.

I read a story once, not sure if true, that claimed that when Von Neumann was told the weather was chaotic and therefore unpredictable, he responded, “That’s not unpredictability, it’s control.” We’re not there yet with weather, but chaotic control is really useful for systems where we understand the attractors well enough.


Without going into a lot of detail about microstates and macrostates, the simple explanation is that macroscopic objects like coins generally don’t exhibit quantum properties, and can, in practice, be deterministic. The energy of an atom on Lincoln’s nose isn’t deterministic, but it also doesn’t determine whether the coin lands heads or tails.


That’s why I always ask the Magic 8 Ball first whether I should flip a coin or not.


My quantum professor would emphatically disagree with you. “A superposition of wavefunctions is still a wavefunction even if it can’t be calculated with a universe-sized computer”. Of course, a professor of quantum mechanics is required (by tenure) to view the entire universe quantum mechanically :wink:





That is completely true, and doesn’t change the fact that most macroscopic objects behave classically, because, beyond a certain point, the quantum fluctuations of individual "particles " cancel each other out. While it’s possible for that not to be the case, the odds of ever observing a case where it isn’t in this universe are effectively zero for normal, fermionic, room temperature matter like a coin.

Quantum is fun, but statistical physics is a lot more relevant for macroscopic ensembles.


This sort of assertion always bugs me. “it’s not random because if you knew all of the initial conditions, you could predict the outcome” may well be true, but you don’t, and there’s a very wide gulf between being able to do so in theory and actually being able to do so in practice.

(Good, Fair) Coin flips are effectively random. Anyone who tells you different is being too clever by half.


I may have erred semantically here, and I am making the assumption that a double pendulum moves “chaotically” since he used it as an example. Knowing that the point at the end of a double pendulum will always eventually trace out a similar pattern is not the same as being able to predict the position of the moving pendulum in the future based on information available in the present. You seem to be suggesting that the particular gyrations the pendulum undergoes, which can’t be predicted accurately very far in the future at all, are trivial. I’m not so sure about that.

He is equating randomness with unpredictability, but then treating certain kinds of unpredictability as trivial, and then asserting that radioactive decay is random. I would imagine that radioactive decay also happens in some sort of fractal pattern that can’t be easily grokked. Imagine if you didn’t have the luxury of understanding the construction of the pendulum and being able to monitor the movement of a specific point on its structure. Radioactive decay might just be an instance of this unknown structure unexpectedly changing direction, so to speak.

I’m not an expert on this subject obviously, but I have been previously exposed to it. The universe would not work if it was “random”, but it does not appear to work quite like a coin flip indoors onto a flat surface.