[oldsecondhand]

I think he is confusing the usage of entropy in physics and computer science. In computer science entropy is conditional probability and depends on what we know about a system.

[kgwgk]

As it does in physics!

"which parameters [thermodynamic variables] you know" ~ "which parameters [thermodynamic variables] you hold fixed"

(or know in average, like the energy for a system in a heat bath where the temperature is fixed)

https://bayes.wustl.edu/etj/articles/theory.1.pdf

http://nicf.net/articles/thermodynamics-statistical-mechanic...

[oldsecondhand]

You can observe the movement of molecules beyond macro properties like temperature.

[kgwgk]

Sure, at least in principle. And if you knew what every molecule was doing the entropy would vanish.

[IX-103]

I've thought that for a while, but I'm not a physicist. Do you know any prominent physicists that hold that view? It seems to contradict at least the popular narrative about entropy as a property of a system.

[kgwgk]

> Do you know any prominent physicists that hold that view?

The view that the entropy of a microstate (i.e. a perfectly defined physical state) is zero?

All of them, hopefully.

[liminvorous]

I only have an undergraduate degree in physics, but I think the point you may be missing is that information requires some physical medium to store it.

So entropy is very roughly speaking the property of the system that determines how big a hard drive you need to store a description of that system.