Hacker News Reader

Comment by oezi | Hacker News Reader

> On the macroscopic average, entropy can not decrease, only increase.

The reason or physical principle from which this law of thermodynamics originats has always escaped me. Why can't there be a way to decrease entropy?

View on HN

Because there's no way to convert energy from heat to mechanical, electrical, or chemical but they inevitably lose some of themselves to heat energy (through resistance, friction, and exothermic reactions).

Note that you can use a difference in temperatures to generate mechanical/electrical energy, but that's also an increase in entropy as the nicely sorted temperatures get muddled and weakened, and there's no way to undo that either (at least not without suffering a loss of efficiency).

Also note that you can always "cheat" by expanding your reference frame to include objects and energy you don't care about. That's why the first law of thermodynamics states that you can't get energy for free, but we can, for all intents and purposes, get "unlimited" "free" energy from the sun, and that a solar-powered machine could have what is effectively perpetual motion from our perspective. I don't know about you, but I'm not going to worry about the machine failing in five billion years when the sun dies.

In QM information can be created (by measurements), but not destroyed* (no-deletion/no-cloning theorem). To "store" that information you need to have enough possible physical quantum states (the kind bounded in Bekenstein bound) which is thermodynamic entropy.

[*] what made blackhole information paradox controversial

I think it originates from the observation that all machines "leak" power in the process of performing work. Vibrations, thermal conduction, infrared radiation, etc all tend to decrease efficiency and hence increase entropy (in the form of heat loss, mostly). Entropy can of course be lowered by bringing a measure of order to a given "isolated" system. At much larger scales however the vast number of possible microstates dominate, and thus the "inevitable march toward disorder". Also bear in mind that "entropy" itself is a fairly general concept which can be applied to just about any field. Information, for example, can expressed in terms of entropy. (Which obviously has nothing to do with the "thermal entropy" of physics.)

> Why can't there be a way to decrease entropy?

It is like asking "why can't things go faster than the speed of light", this is an observation of all known processes, not an absolute law. It is possible there are things we haven't observed that will change this.

Entropy can decrease, it's just statistically unlikely. It's a "law" due to the law of large numbers, not because the transitions are forbidden.

Another way to phrase this is that there's no law of nature the says your pool balls can't end up neatly racked after the break, but you'd still never expect to see that happen.

It is possible to decrease entropy locally, but not globally. You can only push entropy somewhere else, and that pushing creates even more entropy.

This is why refrigerators keep food cold but heat up the room, and why it's impossible to "air-condition the outdoors"

My layperson's understanding of entropy is energy that cannot be used. If something cannot be used, it also cannot be made smaller.

Reply on news.ycombinator.com