Hacker News Reader

Comment by dredmorbius | Hacker News Reader

Most clocks that we can think of involve a clear entropic flow as part of the intrinsic mechanism: an unwinding spring, falling sand, a descending weight, a discharging battery.

But the very first clocks don't rely on this at least not as a principle mechanism. Sundials, early astronomical observatories such as Stonehenge, or simply observing transits and occlusions, don't involve an entropic process. Rather, they track the regular and sustained rotational or orbital movement of the Earth, Moon, and planets. They track the passage of periods of time, but that tracking itself doesn't necessitate entropic flow.

(The light that's necessary to observe the movements, of course, does. But that would be part of any time-based observation, and isn't fundamental to the regular movement itself being measured.)

View on HN

> But the very first clocks don't rely on this at least not as a principle mechanism. Sundials, early astronomical observatories such as Stonehenge, or simply observing transits and occlusions, don't involve an entropic process. Rather, they track the regular and sustained rotational or orbital movement of the Earth, Moon, and planets. They track the passage of periods of time, but that tracking itself doesn't necessitate entropic flow.

The clock here isn't the sundial but the movement of Earth (the sundial or the astronomy observatory is effectively a measurement device). If you see the movement of Earth as a whole, it's clear it's losing energy to heat, like all clocks.

Something else that can be said is that clocks are computing devices; and computers are thermal machines too.

Is the rotation (or orbit) of the Earth a driven process, or one based on momentum alone? What's the thermal process driving the rotation or orbit?

That is: as the Earth rotates or orbits, is it going through some cyclical energy potential flux of any significant magnitude?

I'd maintain that it's not.

The rotation of an astronomical body, or its orbit, is reversible in ways that thermodynamic events are not. If you were to observe a single body spinning in space, or two bodies orbiting one another, you could not tell from a simple observation of the movement whether you were seeing the motion played forward or backward.

(With sufficient numbers of bodies, or a sufficiently long period of time, this would eventually be disernable. But not at a macro level.)

For other thermodynamic clocks, this seems in general not to be true. Not only does a clock measure time and do so using a thermodynamic gradient, but we can tell which way is forward in time by casually observing the clock itself. Weights fall, springs unwind, batteries discharge, chemicals dissipage or entropise, etc., etc.

That is, an orbit is unlike falling sand, a discharging electric cell, an excited atom, or an unwinding spring. It's not even a periodic exchange of potential for kinetic energy as with a pendulum.

Tidal forces slow the rotation down over time, even in space. Also, the sun looses weight as it emits energy.

A black dwarf and dead planet, tidally locked, would still describe an orbit. So long as that functions, any means of being able to signal the positions of the two objects would serve to indicate the orbital period.

Solar emissions are not integrally central to the clock. That source of entropy can be eliminated.

Tidal losses are not integrally central to the clock, that source of entropy can be eliminated.

What are we left with?

Your argument only works in a universe with exactly two newtonian point masses, not a single additional mass. In that universe, time isn't the same as in ours. It is perfectly periodic. That you can't figure out the arrow of time in that universe is actually the smaller anomaly.

Yes: Thought experiments simplify.

That is the point.

My point is, that this is so simplified, that it isn't even a clock anymore.

Just like a spring is first loaded, then released, and the act of loading the spring is driving the whole mechanism, the Earth spin was set up by the primordial configuration of the solar system.

As the Earth rotates, its molten core gets continually mixed (as it doesn't rotate at the same rate as the crust[0]) and this generates heat. Likewise, the atmosphere doesn't rotate at the same speed as the crust, and this generates heat too. And, well, since perfectly rigid bodies can't exist[1], rotating bodies will always release heat, even those who are solid [2]. On the interaction of many bodies, tidal forces generate heat too, again because the bodies aren't rigid.

Those are all dissipative forces. I think it's not possible to have an orbital system composed of purely reversible processes. In this case, any orbital system counts as a clock in the sense of the article.

[0] https://en.wikipedia.org/wiki/Inner_core_super-rotation

[1] due to relativity, https://physics.stackexchange.com/a/201499

[2] actually.. I'm not so sure. I was searching a bit and https://phys.libretexts.org/Bookshelves/Classical_Mechanics/... says the non-rigid body eventually settles into rotating around its principal axis of rotation, which makes it not have any internal stresses, and thus not release heat. I think Earth revolves around its principal axis of rotation though (it had billions of years to settle).

At this point I'm trying to read about orbital mechanics of non-rigid bodies, like this https://impa.br/wp-content/uploads/2017/04/30CBM_03.pdf - the only thing I can gather is that indeed there are dissipative forces even in solids.

[T]he Earth['s] spin was set up by the primordial configuration of the solar system.

Valid point, and a possible out.

Your other suggestions could be eliminated with the tidally-locked black darf / dead planet thought experiment. Here you'd have bodies about as inert as we can conceive of bodies being, keeping a regular time.

But on such a body sundial would not work.

This is mentioned in the article, it isn't a contradiction. You need both periodic phenomena and the generation of entropy for a clock to work.

Consider what would happen if we used the earth's rotation as a clock, but without generating entropy. We would have a system which would go back to its original state every day, each rotation. This means there would be no way to discern how many days have passed, as the system would be in the same configuration as it was at the beginning of the rotation.

In order to encode how many days have passed, you need to increase the system's information, and by doing so you increase its entropy. The same problem applies to a pendulum clock: you can't have a pendulum clock as anything but an entropy-generating device, or else it will keep going back to its initial state without any changes, not measuring any passage of time at all.

In the pendulum clock, the mainspring provides the entropy flow that allows it to time-keep and register the passing of time on the indicator. In the sundial, the flow of light from the sun into the sundial and into our eyes provides the flow. The light flow gets interrupted by the gnomon, casting a shadow (an irreversible process). In order to register that this has happened, you look at the sundial and take note of the time (another irreversible process). You can replace yourself with light-sensitive paper, which gets etched by the light (yet again another irreversible process).

The other sources of entropy (like tidal friction, as well as atmospheric friction and friction between the layers inside the Earth) are irrelevant for the sundial's working, but could possibly be used to create timekeeping devices themselves.

I'd say the main difference between the Sundial and the Pendulum clock's working is that in the pendulum clock, the energy to keep the pendulum oscillating and to register the passage of time are both provided by the same source (the mainspring), while for the Sundial there are two different sources of energy: sunlight provides the energy to register the passage of time, while the Earth's inertia is the only source of energy that keeps it rotating.

We could make the two systems more similar by changing the pendulum clock so it's only a pendulum, creating a separate indicator consisting of a light shining onto a pendulum and a sheet of photosensitive paper slowly moving down and receiving the pendulum's shadow, registering the passage of time onto the paper without using the pendulum's energy for that.

We would have a system which would go back to its original state every day, each rotation. This means there would be no way to discern how many days have passed, as the system would be in the same configuration as it was at the beginning of the rotation.

That's precisely what bodies in angular rotation or orbits do.

"One day looks pretty much like another."

For a sundial, the (apparent) movement of the shadow provides the marker of time throughout the day. It (mostly) doesn't indicate the passage of time greater than a day (not strictly true: see the annalema). But within any 24 hour period, you've got a timekeeping piece in which entropy plays no role in the measurement of the passage of time itself, but only as a signalling mechanism (casting a shadow).

In order to complete the clocked-ness, you either need a second repeating phenomenon (e.g., Earth's rotation vs. Moon's orbit, Earth's rotation vs. Earth's orbit). In which case with an external recording mechanisms the passage of time can be noted.

But the fundamental timekeeping mechaism, the phenomenon which is regularly occurring with time occurs without discernable entropic decay, once set in motion.

The rotation or orbit is the escapement here. It's functioning without an entropic drive.

Pointing out that an entropic flow might be used independently of a sundial does not explain how it relates to the sundial.

A sundial uses more than just angular rotation and orbits, though. You can't measure time with angular rotation alone. Using a sundial as a clock (by registering its cycles or by looking at it and remembering how many cycles it went through) inevitably increases the system's entropy.

You don't need a second repeating phenomenon, just as you don't need a second pendulum in the pendulum clock. The pendulum might repeat its motion at a short interval, but is able to measure intervals longer than its period.

If you have a sundial that works without increasing entropy, it doesn't work as a clock. Just as a perpetual motion pendulum alone wouldn't work as a clock. Using other orbits doesn't change that. You need the entropic flow to register the passage of time. In the case of the sundial, it can be done by recording its cycles.

A sundial on a rotating planet isn't able to measure time within its cycle without increasing entropy any more than a pendulum is able to measure time within its period without increasing entropy. They're not able to do that. A Sundial without light and a recording device isn't a clock, just as a lone pendulum with no energy losses and no recording device isn't a clock.

https://news.ycombinator.com/item?id=28417247

Reframing the analysis helps show that your thoughts still fit within the proposed model:

Sundials and astronomical observations are not principle mechanisms of clocks.. they are a window of observation into the clock system of moving planetary bodies. Those planetary bodies indeed follow the thermodynamic principles outlined.

Imagine you had a rotating disc occluding a light source with a thin slot cut in the disc such that every rotation at one small interval the light source is visible. Rotate it using, say, a coiled spring, and you have a clock, revealing a pulse of light with a constant frequency. This is easy enough to fit into the mold of a thermodynamic clock as proposed in the article.

Now, we can think of earth orbiting the sun similarly to our disc rotating in front of the light source…. Every time earth rotates on its axis, half is exposed to the light and half is not…. Quite a wide ‘slot’ cut into our planetary ‘disc’, not useful for much more than knowing what day it is.

All a sundial does is narrow that ‘slot’ to make a more accurate observation of our planetary clock’s state.

Same goes for observing other planets / stars etc. While these planetary motions appear to be ‘regular and sustained’ on our timescales, they are still subject to the same thermodynamic principles as all other matter, and reach lower entropy over time, until the heat death of the universe :(

Someone posted about two bodies orbiting each other, and not know time forward from backward.. again, you can’t ignore physics, those orbits will measurably and irreversibly decay over time until the bodies collide.. just not on timescales that would effect our species using them as clocks!

> Sundials, early astronomical observatories such as Stonehenge, or simply observing transits and occlusions, don't involve an entropic process.

How is the nuclear fusion happening in the Sun and other stars producing rays of light not an entropic process?

How is that fusion relevant to the orbital motion of bodies?

As I've suggested elsewhere in the thread: imagine a black dwarf and dead planet in orbit.

Clock or no clock?

Entropic or not entropic?

> How is that fusion relevant to the orbital motion of bodies?

Because without fusion we would not observe them.

See my original comment.

That's a common element to any observation-based system.

Or, put another way: if a dead planet orbits a black dwarf whilst tidally locked in an utterly dark Universe, does time exist?

Is the observer necessary?

Self-reply answering a number of points:

- I'm familiar with Eddington's "arrow of time" and the concept's fascinated me since I first encountered it listening to a lecture by Stephen Hawking, and reading his own A Brief History of Time. (The lecture was effectively a chapter from the book.)

- The question of whether or not all clocks are entropic has bothered me ever since.

- It's relatively recent delvings into origins of technology, and recognising that astronomy was in large part the science of time, as well as the fact that most of our historical measures of time were based on astronomical phenomena (the day, month, and year, as well as multiples and subdivisions --- reliance on nonastronomical standards is a modern departure largely undertaken for accuracy but not basis).

- In the Seven Liberal Arts, the trivium concerns data input/output and processing (grammer: parsing input, logic: processing, rhetoric: output), whilst the quadrivum is quantitative: arithmatic is quantity, geometry is quantity in space, music is quantity in time, and astronomy is quantity in space and time. That adds an additional dimension to astronomy, but again, the role of ancient astronomers in timekeeping, prophecy, and navigation remains significant.

Reducing my argument above to its essence, I like the formulation I arrived at in one subthread here, a variant on the "if a tree falls in a forest" conundrum:

If a dead planet orbits a black dwarf whilst tidally locked in an utterly dark Universe, does time exist?

Is the observer necessary?

If the system returns back to the exact same configuration (highly improbable because there are other forces besides gravity), I think we could say that time makes a loop, because any observer can't distinguish whether it's currently in time t or time t+P where P is the period.

Incidentally, I've been thinking about a time travel game along those lines: if you ever return to a save point but the game is at a previous configuration, this counts as a time travel (which can be, like.. a multiverse travel if time branched out.. which amounts to the same thing), just like if you selected a time travel in the UI.. two game states can't be in a different time if they are identical.

Of course if the mechanic is just this, then this is just a regular save/load state with infinite save lots. Which is a funny observation. Then I need to add other mechanics on top, like carrying things from a timeline to another, which is where the physical analogy would break up.

Another observation, if we had two different timelines that end up in the same exact state, then this could be an explanation on why time isn't reversible: when we go back in time there's multiple possible pasts, and by looking at the present state of the world we don't know which timeline is the real past (if there's ever such a thing).

The many-worlds quantum interpretation says that, likewise, when we go to the _future_ we also have multiple timelines, and nature somehow just pick one arbitrarily, and this becomes "our" timeline. Then to reverse time, perhaps it's sufficient to employ the same mechanism, but in reverse. (Perhaps there's already beings that live backwards in time, it's just that "their" timeline works that way. Like the movie Tenet). Or even, perhaps each time the timeline splits in the future direction due to quantum phenomena, it also equally splits when going in the past, in the exact same way.

(Well I'm trying to recover some symmetry here, but not sure if it works)

well, the way to keep time with a sundial is to look at the shadow it casts. sun is the heat source here and the dial itself heats up in the process of casting a shadow.

also note that if there's no sun (no energy), time keeping functionality is severely limited.

What then would be the minimal requirements for a timekeeping function?

to tell time, i guess.

The light is fundamental... without particles bouncing between planetary bodies there would be nothing keeping them in sync.

Thought experiment: imagine a box through which 0 information passes; no light, heat, radio waves nothing. There’s a clock in the box. Does the clocks progress reflect the progress of time outside the box?

No, because time is based on causality and in space that’s propagated largely by massless particles.

Light is not keeping the Earth and Moon "in sync". Gravitational force is.[1] That force propogates at the speed of light. It isn't light itself.

Light does play a role in illuminating the system, rendering it observable. It's a signalling component. But again, that's exogenous to the system itself. (See discussion of Maxwell's Daemon and the resolution of that paradox for the role of light in that specific thermodynamic process.)

But, and this is key: ANY clock needs some way of passing information on to the observer. From TFA, that might be light, or sound, or chemical potential (received as the smell of rotting garbage). But that is a common element, part of the signalling mechanism, but not of the timekeeping mechanism itself.

For an astronomical clock, in most cases, the regular rotational or orbital motion is the timekeeping mechanism. The signalling mechanism is entirely separate.

E.g., in the Earth-Moon system, timekeeping is noted by both the daily rotation of the Earth, and the monthly orbit of the Moon. The kinetic energy of both those regular motions is not driven in any sense, nor is it measuring a decay.[2]

The signalling comes from an exogenous light source (the Sun) which illuminates both bodies. But the Sun isn't powering the timekeeping mechanism in question. And is equally significant in providing the light necessary for you to read a wristwatch or wall clock (of a non-illuminated variety).

________________________________

Notes:

1. I'm going to leave discussion of just what gravity is and/or how it's manifested out of this discussion. Veritassium tackled this recently though: https://youtube.com/watch?v=XRr1kaXKBsU

2. Yes, there's tidal friction. That's removing energy from the system, and is measurable with sufficiently precise clocks and over time. But that's also incidental and secondary to the motions in quesiton. We could use far more remote bodies (e.g., the Earth and another planet) in which mutual tidal forces are minimal, to eliminate this consideration. Or presume tidally-locked bodies in which orbital / rotational kinetic energy and angular momentum are constant.

Reply on news.ycombinator.com