Tripping the Light fantastically
August 24, 2021 1:39 PM   Subscribe

No One Has Ever Measured the One Way Speed of Light. (Veritasium video)... in fact the one-way speed of light is not just unknown, it is undefined. Crazily enough, it is possible that light travels at 1/2c in one direction, and instantaneously in the reverse direction.
posted by storybored (85 comments total) 24 users marked this as a favorite
 
Can't recommend this channel enough. The videos cover science topics so much better than any program I grew up with on TV and certainly better than any of the programs now shown on networks that once aired halfway-decent educational content -- Veritasium is full-blown decent.
posted by GoblinHoney at 1:53 PM on August 24, 2021 [6 favorites]


Crazily enough, it is possible that light travels at c in one direction, and instantaneously in the reverse direction

Wouldn't that be 1/2c in one direction...?

EDIT: You know what? This is making my brain hurt. Comment retracted.
posted by mrnutty at 2:02 PM on August 24, 2021 [2 favorites]


So they are saying the suns and the novas are omnidirectional? Then the sun's light goes back into the sun as well? Or, are they describing the real interconnectness of all things?
posted by Oyéah at 2:21 PM on August 24, 2021


Wouldn't that be 1/2c in one direction...?
You are correct, and in the video he says ½ c.
posted by mbrubeck at 2:25 PM on August 24, 2021 [1 favorite]


Surely this is a job for quantum entanglement, or something else I don't really know anything about? Even defining "direction" is a problem in this context, isn't it? Does the event horizon of a black hole depend on which "direction" light is trying to leave it?
posted by maxwelton at 2:26 PM on August 24, 2021


Michelson-Morley proved that light travels at the same speed in all directions.

Additionally if the speed of light varied, we'd see different results of double-slit diffraction vary depending on orientation, which we don't see. Also we can indirectly measure the one-way speed of light by calculating it based on diffraction patterns.

The proposal to measure the speed of light through a huge spool of wire or fiber optic cable is correct as, again, Michael-Morley proves that the speed of light is the same in all directions. There's no biasing.

Finally the proposal that the speed of light is different in different directions but that we can't detect it because everything happens relative to c is interesting but irrelevant. Even relativity doesn't care about hidden variables and ocne you're into string theory with hidden variables, well, whatever, you've gone way off topic.
posted by GuyZero at 2:38 PM on August 24, 2021 [12 favorites]


The Mars experiment he goes into is up there with "do we all see blue the same way?"

We can never know and it also never matters.
posted by GuyZero at 2:40 PM on August 24, 2021 [3 favorites]


The more I watch this video the more it needs a fat spliff and a bag of doritos. This isn't physics.
posted by GuyZero at 2:41 PM on August 24, 2021 [7 favorites]


Crazily enough, it is possible that light travels at c in one direction, and instantaneously in the reverse direction.

So the issue here is that there's no such thing as a "reverse direction". Light always moves forwards. The notion of it moving towards or away from something is purely something created by an observer. The photon doesn't know. In many cases what you think of as a "reflection" is, in fact, an absorption and reemission and the photons, in so far as we can give photons an identity, aren't even the same photons. Thus they can never even know if they were reflected or whether they were emitted.

This whole train of thought is based on the notion that "reflection" has some special property that doesn't actually exist.
posted by GuyZero at 2:45 PM on August 24, 2021 [7 favorites]


What if light just is and we look at it, along it, watch it go? (Tiptoeing away from this thread. Bye now.)
posted by Oyéah at 2:51 PM on August 24, 2021 [2 favorites]


This is an interesting idea but it seems to depend on a two dimensional approach. GPS uses four dimensions to enable a receiver to work out its position in not only space but time - doesn't that solve the clock synchronization problem? As put here https://www.eso.org/public/outreach/eduoff/seaspace/docs/navigation/navgps/navgps-3.html


The trick lies in the fact that the time offset of the clock in the GPS receiver is considered as the fourth unknown (the first three being the three space coordinates of the receiver). In the first approximation, the offset is considered to be zero. Then, if a fourth satellite signal is received and a fourth distance is measured, it will also be possible to determine with high precision this time offset and then to find the correct space coordinates. Said in other words, the four distances to the four satellites will only fit and determine one particular point in space, if the time offset has a certain value. This calculation is done automatically by the software in the GPS receiver.


posted by jamespake at 2:57 PM on August 24, 2021 [2 favorites]


Sorry, couldn't resist: https://www.youtube.com/watch?v=RbTUTNenvCY

posted by jamespake at 3:16 PM on August 24, 2021 [2 favorites]


Yeah, how does the light know whether its a reflection or not? But maybe it does.
posted by dmh at 3:29 PM on August 24, 2021 [2 favorites]




what about quasars?
posted by clavdivs at 3:37 PM on August 24, 2021


Also: "From the perspective of a photon, there is no such thing as time. It's emitted, and might exist for hundreds of trillions of years, but for the photon, there's zero time elapsed between when it's emitted and when it's absorbed again." from https://phys.org/news/2014-05-does-light-experience-time.html
posted by jamespake at 3:42 PM on August 24, 2021 [3 favorites]


I haven't watched the video yet but did skim through the wikipedia article which is helpful, and I think there's a bit of confusion about directions and reflections here. So the Michelson–Morley experiment is more like a four-way path for the photons but since they make a round trip and get measured where they're emitted it's a "two-way speed" experiment. Same with sending the photons down a fiberoptic cable that twists in all sorts of directions and they probably take a zillion-way path but it's always the round trip that's measured.

It sounds like this is more akin to the first half of the twin paradox where, if the astronaut twin never turns around and comes back to Earth there's no way to say who's older, if special relativity says there's no distinction between reference frames where the Earth twin was static and the other moving, or vice-versa. Interestingly enough I had been taught the role of acceleration was crucial to that but, on a quick doublecheck, apparently not? Physics really is weirder than you expect.
posted by traveler_ at 3:53 PM on August 24, 2021 [8 favorites]


Maxwell’s equations

Yeah, it's fun to smoke weed and get high but if you're really going to discuss the speed of light it's Maxwell's Equations or gtfo. Don't like doing integrals of field cross-products? Again, gtfo.

And again, the proposal that c could be non-constant but we can't tell because we're always dividing by c is irrelevant until you can come with an experiment to make it matter.
posted by GuyZero at 3:53 PM on August 24, 2021 [5 favorites]


Same with sending the photons down a fiberoptic cable that twists in all sorts of directions and they probably take a zillion-way path but it's always the round trip that's measured.

Nope. You can take a spool of a mile of fiber optic cable and stick both ends beside each other. It's one-way.
posted by GuyZero at 3:55 PM on August 24, 2021 [2 favorites]


mrnutty: Wouldn't that be 1/2c in one direction...?

Right you are! Good catch. I asked the mods to fix it.
posted by storybored at 3:57 PM on August 24, 2021


And remember that M-M doesn't measure speed. It measures phase difference.
posted by GuyZero at 3:58 PM on August 24, 2021 [1 favorite]


Mod note: I lent the post 1/2c of photons, carry on
posted by cortex (staff) at 4:03 PM on August 24, 2021 [12 favorites]


Nothing like a cuppa photons. Thanks neighbor :)
posted by storybored at 4:07 PM on August 24, 2021 [3 favorites]


I think this is might be an actual place to apply Occam's razor properly: If two theories make indistinguishable predictions, the simpler one is true.
posted by Zalzidrax at 4:12 PM on August 24, 2021 [5 favorites]


What if you're seeing one of those theories in a mirror, after it's been spit out of a mile-long fiber-optic cable?
posted by riverlife at 4:17 PM on August 24, 2021 [2 favorites]


Nope. You can take a spool of a mile of fiber optic cable and stick both ends beside each other. It's one-way.

Yeah having now watched the video it's clearer that the fiberoptic cable idea is meant to be one-way, and in fact has been carried out in some form in the "JPL study". The Veritasium guy describes the problem with that idea as having to do with the different internal directions the light takes in the coil but I'm not sure how well I understand that / how well that describes the problem? I wonder if it's that sending photons along any closed path can only rule out some, not all, anisotropies. Anyway the cited rebuttal to that experiment is a 296-page textbook that I don't have access to and probably couldn't understand anyway.
posted by traveler_ at 4:37 PM on August 24, 2021 [2 favorites]


I think the video guys makes one major mistake which is to conflate relative and absolute directions. He talks about the speed "back" from Mars but Mars and the Earth move and in absolute terms (such as they are, let's say relative to the Sun) the path is actually moving in a different direction. So he's making two, possibly unrelated claims, that a) c is different in different "absolute' directions and b) c is different relative to an observer

a) is disproven by M-M
b) is disproven by Maxwell's equations and/or relativity where the existence of an observer is irrelevant

Anyway apologies for making the same comment like 10 times. It's not just sloppy thinking, it's sloppy research as he just totally ignores a bunch of different ways people have tried to measure the speed of light.
posted by GuyZero at 4:43 PM on August 24, 2021 [3 favorites]


GuyZero, if this is just a mistake by Derek Muller then why is this a sufficiently well known philosophical issue with special relativity to have an extensive wikipedia page and an entry in the Stanford Encyclopedia of Philosophy?

TBH I thought the synchronization convention issue was a pretty well known feature of special relativity, pretty sure my physics prof covered it in undergrad along with the other counterinuitive issues like length contraction, so I'm surprised to see it getting such pushback here.
posted by Proofs and Refutations at 5:00 PM on August 24, 2021 [7 favorites]


I don't get if you have pictures of a super nova or some star exploding and the exploding bits are equal on both ends... doesn't that prove it or am I misunderstanding it?
posted by geoff. at 5:01 PM on August 24, 2021


Either the one-way speed of light is measurable in principle, or it is not. If you manage to measure it as not-c, congratulations, you've got new physics!

Otherwise, it's either physical law (if you measure it, and it's equal) or a convenient but correct bookkeeping trick (if, as it seems, it is impossible to measure even in principle) that informs us about how, in practice, we should set our mars clocks to simplify computations about the times of events on multiple planets.

All that said, I hope Veritasium's video makes people curious without creating net "physicists don't know anything" anti-science types. Being somewhat contrarian is part of his deal, and in this age I worry about contrarian celebrities who end up creating the wrong kind of doubts & unease about the edifice of science.
posted by the antecedent of that pronoun at 5:11 PM on August 24, 2021 [6 favorites]


if this is just a mistake by Derek Muller then why...

I'll be the first to admit that my very basic level of undergrad physics is pretty limited, so the answer may be me simply assuming these are rebuttals when they are not. I'll also be the first to say that getting worked up and being right are two different things.

I honestly don't see anything on this in the wikipedia entry for special relativity so if you have a more specific reference, I'd love to see it. personally I feel like mentioning M-M, which was high school physics for me, would be a pretty basic thing to do if it somehow isn't actually a rebuttal to what he's describing. It's kind of the whole point of a pretty famous experiment.
posted by GuyZero at 5:13 PM on August 24, 2021 [2 favorites]


I was referring to this Wikipedia page and this Stanford Encyclopedia of Philosophy page which are pretty good primers for the issue, which is mostly one of definitions.
posted by Proofs and Refutations at 5:18 PM on August 24, 2021 [2 favorites]


Yeah, ok, so reading through Conventionality of Simultaneity I get that this is a philosophical approach to simultaneity but I feel like his examples don't prove what he's trying to prove.

The example of communicating with Mars relies on one of the two assumptions of non-constancy he describes, both of which are disprovable IMO.

I guess the whole video boils down to a longer version of this:

Grünbaum (1973, 356) rejects this argument on the grounds that, since the equality of the one-way speeds of light is a convention, this choice does not simplify the postulational basis of the theory but only gives a symbolically simpler representation.

Yeah I guess. But we can prove that in free space light goes the same speed North as it does West. I guess I'm not clear that if c is really c(x) what x is. I'm not the guy who gets to draw the line between philosophy and physics is, but I feel like this got sold as physics in the video when it's not.

Anyway, thanks for the info and it makes a lot more sense what he's trying to describe.
posted by GuyZero at 5:26 PM on August 24, 2021 [2 favorites]


It's interesting because we have very good reasons to believe there is no anisotropy in the speed of light, but have yet to be able to prove it experimentally. I guess it's more about the speed of causality, though. After all, light itself isn't the only thing that propagates at the speed of light. Gravitational waves do as well.
posted by wierdo at 5:31 PM on August 24, 2021 [1 favorite]


And, again, apologies for perseverating and being half-wrong, the real takeaway is that you can't synchronize clocks rather than that we can't measure the one-way speed. Which, sure, one implies the other, but I think a lack of synchronization is a more important issue.
posted by GuyZero at 5:33 PM on August 24, 2021 [3 favorites]


I'm pretty sure light doesn't really have anything else to do. Goodnight, y'all.
posted by halfbuckaroo at 6:21 PM on August 24, 2021


quoth wikipedia:
Since 1983 the metre has been defined as the distance traveled by light in vacuum in 1⁄299,792,458 second.[7] This means that the speed of light can no longer be experimentally measured in SI units, but the length of a meter can be compared experimentally against some other standard of length.
which brings us to the possibility that per this definition + asymmetric speed of light, in one direction, the meter' is two meters long and in the other direction it is zero. But there's no experimental way to measure this fact.

By the way if anyone was having trouble figuring out what M-M is, it's the Michelson–Morley experiment, which wikipedia says in the one-way speed of light article only measures the two-way speed, as the light is returned to the observer; the same, so it would seem, for fiber optics experiments.
posted by the antecedent of that pronoun at 6:34 PM on August 24, 2021


GuyZero, the Michaelson-Morley experiment relies on reflecting light from the source to a pair of mirrors and from there to an eyepiece. The reflection means that the beams of light are reversing their direction, so it's a two-way measurement. Even if you arrange your mirrors so that at least one of the light beam takes a circuitous path, the sum of each beam's motion towards and away from the hypothetical "preferred" direction will be the same.
posted by Joe in Australia at 6:51 PM on August 24, 2021


this is kind of like the brain-in-a-vat thought experiment right? it's possible, but we'll never have any way of proving it, and it doesn't matter, at least to any of the models that have been come up to try to explain things so far, and we also don't really have any reason to suspect it is true.

i mean i do like the idea of a giant "this side up" sticker somewhere in the universe though
posted by pingu at 7:00 PM on August 24, 2021 [4 favorites]


I do wonder if the so-called "spooky action at a distance" quantum-unit entanglement they've been playing with could eventually be used to measure this. I mean, not tomorrow. Probably not in my lifetime. But maybe it could be done using that kind of weird thing that can be made to happen.
posted by hippybear at 8:43 PM on August 24, 2021 [2 favorites]


'Light from ancient quasars helps confirm quantum entanglement'

whatta bout the quasars.

"Last February, the MIT team and their colleagues significantly constrained the freedom-of-choice loophole, by using 600-year-old starlight to decide what properties of two entangled photons to measure. Their experiment proved that, if a classical mechanism caused the correlations they observed, it would have to have been set in motion more than 600 years ago, before the stars’ light was first emitted and long before the actual experiment was even conceived."

spooky.
posted by clavdivs at 8:45 PM on August 24, 2021 [3 favorites]


And, again, apologies for perseverating and being half-wrong, the real takeaway is that you can't synchronize clocks rather than that we can't measure the one-way speed. Which, sure, one implies the other, but I think a lack of synchronization is a more important issue.

Yeah, as best I can tell, the idea of a "one way speed of light" being a meaningful concept is fundamentally entangled with a notion of simultaneity - in that, you need simultaneity to be a thing for a "one way speed" to mean something, and unfortunately simultaneity isn't a thing. Lucky for us, the math and experiments both behave well if we make the simplifying assumption that c is universal. Click bait aside, the video isn't so much about "physics is wrong!" as "huh, it's kind of an interesting bit of trivia that due to how relativity works, it's not possible to prove that the speed of light is (or isn't) the same going towards a destination as coming back from it".

I kinda group "a measurement of the one way speed of light" with "a magnetic monopole" and "a tiling of the Euclidean plane with regular pentagons" as things you just can't have, as a result of how the universe and/or math works.
posted by NMcCoy at 8:50 PM on August 24, 2021 [3 favorites]


If light speed was instant in one direction and 2/c in the other, then for Hubble to see distant stars, galaxies etc in both directions, that are in the same point in their evolution, would seem to need the big bang to have happened earlier in one direction than in the other. That would be pretty wild! But also Occam..

A small enough difference in speed would be impossble to detect this way now, but far enough in the future, it would be detectable the same way.

I demand a Greg Egan novel.
posted by joeyh at 8:54 PM on August 24, 2021 [1 favorite]


(c/2)
posted by joeyh at 9:01 PM on August 24, 2021


If they ever manage to do this it'll be amazing. I'm still pretty amazed at the whole gravity wave detection thing they have going on now. THEY FIGURED OUT HOW TO DETECT BENDS IN SPACE-TIME.
posted by hippybear at 9:04 PM on August 24, 2021 [1 favorite]


(which I just realized depends on the differences in speed of reflected beams of light, so it depends on the whole thing this is all talking about... Physics is all so interrelated!)
posted by hippybear at 9:19 PM on August 24, 2021 [1 favorite]


If the speed of light were instantaneous in one direction and c/2 in the other, what speed would it be at an angle?
posted by Joe in Australia at 9:52 PM on August 24, 2021 [1 favorite]


42?
posted by hippybear at 10:10 PM on August 24, 2021 [5 favorites]


I kinda group "a measurement of the one way speed of light" with "a magnetic monopole" and "a tiling of the Euclidean plane with regular pentagons" as things you just can't have, as a result of how the universe and/or math works.

In the end, we may discover the working of the universe and/or math is the fundamental reason why we can't have nice things.
posted by otherchaz at 10:32 PM on August 24, 2021 [1 favorite]


If the Michelson-Morley experiment didn't mean anything about the round-trip speed of light in different directions, why was it so significant?

I guess that relativity cancels the terms, but under that framework how does instant travel make sense? Wouldn't that require infinite time dilation (speed or gravity) along that direction? You aren't proving anything by dividing by zero.

I have a lot of respect for Veritasium, but I can't understand the point of this video.
posted by netowl at 10:43 PM on August 24, 2021


>If the Michelson-Morley experiment didn't mean anything about the round-trip speed of light in different directions, why was it so significant?

The context is important -- light was waves like sound and sea waves, riding on a medium -- so the experiment sought to measure more properties of light's medium, something they called 'lumeniferous ether'. Instead it was inconclusive and was easily replicable, so joined the search for more information about light.

>THEY FIGURED OUT HOW TO DETECT BENDS IN SPACE-TIME.
Well, much later, with far more sensitive interferometers, a version of the Michelson-Morley correlated some ripples on the rubber-sheet of the cosmos -- what should bend your noggin is whether this is the lumeniferous ether.
posted by k3ninho at 11:35 PM on August 24, 2021 [2 favorites]


Not something I have much energy to think through in detail, but since there have been time dilation experiments successfully done with satellites, then isn't that basically indirectly measuring the speed of light? Otherwise the experimental data wouldn't make any sense. Or are GPS experiments somehow independent of c? Also, how is "measure" being defined here?
posted by polymodus at 1:32 AM on August 25, 2021 [1 favorite]


I was referring to this Wikipedia page and this Stanford Encyclopedia of Philosophy page which are pretty good primers for the issue, which is mostly one of definitions.
posted by Proofs and Refutations at 8:18 PM on August 24


Eponysterical!
posted by LizBoBiz at 2:14 AM on August 25, 2021 [1 favorite]


I think the point of the video is just to popularise the idea that the speed of light can only be measured if the light starts and ends at the same observer. He's not really trying to say light does travel at different speeds in different directions, only that the suggestion is unfalsifiable. And that the physics still "works" even if you do assume different speeds away from you and towards you.
posted by mokey at 3:59 AM on August 25, 2021 [4 favorites]


ok, if there was some kind of directionality woven into the fabric of the universe that caused light to travel faster in one direction than another, would we be able to notice differences in the state of distant light sources? Like would there be a greater variation in the age of stars, etc we can see in one quadrant of the galaxy than in another?
posted by Jon_Evil at 6:49 AM on August 25, 2021 [2 favorites]


Jon_Evil: if there was some kind of directionality woven into the fabric of the universe that caused light to travel faster in one direction than another, would we be able to notice differences in the state of distant light sources?

Absolutely correct. The fact that the cosmic microwave background radiation reaches us at the same temperature from all directions indicates that this light has taken the same amount of time (and hence sampled the same expansion factor of space) to travel to us (one way!) since it was emitted 13.7ish billion years ago.

If you want a more "local" example, Ole Rømer measured the difference in the one-way light travel time from Jupiter to the Earth when the Earth was on the near side versus the far side of its orbit from Jupiter, way back in 1676.

There is also direct experimental evidence that light and gravitational waves travel at the same speed, based on a measurement of the two arriving at Earth within less than 2 seconds of each other after traveling one way across a distance of 130 million light years.

if this is just a mistake by Derek Muller then why is this a sufficiently well known philosophical issue with special relativity to have an extensive wikipedia page and an entry in the Stanford Encyclopedia of Philosophy?

Looks to me like philosophers have got some 'splaining to do.
posted by heatherlogan at 7:07 AM on August 25, 2021 [3 favorites]


polymodus -- the time dilation experiments are closely related to what's being described here, namely synchronization. For those, you synchronize two clocks locally in space, then move one of them. When it returns, the clocks will be de-synchronized in the way predicted by relativity.

The issue with trying to measure one-way speed of light is that you need the clocks at the beginning and end of the one-way trip to be 'synchronized' -- that really just means comparable in a meaningful way. But if you just use theoretical time-dilation results to compare the clocks, then you are assuming the one-way isotropy you want to measure. So you can't do it that way. In fact, it seems kind of fundamental that there's no way to synchronize the clocks that doesn't implicity assume isotropy of c in the first place. That's my read of this material, anyway.
posted by dbx at 7:29 AM on August 25, 2021 [4 favorites]


Conservation of angular momentum is fundamentally linked to the isotropy of spacetime, as proved by Emmy Noether a century ago. No violations of angular momentum conservation have been observed.
posted by heatherlogan at 7:35 AM on August 25, 2021


I found the ending remarks about simultaneity dissatisfying, since it is impossible to say categorically that two events separated in space are simultaneous in time, even if you accept the Einstein convention. Two observers moving relative to one another will perceive different events as simultaneous, and since there is no 'standard' inertial fame neither can be said to be 'correct'.
posted by Omission at 7:37 AM on August 25, 2021


If you want a more "local" example, Ole Rømer measured the difference in the one-way light travel time from Jupiter to the Earth when the Earth was on the near side versus the far side of its orbit from Jupiter, way back in 1676.

Wikipedia says: "The first experimental determination of the speed of light was made by Ole Christensen Rømer. It may seem that this experiment measures the time for light to traverse part of the Earth's orbit and thus determines its one-way speed, however, this experiment was carefully re-analysed by Zhang, who showed that the measurement does not measure the speed independently of a clock synchronization scheme but actually used the Jupiter system as a slowly-transported clock to measure the light transit times."

There's lots of good reasons to believe the one-way speed of light is c, but there seem to be unavoidable complications w/r/t clock synchronization that show up when people actually try to measure it.
posted by BungaDunga at 7:47 AM on August 25, 2021 [6 favorites]


It seems an epistemological issue. You could say that nobody has ever seen themselves, only their reflections. Whether this matters depends on whether you believe all truth claims can and need to be empirically grounded. In the case of measuring the one-way speed of light it appears to be impossible to do so. What you can do is measure the time it takes for light to reflect back to you and assume it had the same speed both ways. But that's not an empirical fact, just a reasonable idea.
posted by dmh at 7:58 AM on August 25, 2021 [2 favorites]


What you can do is measure the time it takes for light to reflect back to you and assume it had the same speed both ways.

You could just measure the time at each end of a one-way trip, what trips these experiments up is less epistemological concerns and more special relativity, which makes clock synchronization depend on the same thing you're trying to measure. Two-way speed measurements only need one clock, so there's no synchronization required.
posted by BungaDunga at 8:04 AM on August 25, 2021


It's also impossible to disprove that physicists are living inside Roko's Basilisk and these philosophers are just a simulated torture mechanism.

There is a thing called the classical limit. It is a very good approximation in some situations, including the Jupiter timing measurement. In the classical limit the clock synchronization problem disappears. Deviations from the classical limit are calculable, and can be compared quantitatively to experiment. Most importantly, physics holds together, and the supposed "loopholes" that would allow light to travel at c/2 in one direction and infinite speed in the "other" direction (which direction would this be?) are akin to the "loopholes" whatabouted by flat earthers that wilfully ignore the entire rest of human knowledge.

Also, optical tweezers.
posted by heatherlogan at 8:11 AM on August 25, 2021 [2 favorites]


Isotropic light makes sense and is a good assumption. But I also think physicists are also too easily lured by symmetry. Broken symmetries (matter/antimatter, direction of time/entropy, CPT violations, etc) are arguably what allows the Universe to exist. I still recall my Astrophysics professor teasingly remarking "Observations show that the Universe is not highly curved but nearly Flat. Then symmetry arguments tell us it is must be exactly Flat. Right?"

I'm not sure that the CMB argument holds either. Anisotropic light speeds also imply anisotropic time dilation so I think the bookkeeping comes out the same.

If physics stays the same then is it merely a philosophical argument? Yes, it is, that is until it isn't. Until some effect is discovered that would be affected by light anisotropy.
posted by vacapinta at 8:23 AM on August 25, 2021 [3 favorites]


As far as I can see, the fundamental issue here is that there is one theory, special relativity, that assumes that the one-way speed of light is constant in all directions. Under that theory, you can set up experiments to observe the two-way speed of light, but not the one-way. There are lots of seemingly similar things you can observe, but one-way speed of light is illusive because space and time are tied up together, and once you assume special relativity is correct (to separate them) then you get your constant one-way speed baked in. In principle you could observe violations of special relativity, but no one ever has (in the appropriate weak-gravity limit). That plus the fact that it's really simple (basically just linear algebra) makes special relativity really appealing, and I think that's where a lot of people are coming from.

On the other side of things, you have "what if the one-way speed of light isn't constant?" The problem is there isn't any particular theory with predictions to compare to. The basic physicist reaction to "ideas" like this is "theoretical prediction or GTFO". No such predictions are available because there isn't any actual theory, just "what if not special relativity". Plus it seems inevitable that any such theory would be more complicated, and could only differ in situations that are very extreme (and thus haven't been observed) or in places where special relativity matched observations and the new theory didn't.

TLDR it's an idea that screams "this isn't a productive issue to worry about".
posted by Humanzee at 8:24 AM on August 25, 2021 [2 favorites]


the fundamental issue here is that there is one theory, special relativity, that assumes that the one-way speed of light is constant in all directions

My (layman) understanding is that SR works just fine if you assume the one-way speed of light has a preferred direction: it makes the same predictions about what you can actually measure, because any deviations from a constant one-way speed of light would be exactly balanced by the time dilation experienced while you're moving your clocks to set up the experiment. Einstein just chose as convention to treat c as the one-way speed of light because the alternative is to have a bunch of extra "what if there's an anisotropy" terms which all cancel out in the end anyway.
posted by BungaDunga at 9:00 AM on August 25, 2021 [3 favorites]


For all the discussion here, nobody can actually prove that light travelling away from you, hitting a mirror, and returning, went the same speed in both directions. I mean, obviously it did, since how would it know to change speed based on where you were? But you can't prove it. Yes, it makes no difference to anything, it's just interesting.
posted by mokey at 9:23 AM on August 25, 2021 [1 favorite]


If light had a dramatically different speed in one direction, wouldn't someone in a fast spaceship see relativistic effects like red shifts on one half of a voyage but not the other?
posted by Joe in Australia at 9:42 AM on August 25, 2021 [1 favorite]


I find it interesting because it reminds me of mathematical incompleteness, e.g. Godel, whose theorems contributed to deep understanding of mathematics.

I still don't get the dilation issue though, maybe this is naive but to get GPS experiments to work don't you need to plug in the actual value of c at some point? And not just some theoretically derived one but the actual constant. So how did we, measure, that number.
posted by polymodus at 9:45 AM on August 25, 2021


obviously it did, since how would it know to change speed based on where you were?

not if the universe has a preferred direction. Suppose there's a direction, that's absolute- it's the same everywhere- along which light travels at 2c. In the precisely opposite direction, it travels at half c. In every other direction it travels something between .5c and 2c. It's like there's a cosmic conveyor belt or something. Depending on where you place your mirror, it will sometimes travel faster going away, and sometimes travel faster coming back (or, if you pick an orthogonal direction, they'll both be c, but it's rare), but the two-way distance will always be c.

This does seem like a strange sort of way to arrange the universe, but it seems like experiments can't actually prove we're not in it, because the flow of time would also vary depending on which direction things travel, and the effects precisely cancel out if special relativity is correct.
posted by BungaDunga at 10:21 AM on August 25, 2021 [1 favorite]


how did we, measure, that number.

You measure the two-way speed of light (2*c) and divide by two. That's definitely c. It's just that the one-way speed of light might not be c in every direction, and attempts to measure c directly always end up accidentally measuring the two-way speed of light or making assumptions about clock synchronization that assumes what they're trying to prove.

If light had a dramatically different speed in one direction, wouldn't someone in a fast spaceship see relativistic effects like red shifts on one half of a voyage but not the other?

Time dilation is also moderated by the speed of light, so the effects end up canceling out. That is, the "speed of time" is also different in one direction, so you can't tell the difference. This is not very satisfactory and therefore most people can reasonably decide it's a silly question and not worry about it.
posted by BungaDunga at 10:35 AM on August 25, 2021 [2 favorites]


So light maybe changes speed depending on direction, but so does time dilation, so it all comes out even in the end? I can’t decide whether this is analogous to, or literally the same thing as, the thing where so long as there’s no acceleration, it’s impossible to tell whether you’re in motion or in a “fixed” position.
posted by ook at 10:37 AM on August 25, 2021


Einstein just chose as convention to treat c as the one-way speed of light because the alternative is to have a bunch of extra "what if there's an anisotropy" terms which all cancel out in the end anyway.

This, exactly. Isotropy of the speed of light proceeds directly from a chosen convention for defining simultaneity and spatial position for events that don't happen at the observer's own exact location. The convention exists for no other reason than that all the consequent formulae are simpler with that than with any other.

Similarly, there is nothing incorrect about a mathematical model of the Solar System that takes all positions relative to the centre of an assumed-stationary Earth, as long as the predictions it makes agree with those made by a different model that takes all positions relative to the assumed-stationary centre of mass of the entire system (which is close to, but not always inside, the Sun). Heliocentrism is not "more correct" than geocentrism; it's just a hell of a lot simpler to work with because the associated formulae (and the approximations of them that actually get used for most things) are way way way less fiddly.

I like to think of physics as part of a lossily compressed representation of a substantial portion of reality; the simpler the formulae the more efficient the compression for any specified degree of loss, and I'm a fan of efficiency.

Philosophically, this is also the basis for my opinion that there will never exist a workable Theory Of Everything. It seems to me that there will always exist some level of predictive detail at which it takes longer to decompress the relevant portions of theory than it does just to observe what actually happens.
posted by flabdablet at 2:16 PM on August 25, 2021 [2 favorites]


If the speed of light were instantaneous in one direction and c/2 in the other, what speed would it be at an angle?

Good question! I think it could possibly be any value between instantaneous and c/2, depending on how much anisotropy there was in the speed of light perpendicular to the first dimension already covered. Then it would be the proportional average between the speed of light in the one dimension and its speed in the second dimension, with the cosine of its angle being the proportionality factor.
posted by traveler_ at 2:56 PM on August 25, 2021 [2 favorites]


I don't see how that could work, though. Suppose there are four mirrors at the vertices of a square, each one three light years from Alpha Centauri. AC is four light years from us, and each mirror's distance from us is five light years.

Let's say AC flares, and the light from the flare travels to the mirrors and bounces to earth. Two of the mirrors are at right angles to the preferred direction, so it has no effect on the transit time and the light takes 3+5=8 years to reach us. One mirror is in the preferred direction, so the light takes 0 time to reach the mirror, then ... 25/4 years to reach us? And the light from the other mirror takes 6+4=10 years? Whatever the figures, we now know the one-way speed of light. And if there's a problem with us knowing the mirrors' distances, they're orbiting AC and exchange places with each other over time, so if their relative distances are changing we're going to have to do a lot of physics rewriting to explain this.
posted by Joe in Australia at 5:57 AM on August 26, 2021


I did not watch TFV. The squib caused me to lose respect for Veritasium, because first "nobody ever measured the one-way speed of light" elicits the response "well, duh." It's just trite.

As for "nobody can say if light moving toward the mirror travels 1/2c outbound and instantaneously inbound," that's just silly. As well as making the error of saying "/2" when he must have meant "*2."

I did scan the whole thread, and saw a lot of learned-seeming invocation of Maxwell's equations, which isn't how I thought to point out the absurdity. A few people have pointed out that from the photon's POV there's just "moving forward" (though they elide that there's an instant when it's "being reflected").

What nobody'd mentioned, as well as I could scan, is that we have a whole fucking lot of observational data about objects moving slower than light, but at a reasonable fraction of that speed, and it agrees just about perfectly to the limits of the precision we can measure with a theory that holds that the speed of light is invariant in all reference frames. To the extent it doesn't agree, there's a more complicated theory that's harder to compute solutions for which also requires lightspeed to be invariant.

Anybody who does not realize that, if we know anything, we know that the reflected light travels at the same speed as the outbound light, just shouldn't be making "explainer" videos about physics.
posted by Aardvark Cheeselog at 8:05 AM on August 26, 2021 [4 favorites]


As for "nobody can say if light moving toward the mirror travels 1/2c outbound and instantaneously inbound," that's just silly.

It's only silly inside a framework that defines an event D distant from observer O as simultaneous with local event L if and only if the observation of D by O occurs at time d(L,D)/c later than L, where d(a,b) is the spatial distance between two events regardless of direction, c is lightspeed, and the spatial distance between the observer and L is negligible.

Under that framework, the only cases where observations of distant, simultaneous events will actually occur simultaneously is if the events occur at the same distances from the observer. However, it does have the advantage that all observers who are stationary with respect to one another can work with a shared spatial reference frame and a common timebase, and make independent observations of common events that should agree without needing inter-observer adjustments.

There's another quite natural definition of simultaneity that defines events to be simultaneous if and only if they are directly observed to occur at the same time. For events at distances smaller than light-milliseconds - that is, human-scale distances; a light-millisecond is about 300km - it's nearly impossible for unaided human senses to perceive the differences arising between this definition of simultaneity and the first kind. That makes it hard to defend a claim that one of these definitions is "just silly" and the other one isn't. You need quite precise instrumentation to tell them apart.

And of course it's completely feasible to work up a version of Special Relativity based on this definition of simultaneity as well, just as it's completely feasible to work up an account of planetary motion where the Earth neither moves nor spins and everything else is moving relative to us, as opposed to the commonly-used heliocentric model that takes the solar system's centre of mass to be the stationary reference point.

We don't use that version of SR, though, for the same reason we don't use geocentric models to do astronomy: the equations are horrendous. Not only do the transformations required to reconcile one observer's observations with another involve corrections based on the relative velocities of the observers, but on their relative positions as well.

In other words, under hideous bizarro world SR, not only will observers in motion with respect to one another need to apply relativistic transformations to each other's observations in order to reconcile them, observers in different positions will need to as well. Less convenient than our commonly used SR, sure, but again hard to dismiss as "just silly". After all, it's exactly the same kind of issue we're already dealing with in SR, it just crops up in more circumstances is all.

There are payoffs, though. Given that definition of simultaneity, I can set up a remote clock that I can verify by direct observation is synchronized with mine. I don't have to take it on faith: I can directly observe t=0 there occurring simultaneously with t=0 here, and all the subsequent ticks being simultaneous as well. Not only that, but I should be able to see motion-induced desync directly. It's not going to boggle my brain the way it does when I'm trying to understand SR.

Having defined the spatial extent of my "now" according to this version of simultaneity, it becomes axiomatic that all inbound signalling is instantaneous regardless of direction.

If we're committed to thinking in rectilinear spatial coordinates, this is of course a grossly physically implausible proposition, almost to the point of being offensive. But it works very naturally in polar coordinates with the observer at origin, even if something like a personal consciousness field generating something akin to a gravity well needs to be posited to give the system an appropriately intuitive patina.

Hell, just posit one and be done with it. What we're observing is events in reality, and reality doesn't give two shits how we go about looking at it. We just cross-reconcile observations made by different observers using HBWSR and everything still works.

And we can also directly measure the one-way speed of light in this system. Because all inbound communication is instantaneous by definition, we just set up experiments where we directly observe an emission event somewhere and directly observe the instrumentation recording the absorption event somewhere else. Easy peasy. And again, we'll need to use HBWSR to reconcile our numbers from observer to observer, but hey, that's what computers are for.

What we're going to find, given that we're measuring exactly the same reality, is that directly outbound electromagnetic signalling propagates in vacuo at about 150m/μs. EM signalling that's neither directly inbound nor directly outbound is going to be seen to happen at a range of speeds between that and infinity.

FUCK knows what the extension to GR would look like, but there would certainly be one. And it's kind of nice to be intuitively prompted to see the Big Bang and attendant cosmic background radiation as something still happening, everywhere and always, right now rather than some mind-mangling number of years ago.

"Just silly", you still say? No, mate, this is conceptual art. You might as well call the Mona Lisa a silly painting on the basis that it took Leonardo more than a few minutes to photograph, filter, crop and print. Yes, I did just compare myself to the canonical Renaissance Man. I will leave deriving the transformations required to reconcile that observation with your own to you.
posted by flabdablet at 11:21 AM on August 26, 2021 [1 favorite]


Aardvark Cheeselog, what you may be missing from not watching the video is that the alternate models like Lorentz ether theory are completely equivalent to special relativity; you can think of them as alternative interpretations or formalizations of the same model. They don't predict any difference in any experimental outcome, so they are not falsifiable by the types of experiments you are talking about. In fact, if special relativity is correct, then these other models are equally correct.

The point isn't that "one way speed of light" hasn't been measured, it's that it fundamentally can't be measured, and so we are free to define it as constant in all frames (as Einstein did), or in other ways that give the same end result. In practice, Einstein's convention is always used because any other interpretation makes the math more complicated for no reason.

For the same reason, in Joe in Australia's experiment, such a model would necessarily predict that along the "slow" mirror-to-Earth diagonal, light takes 8 years to travel 5 cy, and along the "fast" diagonal it takes 2 years (!) to travel the same distance. So the total time along the different paths is (3 + 5) = (0 + 8) = (6 + 2) = 8 years.
posted by mbrubeck at 11:26 AM on August 26, 2021 [2 favorites]


GR wrecks simplistic notions of spatially separated comoving observers seeing each other's clocks tick at the same interval. Space is warped, yo.
posted by wierdo at 11:46 AM on August 26, 2021 [1 favorite]


Who knows? Maybe HBWGR would come out simpler than HBWSR. Or at least no worse than GR. Or at least not much worse.
posted by flabdablet at 11:48 AM on August 26, 2021


Why couldn't you synchronize your clocks at the midpoint of the light speed test, move them (at the same speed) to the endpoints, perform the test, then compare the clocks? I'm not a physicist.
posted by smokysunday at 4:43 PM on August 26, 2021 [1 favorite]


Isn't there another problem with this, in that a moving observer will see different relativistic effects depending on the direction they're moving in?

Me burning 100 tonnes of super lightspeed fuel heading Galactic north: "Wow! Look at the cool red shifts and distorted constellations!"

Me burning 100 tonnes of super lightspeed fuel heading Galactic south: "WTF? This sucks. It feels like I'm crawling."
posted by Joe in Australia at 6:06 PM on August 26, 2021


Why couldn't you synchronize your clocks at the midpoint of the light speed test, move them (at the same speed) to the endpoints, perform the test, then compare the clocks? I'm not a physicist.

That's the "slowly transported clocks" scenario. It turns out that if you assume two separated events A and B are simultaneous if two slowly-transported clocks at those two locations read the same time, then those clocks will also measure that the speed of light is the same from A to B and B to A. It ends up being a theorem in Special Relativity: those two things are really interchangeable assumptions and one can be proven from the other.

I'm not enough of a physicist to understand precisely why that is, let alone explain it properly.
posted by BungaDunga at 8:00 PM on August 26, 2021 [2 favorites]


mbrubeck:

> The point isn't that "one way speed of light" hasn't been measured, it's that it fundamentally can't be measured

This is precisely my point. And also the point that, if you don't grasp it already, you shouldn't be making "explainers" about physics.
posted by Aardvark Cheeselog at 6:17 AM on September 1, 2021


This is precisely my point.
Well, it's also Veritasium’s point, so I’m not sure where you think your understanding disagrees with his.

This isn't some crackpot idea that Veritasium came up with. These are concepts that were first discussed by Einstein's colleague Hans Reichenbach, as well as by Einstein himself. It does explore some interesting theorems about relativity that my own undergrad classes only briefly alluded to, so I thought it was interesting and informative.

Maybe you don't find it interesting or surprising that there are fully-consistent models of special relativity that obey the Laue–Weyl condition but not the Einstein synchronization convention. Some of us do, however.
if you don't grasp it already, you shouldn't be making "explainers" about physics.
Perhaps, if you haven't watched the video, you shouldn't be making critiques of it.
posted by mbrubeck at 7:01 AM on September 1, 2021


« Older This is an okapi   |   RIP Charlie Newer »


This thread has been archived and is closed to new comments