That video was pretty bad, though. It completely ignores everything we know about the CMB. If it were true that the speed of light was different in different directions, the CMB would look very different.
No disagreement on what you write, but the CMB spectrum is about as useless as radiocarbon dating for deciding which of two Hydrogen maser clocks on opposite sides of a university laboratory is wrong when they differ by ten nanoseconds.
However, we can generalize your point and say that real configurations of matter let us make those decisions reasonably and with good accuracy. (The BIPM UTC Circular-T contributors seem to manage fine, as do the various GNSS operators).
If the video had taken care to set their experiment in vacuum flat spacetime (perturbed only by the experiment), as the shown Einstein 1905 work <https://www.fourmilab.ch/etexts/einstein/specrel/> (top of p. 3) did, it would be better. That's because that setting offers no convenient-to-agree way to slice spacetime into space and time. The CMB does, but we suffer from practical limits on measuring the CMB's blackbody temperature.
I'm not sure that claim is true. Take this experiment, where A sends B a message following two paths:
/-->--B (/ and \ are mirrors)
| |
| |
\--<--A
with speeds: (assume lengths are all 1m)
/--cr--B
| |
cu cu
| |
\--cl--A
The time it takes for path 1 (left,up,right) is cl+cu+cr. The time it takes for path 2 is cu. B can measure the difference cl+cu+cr-cu = cl+cr. A can compare cl+cr to cu: if cl+cr != 2cu the velocity is not isotropic. To see that's always possible, A and B can simply bounce back path 2, so B receives pulses every 2cu (and hence can measure and compare to other time intervals).
Directional speed of light would be very weird, I think it'd show up everywhere in experiments if it weren't true.
---
Also, I think there's a notable distinction in physics: they are usually called postulates to distinguish from axioms. A postulate is an assumption about a physical theory (usually something simple and "beautiful" -- mathematically neat and satisfying Occam's razor); if a theory doesn't match reality, one of its postulates is incorrect. An axiom in mathematics of course can't be proven wrong. Because axioms are the basis for your mathematical theory describing reality, they can't be incorrect (as long as they form a mathematically consistent theory); the most could happen is they're insufficient to describe reality (you need other axioms and another mathematical theory), but they're not (somewhat) falsifiable in the sense of physical postulates.
Correct, cl+cr = 2cu is the best you can measure. But it doesn’t follow that cl = cr! Assume a linear transformation like αcl+βcr = 2cu, where α and β tell you how the speed of light varies in between left and right. We simply assume that α = β = 1.0, but what if α=½ and β=³⁄₂? Or α=2 and β=0? A long as α + β = 2.0, then everything will look exactly the same.
I think you assume one can measure the time between two events at different locations. I believe Einstein proved you can’t because there’s no way to truly synchronize clocks at different locations.
Another way to say it is that there’s no instantaneous "now" that all observers can agree on. In special relativity, "now" is meaningless. Or rather, "now" depends on the observer’s inertial frame of reference. There’s a nice diagram one can plot, with 1 dimension of space and 1 dimension of time, that shows lines of simultaneous events based on the velocity of a moving observer in another frame of reference.
> Veritasium has a nice video about it on YouTube.
He is often wrong in subtle and not—subtle ways, though. In this case, regardless of the issue of one-way measurements of c, his sources do not really support an anisotropy in the speed of light.
His video was mentioned already 5 or 6 times here. Isn’t here anything better at all?
