We don't know anything. If you want to know how we are sure of it, we aren't.
But the information that light always move on the same speed, on any direction and for every observer has been tested again and again since the end of the 19th century. Nobody ever saw it fail.
There is of course a huge number of variables nobody tested yet. And we can't know if those are important or not. But we are very logically biased into assuming they aren't until we have a real need for them to explain something.
(Anyway, the thing you are trying to invent is a kind of modified Newtonian mechanics - even though it's not quite Newtonian, but people use that same name. There are other people trying to get a workable theory with similar parameters. Up to now, nobody got any.)
I'm sorry, but I'm not watching a 19 minutes video to discover what you mean by "two-way speed". Can you explain it?
But anyway, we've been comparing the speed of light over different directions since the 19th century. Modern relativity was created because of those measurements, and nowadays you can buy accelerometers and gyroscopes that work based on that constant speed.
It basically means round-trip tests where the source of light and the detector are the same, and the light reflects back from a mirror. One-way would be having the source send the light to the detector directly, but the problem with testing that way comes down to synchronizing the clocks of the source and detector.
I don't know anything about this, I just found a Wikipedia page [1]
Yeah, the one thing we tested again and again is that if there is any difference on the speed of light going and coming back, it must average to the same thing in every direction (actually in every circuit, with any number or directional changes).
There are some experiments that measure one-way speed even on the link you posted. The first measurement of the speed of light was one-way. Those are just low precision ones.
Can't measure one-way speed of light because you need to get the result back somehow (and that's also limited by speed of light). Usually we just send light and wait for it to bounce back.
> we can’t falsify it. According to some HN-er, this must be unscientific
Apparently more than merely "can't falsify it": the maths is such that if it was different, there would never be any observation that could distinguish between them.
That extra step — that it genuinely makes no difference — would, I think, cause it to be unscientific.
At least, unscientific within relativity; given relativity can't be the whole picture (singularities popping out of maths that presumes spacetime is differentiable), we may well find another better model that would allow us to make the test, which in turn allows us to ask the question scientifically.
That said, after watching the Veritasium video ages ago, I was immediately asking myself: what about the CMB? Surely that would have a stonking big anisotropy if the speed of light wasn't close to the same in all directions? (It does have one which is assumed to be because of our motion relative to it, and I don't know enough to guess what to expect of a different-way-speed-of-light-anisotropy other than its existence, and therefore cannot compare and contrast with the observed one).
> I was immediately asking myself: what about the CMB? Surely that would have a stonking big anisotropy if the speed of light wasn't close to the same in all directions?
No, you can always cancel it out by making the whole rest of physics suitably anisotropic too. Which would, of course, be ridiculous.
changes in the speed of light would change spectroscopy, it really isn't observed in our cosmic neighborhood (maybe in the very early universe at very high redshift).
But the information that light always move on the same speed, on any direction and for every observer has been tested again and again since the end of the 19th century. Nobody ever saw it fail.
There is of course a huge number of variables nobody tested yet. And we can't know if those are important or not. But we are very logically biased into assuming they aren't until we have a real need for them to explain something.
(Anyway, the thing you are trying to invent is a kind of modified Newtonian mechanics - even though it's not quite Newtonian, but people use that same name. There are other people trying to get a workable theory with similar parameters. Up to now, nobody got any.)