Quote (russian @ Apr 10 2014 09:06pm)
That's not how I interpreted his question. I saw it as "you put light in at point A, it comes out at point B. But if you put light in at point B, it comes out in places other than point A (and maybe A also)". So it took one path going one way, and another path going the other way.
Well, you will get youself into trouble if you start talking about which path the light took. All you can really say is it started somewhere and was then detected somewhere else.
The picture you drew is another great challenge to find the error. Here the problem is the precise angles. You can not feed the light in at exactly the critical angle, and you can not feed it back the other direction exactly perpendicular to the glass. So you can not get the behaviour as drawn with a real box. Some of the light from A will not be reflected as drawn, and some of the light from B will be reflected back to A.
There is always going to be a problem since the black box I described earlier would allow you to reduce then entropy of the universe and make free energy. Something like this:
You start the system with the hot reservoir hot, the cold reservoir cold. Heat spontaneous moves from the hot reservoir to the cold reservoir and produces work with a heat engine. This quickly comes to a halt when the cold reservoir and the hot reservoir come to the same temperature. So then you use the one way black box to allow photons to pass from the cold reservoir to the hot reservoir, but prevent the opposite. Now your cold reservoir is cold again, and the hot one is hot again. Rinse and repeat, get as much free energy as you want. Which leaves us with a contradiction. IMO the way to resolve the contraction is simple, that black box don't exist.