Joe Strout wrote:
I was recently
reading about this actually being put to use in real buildings -- light
collectors on the roof, fed to fiber optics, which then light up glow
tubes right next to the fluorescent lights. Light sensors automatically
shut off the fluorescents when the piped light is bright enough.

However, the losses in the fiber optics are pretty severe -- IIRC, both
because of the wide bandwidth of the light, and because of its
intensity. I regret that I don't have the figures handy, but the upshot
was that this was considered impractical for anything but the top floor
of the building.

I'm surprised by this. End losses (getting light into and out of
fibers) can be 10% or so at each end, but the best fibers have only a
few dB/km loss internally and are good from visible light through a
couple of microns wavelength in the infrared. Are those fibers too
expensive for buildings? And electromagnetic radiation is linear. Why
should intensity should make any difference until it's high enough to
start heating the fibers.