I mostly love this article, but it kinda glosses over how much more difficult it is to read out information out optically than it is to stimulate neurons with light. When you stimulate neurons you just need any ole photon, doesn't matter how many times it bounced around, or where it came from.. which is good because the brain isn't so much transparent, its kinda a milky haze. However, when you want to record optically from them you have to make an image of the neurons (unless you want all the neurons signals to get mixed together) and so you care about where all the photons came from. In order to take really effective pictures in the brain you need a fancy two photon microscope, and although some people are playing around with making tiny ones that one could potentially carry around on ones head.. they aren't really going to every be practically chronic implants for anyone, for many reasons.. but first of all you need to hook them up to a large, expensive infrared laser to make it work. That's not to say all this optical reading isn't really awesome, because scientists can make use of it to learn things about brains in more constrained situations.. i just wouldn't look to it to be the missing link in brain machine interfaces anytime soon.