I've just seen a presentation of a great piece of work from Mitsubishi Research. Basically, it allows you to use a normal LED to sense light levels, with only a microcontroller and a single extra trace on your PCB. This means it can be added to a device that already has an LED and microcontroller at give or take zero cost. What's more, as you now have both an emitter and a receiver, you can use it for short range bidirectional communication. It's very clever, very useful, and, unfortunately, very patented.
Now I'm not sure how I feel about this. Unlike, say, the one-click shopping patent, there are no technical grounds to object to it. It's original, non-obvious, and hasn't been done before. It's a piece of apparatus, and it solves a real problem (many patent systems require a physical instantiation and/or usefulness). However, it seems a crying shame that the technology can't be freely used and incorperated into other devices. The patent system was designed to protect the original inventors, allowing them to exploit the invention for a time while guaranteeing that it's (eventually) available to the public. This seems a laudable aim, but loses it's lustre when the exploitation rights are assigned to a corporation. I don't think that's really what's bothering me, though. I think it's that in this case, patenting the technology both reduces the potential for research, but also reduces the chance of wide dissemenation of the technology - even a slight per-device license fee would significantly reduce the cost benefit of the technique. Still, it was a very cool paper.