I've been reading and re-reading Bret Victor's "A Brief Rant on the Future of Interaction Design" recently, and thinking about how we can provide the kind of tactile feedback he espouses as ideal for human physiology. Yesterday, I remembered "small cubes that self-assemble," a project at MIT that used high-RPM flywheels inside small cubes to allow them to generate torque and animate themselves with no moving parts on their exteriors. Here's the video:

I wonder whether using the braked flywheel technique could generate enough angular momentum in a small enough package to serve as the feedback mechanism for a human-computer interface. Tempted to start doing some research now, but I know that I already have enough on my plate, so I'm noting it here to come back to it later.