The YouTube way is the simplest solution, if you really want another way, I would use Flash (having worked with it). This would just give you more control for the keybind,etc. It uses ActionScript which is not hard to pickup. Benefit would be making a desktop version of it with Adobe AIR or you could just upload the SWF to a website.
To optimize the video size, try limiting the Video FPS to ~24, and lower the bitrate. For each interaction you could stream in only the videos you actually require.