There are two cases where triple buffering makes sense:
1) If it takes a substantial amount of time to clear the image. Recall that in double buffering, you are displaying one image while drawing another. When drawing the image, the first thing that is often done is clearing the image to a background color (and depth). On some devices, this took a substantial amount of the frame-time, and adding more memory was cheaper than making the "clear" faster.
2) If it takes more than one, but less than two frame times to draw the image, you can have interleaved pipelines. You are viewing framebuffer 0, mostly completed drawing the image in framebuffer 1, and just starting drawing (with a different set of hardware) into framebuffer 2. When you are done drawing, display framebuffer 1, clear framebuffer 0 and begin drawing, and finish drawing framebuffer 2. Note that this kind of triple-buffering decouples update from latency -- you can get very smooth playback at, say, 120 Hz, but the latency is still 1/60th of a second a best.
Both of these were done when I worked at Silicon Graphics in the early 90's, on machines several orders of magnitude larger than the nexus 7.