He doesn't have to be lying, just unaware of how touchscreen calibration works. With resistive and infrared touchscreens, some portion of the screen can be miscalibrated or faulty even if other portions of the screen work correctly.
His "calibration test" doesn't actually demonstrate anything. Having some portion of the screen work incorrectly, while other sections of the screen work correctly, is entirely consistent with a calibration error for these types of touchscreens.
The only way to eliminate calibration error as the cause here is to: 1) re-calibrate the screen and see if the touch mappings are still incorrect. 2) if they are, the touchscreen itself could be faulty, so the next step is to replace the touchscreen with a known working version, and see if the issue persists.
Obviously there's no way he could have performed either of those steps, so there's no way he could eliminate calibration error as the cause.
Source: I've developed touchscreen kiosks in restaurants for years, and this behavior is entirely consistent with the miscalibrated touchscreens I've seen in the past. Often (generally with crappy touchscreens), a screen will hold its calibration for some period of time, and then stop working correctly until it's recalibrated once again.
Now, whether it's a good idea to use these types of crappy touchscreens in electronic voting machines is an entirely different question...
For such a major accusation, one would think to video all other choices working as expected, and only Obama being an exception. The simplest explanation for what he showed is a calibration error, but the matter is dramatically сomplicated if you take what he has to say for granted.
