I wonder why no one is examining the possibility that the manufacturers of the digital voting machines may be programming them to alter votes. It would be so easy. Look at what Volkswagen did with emission controls. These manufactures need only build a subroutine that is called based on the date and time. Any testing done outside this date and time will not call the subroutine and show no altering of the votes. I read one argument that the system does not know the entries, because the displays are images. However, the image has to follow a defined format and optical character recognition (OCR) is easy to embed into a separate chip that is called in the hidden subroutine. It is also my understanding that the software on these machines is proprietary and not accessible for review.