Let’s Be Clear About That Michigan “Computer Glitch”

The computer glitch that was found to have turned 6000 Trump votes into Biden votes in Michigan.  Software that’s used in 47 other Michigan counties, not to mention pretty much all the states with miraculous vote changes for Biden.

That was not a computer glitch. That was a programming decision. Whether that was some sort of programming error, which seems highly unlikely, or if it was deliberate cheating I don’t know, but I highly suspect deliberately bad software.  Let’s face it; adding integers is just about the easiest thing to get a computer to do (and if you’re thinking of adding votes as floating point numbers, go back to school).

Let’s get one thing clear.  Computer “glitches” cause a system to fail, or cause it produce random-looking outputs.  They can even possibly reboot the system.  If it’s good software, it gives some sort of indication that something bad happened.  If it changes the results of a count, or moves it to another candidate that’s something else.

Part of my perspectives on life come from being an old enough graybeard to have worked on computers in the 1970s, when you programmed by toggle switches on the front panel, and a single chip microprocessor was the highest of high tech.  At the time, I was working as a technician in a company that made industrial process equipment used in things as diverse as commercial chlorine gas production, to Hershey’s chocolate factories, to nuclear power plant controls.  I was also taking my first programming classes at the time.  The engineer teaching programming at night to college juniors was quick to point out a computer error would be to print absolutely gobbledygook instead of the nicely formatted output you wanted, not accidentally changing your bill or bank balance.

The class and real life came together at my day job to give a lesson in hardware failure vs. software error that was so vivid I recall it now, over 40 years later.  The company started producing computers to control the peripherals they made.  This was 8085 based (kids, ask your grandparents).  Like most computers of those days, its serial output to a printer was done through a chip called a UART, or a Universal Asynchronous Receiver/Transmitter.  After regular production test, the systems would be put into a burn-in chamber to run for a few days at 50oC.  We went through a problem where the UART chips would fail in burn-in; instead of printing a neat little summary of the number of times it had run each test in the sequence, and the total time it had been on, it would print a few lines of random text.  It was something like this (from memory – any similarity between this and the actual text it spit out is a slim chance):

##############$$$$$$$$$$$$##################$$$$$$###########S**********###$$$########A####@@@@@@@@@US#######%%%%%%%%%$$$$$$$$$$$$$A#########$$$$$$$$$####G*******E############

If you look at that closely, you’ll find the only letters in the printout spell ‘sausage.’ Of course, that’s a purely random event and makes just about as much sense as saying votes for one candidate really belong to the other candidate, but it does a good job of showing the difference between a hardware failure and software … feature.

For obvious reasons, I began to call this the sausage failure and every time one of our computers started showing the sausage failure, replacing the UART fixed it.

What I’m saying is if the computer “glitched,” it would have done something far more random looking and easier to detect than moving votes from Trump to Biden.  That’s no glitch.  It’s a software feature.

A clip I originally got from Zendo Deb at 357 Magnum goes better than anything else I can think of to wrap this up (you do read her, don’t you?).

EDIT 1045 AM EST: to add a relevant link to 90 Miles From Tyranny.  If true, the software in question is called Dominion; it’s used in every major swing state.  The DC Lobbyist is an aide to Nancy Pelosi.  Also see the link from McThag to Benford’s Law which shows a way to determine if the added votes are mathematically likely to be real.