Category Archives: Hardware

Electronic Voting Machine Roadmap

1844 Presidential Ballot

Paper ballots have a number of benefits over electronic voting. Being easy to count isn’t one of them.

The state of electronic voting machines is rather scary to me.  According to Wikipedia only 27 states require a voter-verified paper audit trail.  That seems rather foolish given the importance of the electoral process.  If individuals are willing to spend millions of dollars, collectively billions of dollars, to try to get people elected it seems reasonable that there may be some individuals willing to defraud the vote to achieve their goal. [To be clear, I am in no way accusing anyone of doing so.]

I’ve believed for some time now that voting could be done conveniently and securely electronically. So I’ve put together my vision of an electronic voting system.

User Interface

From a UI perspective I’ve used two types of electronic voting machines.  The first one was hideous.  It was literally a large board that had every race and candidate on it visible at a single time.  Each candidate had a button and when you pushed the button next to their name a red LED lit up.   It was easy enough to use, but it was really intimidating to look at.  I have no idea how write-in candidates were handled although I do recall there being a button for them.  I’m unable to find a picture of them so I’m hoping they are no longer in use.

The second voting machine I thought was pleasant enough, although I thought the UI buttons should have been bigger to avoid screen calibration issues.  I think each ballot item should be an individual screen, that way the options would be ridiculously large and even those without fine motor skills would be able to vote easily.  Other than that, the UI seemed reasonable enough to me.

Voter-Verified Audit Trail

My major problem is with the lack of a voter-verified audit trail on the machines that I’ve used.  I’m simply taking the computer’s word for it that my vote was counted.  My major modification to electronic voting machines would be the addition of a thermal printer that prints a human readable ballot onto a traditionally sized card-stock ballot.  In addition a QR code would be printed on the bottom of the ballot which contains the vote information in a very computer friendly form as well as a ballot id number.

The choice of a thermal printer is due to their reliability and low-maintenance costs.  The card-stock is a requirement so that the ballots can be easily fed into vote scanners.  The inclusion of a QR code is to allow a computer to easily verify that the optical character recognition of the physical ballot was successful and to allow for the voting machines to record ballot ids of authentic physical ballots.  The ballot ID would never be associated with a person, it would be recorded as being cast.

Once the ballot is printed a final confirmation screen appears that asks the voter to ensure the printed ballot matches their desired selection and gives them the opportunity to raise a flag if it does not.  If they’re happy with their vote they will take the ballot and place it into the central ballot box.

My second modification would be the creation of a digital ballot box.  Upon the completion of the final confirmation screen the vote would be tallied individually on the machine as well as sent to the electronic ballot box for the voting location.

Three Independent Vote Counts

Finally I would institute that the voting machines be collected by one party, the electronic ballot boxes another and the paper ballots by another.

In order for an election to be finalized each party would need to independently deliver their tally of the results to a fourth party that will confirm that first and foremost the outcome of the vote is the same, secondly the electronic voting data should be identical, finally the physical ballot tally should be within a small margin of error of the electronic data.  This margin of error is only to account for scanner errors.

I believe this would make sure that no one party would have the ability to corrupt an election.

The big question is, what happens if the data doesn’t match?  It’s my opinion that the physical ballots should have the final word.  That’s because it’s a physical item that the voter was able to verify matched their intentions.  It’s also how voting has been done in the United States since 1892.  If it’s managed to work over over a hundred years it seems reasonable to me.

Over-all, I think this would be a very secure way to vote.  There are 4 copies (Machine, Electronic Ballot, Paper Human Readable Copy, Paper QR Code] of the voter data that should all match.  Any significant deviation would warrant an investigation to explain the variance.

Home Server Hardware Construction Lessons Learned

I’ve learned a few lessons learned while building my home server. The biggest thing I’ve learned is no matter how much you research you’re going to miss a few things. My original thought was to use a pci-sata card with 2 internal and 2 external sata slots. Combining the two on-board ports with the two internal ports would give me the four maximum sata drives supported by my hot-swap case and would give me two external slots that I could potentially connect drives to either to expand storage in the future or as a quick way to transfer files to/from the server.

The biggest problem of course is that this case does not expose the pci card to the outside of the case. This rendered the esata ports useless and actually a bit of an impediment since it just slightly hits the case. It works just fine, but I did end up ordering a new sata card with 4 internal ports instead of the 2e/2i. That was a $46 dollars whoopsie. Not so bad and it’ll allow me to compare performance of two on board and two on the PCI card versus all four on the PCI card. I think that will be interesting.

The next lesson that I learned was don’t order a part that you realize you forgot to in a hurry. I ended up ordering a 1U Left facing PCI riser card when I needed a 2U Right facing PCI riser card.

Once I realized the mistake I found that finding a 2U Left facing single pci slot riser was quite difficult. I found pretty much all of them ended up being made by the same manufacturer. I ordered this and it arrived from hong kong in about a week. Which leads us to the most valuable lesson learned.

When the 2U Left Facing PCI riser arrived I found that it caused the motherboard not to boot. It was noted that this happened in the original blog that I read about building a similar setup. Since I knew not too many people made this card I decided to read up on what he did to fix it. Using a copy of the pci standard and a voltmeter I found that pin A40 was grounded when it shouldn’t be. I cut this trace and like the blogger it seemed to work properly afterward.