Would you buy a pocket calculator that told you 15+20+25=61? Would you buy one sold by someone referring to you as a member of the "Democrat Party"? How about buying one from someone who threatened you if you tried to find out why that result is 61 instead of 60?
If you've answered "yes" to all these questions, then by all means, get on the first plane to New Jersey where your help will no doubt be gratefully accepted by election officials who've already done all three. Oh, and of course they used the taxpayer's money to do it, and of course they used these systems in actual real live elections and of course they signed off on the obviously-wrong results.
This is the third in a series. In Sequoia Voting Systems threatens Princeton researchers, I explained how Sequoia Voting Systems used legal threats to deter security expert Dr. Ed Felten and his crack team of Princeton researchers from taking a close look at the systems. In Sequoia Voting Systems SUCCESSFULLY threatens Princeton researchers, I talked about how the threat played out, and discussed some related issues such as a belated investigation into the 2004 debacle in Ohio. It might be worth your time to skim those before reading on.
There are many, many problems with these systems, but the one that's currently the focus of considerable attention is the discrepancy between the number of votes cast for candidates and the number of votes cast for all candidates in a party. Those numbers should agree -- and on a number of machines, they didn't.
Here's what I mean. Consider a hypothetical four-person election for the Abbey Road Party:
- 8 John
- 6 Paul
- 4 George
- 6 Ringo
- 23 Abbey Road Party total
That's obviously wrong. 8+6+4+6=24, not 23. But this is exactly what came out of multiple voting systems: sometimes more, sometimes less, sometimes off by one, sometimes off by more.
If even one system did this once, that would be sufficient cause to have them all removed from service immediately. But given that multiple systems have done this multiple times, that should be sufficient cause to have them all removed from service permanently and to demand a full refund from the vendor. And in Union County, New Jersey last month, during the primary election -- failures just like this were observed.
Here's the proof: Voting Machine Summary Tapes. On that page you'll find links to scans of the summary tapes that each machine generated, summaries that show (a) this mistake and (b) the failure of election officials to notice it, since they all dutifully signed off on them without bothering to spot-check the math. These scans are difficult to read (I recommend enlarging them) but they're readable enough to demonstrate something that should send a chill right down the spine of everyone who cares about the franchise: they show the machines disagreeing WITH THEMSELVES.
That's not a "glitch" (the preferred term used by voting system vendors to explain away problems): that's catastrophic failure.
Sequoia Voting Systems has issued a memo attempting to explain this failure away; Dr. Felten has reproduced the relevant portions of it here: Sequoia’s Explanation, and Why It’s Not the Whole Story.
That page also contains a lengthy discussion of what the memo is actually trying to say and compares that to what the voting systems did in the field. I won't attempt to reproduce that discussion here, but I will recommend that you read it -- there are some excellent points made. The consensus seems to be that (a) the vendor's explanation is inadequate (b) this problem should have been caught during rudimentary QA, long before the machines were released (c) the fact that such a serious, obvious problem was NOT caught during QA raises a lot of questions about what other, more subtle problems have also not been caught and fixed and (d) this problem is not confined to Union County, New Jersey.
Dr. Felten says in part:
First, it’s obvious from this description, and from the fact that this happened on so many machines across the state, that even if Sequoia’s explanation is entirely correct, there was some kind of engineering error on Sequoia’s part that caused the machines to misbehave. Sequoia has tried to paint the anomalies as poll worker error, but that’s not plausible in light of Sequoia’s own explanation.
because it's clear from Sequoia's memo that rather than admitting this major blunder, they intend to blame poll workers for this. Of course, no matter what a poll worker or a voter does, no matter which buttons they push or how many times, this error should never happen.
Sequoia Voting Systems also seems intent on insulting roughly half the voters of the country, by repeatedly using the perjorative "Democrat Party", as in:
Let’s assume the Democrat party is assigned option switch 6 while the Republican Party is assigned options switch 12. If a Democrat voter arrives, the poll worker presses the "6″ button followed by the green "Activate" button.
Given how carefully worded that memo is, it's difficult to believe that's a mistake.