The reality about "vote flipping" machines in NC and elsewhere

[Kelvin Mace]

148. Let me take these one at a time

Some precincts recording a larger number of undervotes, particularly in important races and located geographically in selected locations. For example, why would someone in a majority registered Democratic precinct vote for dog catcher or local magistrate and NOT vote for the Senate candidate? Undervotes might be 2-4% and jump to 8-10% on given machines in certain locations and races. No weird undervotes seem to occur where ballots are mailed or in some early voting locations. I interpret that as likely hacking where a formula was introduced to the programming. What is your explanation?

In the few cases that I could get exit poll or parallel voting data, we also see some strange patterns in selected races (some state office, some national). An example would be a school tax referendum in a Democratic district that passes easily while the GOP candidate for state office narrowly wins. Again, the strange results are often overlooked or buried in a large ballot, but difficult to explain unless a subset of machines or tabulators were manipulated.

Well, certainly I would have called for an audit of these machines. But here's the thing, if this happened in 2004 then there was, to my knowledge, no laws on the books governing such circumstances. TS systems were introduced with very little thought about the ramifications of the technology on the voting process. That by itself can result in lots of problems, unforeseen problems that while troubling, are not necessarily evidence of chicanery. Election boards started buying computerized voting system while not adding computer techs to their staff (they pretty much depended on the vendors to provide support and training). Undervotes could be the result of tampering, but a more likely explanation is a defective/improperly calibrated machine, or voter unfamiliarity with how the machine worked. Again, if memory serves, 2004 would have been the first presidential election following the 2000 fiasco when there was a mad dash to buy these machine to replace the very confusing and unreliable "punch ballots". When you rush to deploy tens of thousands of votings machines, and then rush to train tens of thousands of poll workers, then dump the whole new system in the lap of millions of voters (many in the 60+ non-techie age demographic) problems are going to occur.

As I explained to election officials when we were debating new laws governing these machines in NC in 2005, new machines require new rules and procedures to insure accurate results. The problem with digital voting is that honest errors can be indistinguishable from malicious intent, and that unless you were very careful, people would assume the latter (the law we passed in 2005 was called "The Public Confidence in Elections Act&quot .

How would you catch programmer manipulation in the cases that code was altered in such a way that did not show on screen?

Well, most states now have laws on the books requiring code be certified by the state before the election. Since these laws didn't exist in 2004, there was no mechanism to investigate these kinds of anomalies. Here is the law in NC that I helped write in 2005:

§ 163-165.9A. Voting systems: requirements for voting systems vendors; penalties.

(1) The vendor shall place in escrow with an independent escrow agent approved by the State Board of Elections all software that is relevant to functionality, setup, configuration, and operation of the voting system, including, but not limited to, a complete copy of the source and executable code, build scripts, object libraries, application program interfaces, and complete documentation of all aspects of the system including, but not limited to, compiling instructions, design documentation, technical documentation, user documentation, hardware and software specifications, drawings, records, and data. The State Board of Elections may require in its request for proposal that additional items be escrowed, and if any vendor that agrees in a contract to escrow additional items, those items shall be subject to the provisions of this section. The documentation shall include a list of programmers responsible for creating the software and a sworn affidavit that the source code includes all relevant program statements in low-level and high-level languages.

(2) The vendor shall notify the State Board of Elections of any change in any item required to be escrowed by subdivision (1) of this subsection.

(3) The chief executive officer of the vendor shall sign a sworn affidavit that the source code and other material in escrow is the same being used in its voting systems in this State. The chief executive officer shall ensure that the statement is true on a continuing basis.

(4) The vendor shall promptly notify the State Board of Elections and the county board of elections of any county using its voting system of any decertification of the same system in any state, of any defect in the same system known to have occurred anywhere, and of any relevant defect known to have occurred in similar systems.

(5) The vendor shall maintain an office in North Carolina with staff to service the contract.

(b) Penalties. - Willful violation of any of the duties in subsection (a) of this section is a Class G felony. Substitution of source code into an operating voting system without notification as provided by subdivision (a)(2) of this section is a Class I felony. In addition to any other applicable penalties, violations of this section are subject to a civil penalty to be assessed by the State Board of Elections in its discretion in an amount of up to one hundred thousand dollars ($100,000) per violation. A civil penalty assessed under this section shall be subject to the provisions of G.S. 163-278.34(e)."

As you can see we tried to cover all possible contingencies brought about by using TS and computer-based voting machines. We created a mechanism for examining the code before the election and made it a crime to change the code after it was certified. We even made the CEO the guy who would go to jail if something went wrong. Was it effective? Well, Diebold left the state rather than comply with the law, and then sold its voting machine business 4 years later. I would call that a win.

As I keep repeating to folks who are worried about rogue programmers diddling voting machines, with the changes in the law since 2004 it is very dangerous to try since mechanisms are now in place in most states to catch such tampering. If I am a crooked election official, far easier to simply reduce the number of available voting machines and spares in minority precincts, thus creating long lines and causing voters to leave without voting (which is what I suspect happened in Ohio in 2004) than it is to find someone with the expertise needed to rig a voting machine. It is next to impossible to prove criminal intent in how many voting machines are deployed, but getting caught with hinky code in a voting machine means the FBI in your office.

I would say that most of the possible manipulations occur in districts where voter suppression would be quite difficult because of the demographic stability. Also, I don't see such patterns in every race - just selected ones. We've seen all kinds of tricks here in addition to voter registration manipulation, ID requirements, and suppression tactics. Mail ballots tossed, signature checking games, and ballot design.

And all these tricks are easier and less trouble than trying to introduce "cheating code" to voting machines. If the machines are audited, the numbers will not add up, plus exit polling can spot weird variances. Again suppression is more effective and arguably semi-legal.

BTW, we've complained that machines were stored in unsecured locations, not tested, etc. Depending on the ES it's hard to get cooperation for observers and security. Some of our ES's and Secretary of State are openly repubs. Remember Katherine Harris?

Still can be a problem, but again, many states now have laws addressing this.

So, to summarize your main questions:

Can people rig a voting machine to cheat? Yes. Is it easy? Not as easy as many people think, and nowhere as easy as it might have once been now that many states look at the code. Did cheating happen in the past? Possibly. Although I have seen suspicious numbers, I have seen no conclusive evidence. Is it happening now? Highly unlikely, since you are far more likely to get caught if you try and there are safer ways to influence an election that don't involve a federal prison stay.