Cringely on e-Voting
Cringely argues that the best solution to e-Voting is Canada’s pencil and paper method. I agree that this has a number of advantages. But Cringely makes the same mistake that so many make when thinking about this issue: he fails to think about the disabled voter at all.
We have a principle regarding voting: A person should vote without any assistance or interference from any other person.
Why is this a good principle? Well, we want people to vote privately so that they vote their conscience, not what the other person might want them to vote. We don’t want people to be strong-armed into voting for someone they didn’t want to vote for. We don’t want anyone to know who someone else voted for. And so on. This makes sense.
Now take the blind voter and the voter with extreme fine motor skill difficulties. How can we preserve their privacy and allow them to vote without interference in a paper and pencil system?
The blind voter, if they can read braille, could be given an alternate ballot that they would be able to read on their own. But already, the paper and pencil solution that Cringely advocates has been abandoned in part. The blind voter using a touchscreen system can wear headphones that audibly confirms her selections.
How about a voter who cannot manipulate a pencil well enough to put their X in the spot next to their candidate? It’s not clear that there is a solution for these voters within the paper and pencil regime that doesn’t involve a third-party filling out the ballot for them. However, many of these who could not use a pencil, might be able to use a touchscreen.
It’s difficult, but not impossible to handle the touchscreen voting machine security issues. You do the following:
- All software used must be open source and reviewed by a diverse committee of computer software experts.
- The compilation of the software must be supervised by similar experts using an open source compiler on neutral hardware.
- The touchscreen machines should be a hybrid-type that prints an optical scan card, visually and audibly verifiable by the voter.
- The optical scan machine should be manufactured by a different company than the one making the touchscreen machines.
- When the optical scan machine counts the ballot it should be audibly (and perhaps visibly) verifiable by the voter.
- The printed optical scan ballot should have a perforated paper receipt that merely allows a voter to match his receipt to his ballot. It would not indicate his votes. (For instance, the optical scan ballot would have #366450 printed both on the ballot and the torn off receipt.)
- The optical scan ballot itself would be placed in a locked box at the polling location and they would be hand counted in re-counts.
- A certain percentage of random recounts would be mandatory.
- The physical security of the polling location and the voting machines would be guaranteed from the time of software compilation until the end of vote counting.
Complex, yes, but achievable. The question then becomes whether guaranteeing the privacy of those with motor skill disabilities justifies the expense of the above system. If not, then perhaps I’d agree with Cringely (with the proviso of alternative braille ballots). If on the other hand it does justify the expense, then I believe the above system is our best e-voting solution.
Here are the reasons for each of the above:
- The software has to be open source so that we know that Diebold and friends are not writing software that counts inaccurately, whether on purpose or accidentally.
- The compilation has to be supervised because even good clean code can be modified to behave differently with various compilation flags and tricks that have been demonstrated.
- The touchscreen prints an optical scan card so that there is a paper trail that a human can read, both for verification at that moment by the voter and later for recounts. It also provides visual verification for most voters and audible verification for the blind.
- The optical scan machine is manufactured by a different company so that those printing the ballots cannot design an optical scanner that systematically misreads the ballot. It’s an added check in the system.
- To know that the optical scan machine is accurately counting the vote, the voter should be able to see or hear it counting her vote, as an added confirmation at the polling location. Visibly for most voters and for privacy, and audibly for the blind.
- The ballot has a detachable paper receipt so that if there is ever any question about a particular ballot, that ballot can be matched up with that voter, who can then confirm their vote.
- The paper ballots are hand recountable so that we don’t have to trust the machines if they appear to be making mistakes.
- Random mandatory recounts force the software makers to be honest, because their mistakes or fraud will be caught.
- The physical security of the machines is guaranteed because all the above is useless if you allow crooks physical access to the machines at any stage in the process.
Make sense? The positive to Cringely’s paper and pencil system is likely low cost, but it adds problems for the disabled and increases the time until results are available. The positives for the hybrid e-voting system I just described are the nearly instantaneous results and the maximizing of unfettered access for the disabled. It’s downside is likely high cost. But I’m not sure that hiring all these partisan vote counters doesn’t cost a chunk of change too. I’m assuming Cringely is putting that cost on the parties rather than citizens at large. If so, we just need to know the actual costs of each system so that we can make an informed choice.
This work, unless otherwise expressly stated, is licensed under a Creative Commons Attribution-Share Alike 3.0 United States License.