There’s been a bit of talk around the Australian blogosphere, particularly in the tech blogs, about how the election would go a lot more smoothly if it was computerised.
Here’s the way it currently works (I’m just looking at the most common case here, I know there are lots of edge cases)
1. You turn up to the polling location, and present yourself to a person who has a list of everyone in your electorate.
2. Your name is checked off on the list.
3. You’re given voting papers and you shuffle off to a voting booth, where you mark the voting papers according to your wishes.
4. You take the completed voting paper, and drop it in a box with other completed voting papers.
5. At some point, that box is taken away, and the votes are tallied (manually).
6. Once all the votes are tallied, we have a result (or, for this election, no result – a hung parliament).
Let’s start with a brief look at what it would take to introduce some technical improvements to this process.
Assuming we keep basically the same model, but introduce some technical efficiencies, here’s a few ideas about how much it would cost us.
First, let’s look at the idea of computerising the voter verification (that’s step 2, above). I’m deliberately being conservative on these numbers – I’m pretty sure that these wouldn’t be achievable in the real world.
We have 8642 polling locations, nation wide. Assuming the location I voted at was fairly average, there’s 3 people checking names off a list there. It’s a 10+ hour day, so they probably have two shifts of workers. To computerise this, we’d need 3 computers, and a spare (just in case), and training for at least 6 people.
Simple little computers, purchased in bulk, set up properly for this purpose, and shipped to appropriate locations, then shipped back after the fact, and the data off-loaded from them, let’s estimate $2,500 each (although that’s probably on the low side, for government work). 8642 locations, 4 computers per location, you’re looking at $86,420,000.
You’d need to make sure the polling locations have power cabling in place, and that would have to meet OH&S standards of safety. Let’s say $300 per location for another $2,592,600.
You’d need a custom application (Excel isn’t going to cut it, but even if it did, you’d still need to buy almost 35,000 copies of it) written to allow the people checking off names to check off names – In the scale of things that’s fairly cheap, and assuming it’s written in a system that doesn’t require per-seat licensing fees, you can probably come up with that for under a million (but that would include the automated tallying of information after the fact). Let’s call it an easy $1,000,000 in round numbers.
Now, the operators will need training in how to use the system, and what to do when it goes wrong (and it will, in some cases, we all know that right?). Let’s assume a decent decentralised training method (train the trainer, trickle-down styles) to train the 51,000 people who staff the voter verification desks at around $250 each – $12,963,000
So far, we’ve spent over one hundred million dollars – $102,975,600 – of money from your taxes. And we haven’t even got to the polling booth yet.
Next: Computerising the actual voting!