Caracas Chronicles

An election is a design problem. A highly complex one. How do you tally the preferences of millions of voters, across what is typically a large geographical space in a way that we can all trust to be the true reflection of individual voters’ preferences?

I wanted to expand some on Quico’s post earlier today. Everybody – including, for the first time, the government – is now questioning the “best electoral system in the world”. For the first time, everyone’s underpants are getting tangled by the extraordinary amount of null votes. This is good! It shows we’re focusing on the real problems – system design – and not the phantasmagoria – electronic fraud.

The design problems the system has are serious, but they’re also easy to fix. If they really want to resolve the issues with null voting, they can do it easily. All it takes is coding the system such that you can only null vote if you make an explicit, positive choice to vote null. You’d need a spot on the tarjetón that explicitly says “NULO”. If you press on that, we know you intend to annul your vote. If you make a mistake but don’t press “NULO”, the system just prompts you to go back and make a valid choice.

It’s not rocket science.

This would save us the current calvario of having to guess what a voter who voted null actually intended to do. And it would save us going through arguments like @puzkas’s about how more people annulled their votes in pro-chavez districts, which are purely speculative.

Even if districts that have been historically pro-PSUV show more null votes, it doesn’t necessarily follow that there are more null votes in the pro-PSUV voters. You can make assertions about having some districts be more prone to null voting, but you can’t demonstrate that certain voters, with certain preference are more prone to vote null. You can’t. @puzkas is on thin ice.

As I see it, the system has four major flaws. Two of which can be directly attributed to Smartmatic and the other two to the CNE.

The first flaw is the muddled relationship between the Tarjetón – the physical ballot paper – and the Screen. When you approach the booth, the first action one must take is on the tarjetón, not the screen. This isn’t really explained anywhere. The voter is just supposed to know. Many, of course, don’t.

After you are done pressing your buttons on Tarjetón, there is nothing telling you that know you must proceed to the screen. Again, the voter is just supposed to know. The screen should now reflect the choice you made on the Tarjetón. It has a big “VOTAR” button right at the bottom. Click on it, and your are prompted to confirm. This you get regardless of your choices (or lack of).

Click confirm and boom, your little ballot paper is printed. Let’s have a second indulgence here. That little paper is what matters. Everything done so far here was to get to the paper. Smartmatic’s machines are a very large, very expensive pencil.

The first flaw with the CNE is that it, for some made up reason, decided to enforce a time limit on the vote. After the polling booth president unlocks the machine, the user has three minutes to vote. Why is this time limit there, beats me. I can only assume they are worried of certain voters taking too long. Even if they do take a long time, you can’t take away a right because they are slow, or because the system is unclear. Such a time limit makes even the most tech-savvy nervous when voting. If time is a concern, it is resolved by adding more voting booths, not by leaning on voters – some elderly, many with limited tech familiarity and literacy – to do it faster.

Of course there are things you can do to make voters more efficient, like explaining the darn system to them. Here is the system’s second flaw, and it’s a big one: the absolute lack of instructions. Anywhere. Not online, not on the CNE website, not anywhere in the polling center or booth. It took an NGO to freelance a website showing how it happens – now down, due to too much traffic.

You basically have to figure out “intuitively” how to work the voting machine on the election day, with a three minute time limit. The result is that you have witnesses and booth volunteers explaining over and over how to vote, without any instructional material on how to do so (because you can’t do it in front of the machine).

The last flaw from CNE is el Tarjetón itself. It violates every principle of good design in the book: it’s cluttered, confusing, unintuitive. Often, last minute “alliance changes” are made and published in a “Gaceta Oficial” but not reflected in the Tarjetón, meaning your can find yourself in the voting booth pressing a button with a name on it, and have your vote be credited to an entirely different person. The placement of the “tarjetas” is tendentious, and that’s clearly done for the benefit of the governing party.

Ballot design has been studied to death. The guys at the Center for Civic Design have excellent guidelines on how to do it well. It would be hard to do it worse than CNE does.

Then there are multiple issues on top of this which have to do with accessibility. The font size and the lack of audio. But let’s not get into that.

Let’s get back to basics. You can resolve this null-voting scandal easily. In four easy steps:

1. Create instructional material explaining clearly how to vote. This is cheap.
2. Remove the time limit on the voting machines. Because, really, if you study the voting process, the majority of folks do it quick. A sporadic old man who needs 10 minutes is an outlier, thats not going to slow the process down.
3. Do not allow for null voting unless it is explicitly requested. In other words, have a spot on the tarjetón for “Null” voting (which is a valid choice in many cases).
4. If you haven’t chosen candidates, prevent the voter from casting an incomplete vote unless they explicitly say they want to vote nulo.

If Smartmatic had done any follow up, any proper research, any upgrades to their UI, many of this problems would’ve be resolved by now. But they seem to completely lack interest. Same goes for the CNE.

And that’s what’s unacceptable.