Democracy 2.0: technology can improve how we elect leaders

The system we use to elect our leader was designed before we had electric lighting, much less computers and advanced  communication systems. Sure, we’ve tinkered around the edges to improve things with technology (mostly with electronic vote-recording and vote-counting system that have been complete disasters), but what about the system itself? With computers and the internet, we could be much more scientific and rational about the way we actually decide on a leader.

For example: Plenty of tools boast the ability to tell you which candidate is closest to your views. You answer questions about your stance on a large number of issues, it compares your answers to the published stances of the various candidates. It uses some method to calculate the “distance” between you and each of the candidates, and tells you which one you are “closest” (or most similar) to.

Calculating “distance” between abstract things like “my beliefs and the beliefs of a candidate” is a fascinating business. If you want to get an introductory sense of how it’s done, one simple way to think about the distance between you and a particular candidate would be to based on Hamming Distance: take the list of “agree/disagree” answers that you have to a set of issues, and compare it to the answers of a candidate, and add up the number of times you disagree. That total number of disagreements is the distance: the greater the distance, the less similar you are to the candidate. This is what most of the “online quizzes” use to calculate which candidate you should vote for.

That’s really just an example of the simplest possible way to calculate conceptual “distance”, and mathematicians have come up with much more clever and complex methods as well.  But no matter what method is being used, you can see that doing this on a massive scale–such as calculating the distance between the opinions of an entire population and a set of candidates–wasn’t possible until our relatively recent developments in computational power.

But scientists have been putting their number-crunching algorithms, along with cool visualization techniques, to some interesting uses. For example, they can calculate the distances between activists, leaders, and political groups.

Political distances among contacts of Moroccan nationalists between 1930-1950

Political distances among contacts of Moroccan nationalists between 1930-1950

I’ve even used similar techniques to calculate the distances between some of my favorite science fiction and fantasy movies and television shows.

Sci-Fi Fantasy Visualization No. 1

How would this help us in an election?

Imagine yourself going into an election booth (or even completing it online… but let’s take this one step at a time!). Instead of being presented with a list of candidates, you are presented with a list of 20 issues.

(Side note: One layer of complexity in this system is determining how the issues would be selected for each election cycle. Presumably it would need to be updated for each major election to reflect the current political mood, events, and needs of the time. Presumably there would have to be a process for selecting questions that people felt was unbiased. This is probably not a trivial matter… but ideally isn’t insurmountable. Maybe it could be folded into our primary election process. Let’s assume for now, for the sake of argument, that we can arrive at a list of issues that people are happy with.)

For each issue, you have to select “agree”, “disagree”, or “no opinion”. Your “vote” is a list of 20 values: An “agree” is +1, a “disagree” is -1, and a “no opinon” is 0.  Mathematicians would refer to your vote as a 20-dimensional vector.

Voting day is over, and now we can take the average score for each question, across 130 million or so people.  If most people disagree with one of the issue statements, then the average value will be negative. If most people agree, the average value will be positive. The “average view of the population” will be a list of 20 values (a 20-dimensional vector) between -1 and +1.

Then, we calculate the mathematical distance between that list, and the list of values that represent the positions of each of the candidates. The candidate with the smallest mathematical distance to the average opinion vector becomes the next President of the United States!

This system wouldn’t have been possible, or even imaginable, back in the 1700’s. There would have been no way to compile all of this information and do all of the basic arithmetic in a reasonable amount of time. But now we have computers. Isn’t it time we took advantage of some of their incredible power to improve the way we choose the leader of our entire country?

There are some details that need to be worked out. We still need a way to figure out who the candidates are, and as mentioned above we need a way to select the 20 issues that appear on the ballot. We will still face challenges of security and making sure people’s votes are authentic and properly counted.

But it would still be a step above voting based on personalities and name-recognition, personal scandals and dopey “character-based” campaigns.

So what do you think? Would you approve of a more mathematically-based election method? Can you think of any reason we shouldn’t upgrade our election process to take advantage of the computational and information-processing technology that has emerged since the dusty old days of 1776?