Photo by Tobias Gellingskog
This a post on the topic of political process. It does not advocate any ideology, but a technology as a means to improve the current mode. A friend recently, after one of my tirades against representative democracy, asked me to come up with an alternative, upon which I surprised myself with an answer that followed my usual stripping down of government to necessary duties of a larger than communal government (such as research and other beneficial fields that require larger amounts of resources). What was the surprise? I linked machine learning at an app level with voting. Imagine a program that would vote for you based on what it knows about you. A program that you could adjust according to your preferences, and whose vote you could adjust if it voted against your opinion. Imagine a world without politicians. Maybe you’re sceptical, so here is my proposition and my answer to some potential objections.
The posited app/program would be based on machine learning. It would continually get better at voting according to your specific wishes. It would basically learn from your moral judgements, the basic principles you hold, as well as from your continual adjustments to its voting. You would sometimes go in and answer some questions to improve its functioning. This technology is already available in increasingly effective versions (search ‘Machine Learning’ for more on this subject). It would basically function as a voting avatar of yourself. Right now one politician represents you and thousands, for some of you hundreds of thousands or even millions of others. Every vote that politician casts will inevitably carry a high fail rate. By that I mean that the vote won’t represent the opinion (vote) of a great number of people whom the politician represents. An avatar working towards representing you specifically as accurately as possible is likely to succeed in doing so better than a politician which should (but doesn’t always) cede to the statistical majority of her or his constituency.
1. The Accuracy Argument
Basically the reasoning here goes: an app like that would not be accurate enough. Let’s put this argument into its proper context. The meaning of accurate enough can only and must be the result of a comparison to the current mode and its accuracy. If this machine can be better than a politician at representing your vote, well then this argument does not hold. So how would we compare? In any representative democracy each member of a house, senate, reichstag, or any other assembly represents a great number of people. In the U.S. each member of the house represents 726,765 people ( U.S. pop. 316,143,000 / 435 members of the House of Representatives). Adding the Senate to this does not help since the 100 members there cannot be added to the 435 of the House, but instead each represent 3,161,430 people. Let’s use the lower figure, that of the House. How many differing opinions on one issue whose answer is not yes or no could 726,765 people have? In theory as many as there are people. But let’s posit that it is a yes or no question. Let’s say that the vote goes 93% for, 7% against (yes, I am being very generous). That means that if the politician actually does vote according to her or his electorate, which (again) is not always the case, then this politician would have nullified the votes of 50,873 people (0.07 x 726,765). In the current mode you cannot ask your politician to change her or his vote. 50,873 people would not be represented. If all these people had an app and it had a 7% fail rate it would achieve the same result. With the fail safe of you being able to change the vote for a period of time, this number would fall, and the fail rate would continually drop as the app learns your preferences. A point to consider is also that the average misrepresentation by politicians is probably not going to be as low as 7% considering that people are divided 50/50 on many issues. The conclusion is that the app would not have to be very efficient in order to be better than politicians at representing you.
2. The Tampering Argument
The app could be tampered with. Let’s also put this into context. In the current system you can influence one person, and that person represents (in the case of the U.S. House) 726,765 votes. One move = 726,765 votes. I would qualify that as a very vulnerable system. The argument here seems to assume that people (politicians) are incorruptible and completely fail safe. Using similar technology to that banks use for security you would have a fairly safe system, able to withstand attempted tampering at a large scale. The main point is, all systems have vulnerabilities. If we can protect money online enough to trade, then we can vote using a similar system. And should your vote be tampered with, you yourself would be able to correct it.
3. The Expert Argument
This argumentis the easiest to refute. It states that we elect politicians to vote for us on the basis that they know better when it comes to issues of economy, law, science, morality, etc. The truth is that the range of fields, disciplines, subjects, topics or whatever categories you want to put in there that politicians have to deal with in number by far surpass the ability of any one person to master or become an expert in. Sometimes politicians rely on expert advice, well so can a program, albeit in the form of parameters which would be measured against your settings or end all/kill all presets. For instance, “I am completely against war”. The more advanced the technology would become the better it would be at learning the consequences of a topic, and therefore it would be able to vote more accurately according to your desire.
I don’t believe that this technology is here yet, but I could be wrong. What do you know about this, and would you like to share it with our readers? I will make space for a debate on this topic if there is interest. Send us your opinions on this subject to firstname.lastname@example.org