Others have mentioned Jason Brennan’s work, but I wanted to specifically add this comment he made about practical initial trials of a first step towards epistocracy in his effective altruism AMA.
Are there any small-scale, experiments with Epistocracy you think that countries or other jurisdictions should try as a first stab at testing this form of government? What would you like to see and where?
I’d like to try enlightened preference voting in Denmark or New Hampshire.
How it works: 1. Everyone votes for their preferred thing (whatever is being voted on). 2. Everyone somehow registers their demographic data. 3. Everyone takes a 30-question quiz on basic political information.
With 1-3, we then estimate what a demographically identical public would have voted for if it had gotten a perfect score on the quiz. We do that instead of what the majority/plurality actually voted for.
There are lots of details here I’m not getting into, but that’s what I’d want to try. No one’s done it to actually decide policy, but researchers have been doing this in labs for a long time with good results.
Are you worried about governments using the quizzes to favour certain groups regardless of political knowledge? Kind of like gerrymandering. Who will decide what answers are correct? Or do you expect that this would only be abused if they were going to do far worse anyway?
Also, maybe demographic data should include stuff like health conditions and personality, although intellectual disabilities may prevent people from scoring well in the first place, and intelligence/knowledge may correlate with actual interests, i.e. what would be good for a person. Has there been much written about this? I guess we’d hope the more informed would look after the less informed? We’re already hoping for this now with representative democracy.
Epistocracy as described by Jason Brennan is similar to things I’ve been thinking about. Also sortition, which is related but different. Here are some of my thoughts:
If we have a system of ‘extrapolating volition’ of segments of the population by taking representative samples of people and paying them generously to learn and think and debate on a key issue for a couple of weeks, then give their best full answer as to what should be done… And then also quiz them on their factual understanding of relevant systems, and register their predictions about outcomes… I think this is in some sense a more ‘fair’ look at what that segment of the population would think if given plenty of time to think.
Then from there we can make a model which lets us average these opinions and extrapolate (compensate for biases in the predictions, etc.), but in order to check that we’ve ‘extrapolated well’ we should check our extrapolations with different members of that population segment. This process can be repeated several times until we get good agreement on the extrapolation. I think extrapolating is a good idea, but I worry it would be vulnerable to abuse if you didn’t then have the step of the people who are being extrapolated for approving of the extrapolation.
Also, I have some thoughts along the same lines of Jason Brennan’s criticism of democracy as being ‘a system of pushing around the minority.’ Growing up in the liberal Quaker tradition, I’ve spent a lot of time in groups of people trying to do decision making via consensus. The quaker consensus process allows for some small portion of the group to ‘stand aside’ to allow a decision to go through that they disagree with, but this is uncommon. When it does happen, it’s usually less that 5% of the group. It occurs more often in groups of over >200 people than in the more typical consensus-seeking groups of 20-100 people.
I feel like the insights from this process for me are that it doesn’t scale well, and takes a lot of effort per-decision even when it works, but does have some really nice properties of finding better agreements. Often, the time taken to hear everyone’s point of view and share evidence thoroughly means that better solutions are found than anyone even came to the meeting with in the first place. So, this consensus process could inform our governance process indirectly, through finding related techniques which work with larger groups (for instance the Polis Process for crowd-sourcing consensus mentioned by Audrey Tang on the 80k hrs podcast ) or by selecting small representative-population-sample committees of 50-100 people and having them go through a consensus process to draft a statement of group opinion on a topic. I think this is valuable in addition to the ‘individual opinion sample’ mentioned above, because my experience is that people who talk face-to-face with each other on a contentious issue in the context of consensus process tend to come up with better (more win-win) and more empathetic compromises through understanding each others’ points of view and deeply felt emotions. It could also be a useful process for a small decision-making group like a board of trustees.
Others have mentioned Jason Brennan’s work, but I wanted to specifically add this comment he made about practical initial trials of a first step towards epistocracy in his effective altruism AMA.
link: https://forum.effectivealtruism.org/posts/z3S3ZejbwGe6BFjcz/ama-jason-brennan-author-of-against-democracy-and-creator-of?commentId=Cdq6C2f7mQZiykELC
Epistocracy as described by Jason Brennan is similar to things I’ve been thinking about. Also sortition, which is related but different. Here are some of my thoughts:
If we have a system of ‘extrapolating volition’ of segments of the population by taking representative samples of people and paying them generously to learn and think and debate on a key issue for a couple of weeks, then give their best full answer as to what should be done… And then also quiz them on their factual understanding of relevant systems, and register their predictions about outcomes… I think this is in some sense a more ‘fair’ look at what that segment of the population would think if given plenty of time to think.
Then from there we can make a model which lets us average these opinions and extrapolate (compensate for biases in the predictions, etc.), but in order to check that we’ve ‘extrapolated well’ we should check our extrapolations with different members of that population segment. This process can be repeated several times until we get good agreement on the extrapolation. I think extrapolating is a good idea, but I worry it would be vulnerable to abuse if you didn’t then have the step of the people who are being extrapolated for approving of the extrapolation.
Also, I have some thoughts along the same lines of Jason Brennan’s criticism of democracy as being ‘a system of pushing around the minority.’ Growing up in the liberal Quaker tradition, I’ve spent a lot of time in groups of people trying to do decision making via consensus. The quaker consensus process allows for some small portion of the group to ‘stand aside’ to allow a decision to go through that they disagree with, but this is uncommon. When it does happen, it’s usually less that 5% of the group. It occurs more often in groups of over >200 people than in the more typical consensus-seeking groups of 20-100 people.
I feel like the insights from this process for me are that it doesn’t scale well, and takes a lot of effort per-decision even when it works, but does have some really nice properties of finding better agreements. Often, the time taken to hear everyone’s point of view and share evidence thoroughly means that better solutions are found than anyone even came to the meeting with in the first place. So, this consensus process could inform our governance process indirectly, through finding related techniques which work with larger groups (for instance the Polis Process for crowd-sourcing consensus mentioned by Audrey Tang on the 80k hrs podcast ) or by selecting small representative-population-sample committees of 50-100 people and having them go through a consensus process to draft a statement of group opinion on a topic. I think this is valuable in addition to the ‘individual opinion sample’ mentioned above, because my experience is that people who talk face-to-face with each other on a contentious issue in the context of consensus process tend to come up with better (more win-win) and more empathetic compromises through understanding each others’ points of view and deeply felt emotions. It could also be a useful process for a small decision-making group like a board of trustees.