The ideal is simple random sampling of a group that is large enough that it is extremely likely to closely match the population in all respects (not just demographics). Some bodies might have smaller optimal sizes, or have many people decline to serve, so random stratified sampling might be done (as is done for polling). There are competing ideas about the best way to do a lottery, but all need to be random to avoid self-selection bias or organizer manipulation. This book is mainly about why elections can't be a dominant tool in a democracy, and while I touch on sortition design in Part III, it will certainly take years of trial and error to perfect it. I explain my preferred design in a paper here:
I read the government design paper you linked. Pretty thorough, with lots of lotteries. I have a couple questions. One, do the policy juries need consensus? Or is it majority or some other percentage? And two, are the decisions up or down decisions? Or some other kind of decision? I’m thinking that sometimes, an up or down decision, like we make on initiatives or ballot measures here in Oregon, is not a very good way to form policy. It’s often much more nuanced.
The nuance, refinement and deliberation happens in prior lottery bodies perhaps with super majorities or consensus, etc... i insist that the body that drafts and refines a bill NOT be the body that gets to make the final decision. Pride of authorship and information cascades or groupthink can distort good decision making. a new jury with fresh eyes needs to hear all the pro and con arguments and then decide.. But whether a final draft bill becomes a law or is rejected needs a final yes/no vote. So as not to unfairly favor the status quo, I assume this would be a majority vote... but the rules panel might also choose to adopt a rule that a vote with a margin of less than X % needs a second jury to take a look, and see if they make the same decision. One point is that with a lottery selected rules committee, this democracy would be subject to un-ending improvement, based on lived experience with the system.
Couldn’t find English errors this time. But I have to comment on the idea that our legislature doesn’t represent us because they don’t belong to all the demographic groups. You do mention this in the text, but it’s of course possible for an old rich white guy to represent the ideas of the majority of their district. There are also many real examples of people of power in minority demographic groups (for instance Clarence Thomas) who don’t represent their demographic group very well.
You are certainly correct, but there is much more to this (as I will discuss in later chapters) having to do with diversity of experience, cognitive style and latent knowledge. An individual man may do an excellent job working for the interests of women. But when extrapolated over an entire body of mostly men, the odds of optimal representation go down and down.
Ok, agreed, but what I was getting at was that it’s obvious that our Congress isn’t like us, and that they don’t represent us well. But it’s not imperative that our congressmen and women look like us or reside in our demographic groups to make good policy, especially in the international arena, which they should be focusing on. But good deliberation is necessary. (It's hard to write down my thoughts – this is a very deep and wide subject area, but hopefully you get the idea, and will be talking about this in subsequent chapters.)
Another question... With sortition, if you include a demographic component, do you lose the random nature of the process, and hence lose the statistical basis for its efficacy? I thought in polls, the whole idea is to be really random. Otherwise, a small sample can’t be representative. I think…
The ideal is simple random sampling of a group that is large enough that it is extremely likely to closely match the population in all respects (not just demographics). Some bodies might have smaller optimal sizes, or have many people decline to serve, so random stratified sampling might be done (as is done for polling). There are competing ideas about the best way to do a lottery, but all need to be random to avoid self-selection bias or organizer manipulation. This book is mainly about why elections can't be a dominant tool in a democracy, and while I touch on sortition design in Part III, it will certainly take years of trial and error to perfect it. I explain my preferred design in a paper here:
https://delibdemjournal.org/articles/abstract/10.16997/jdd.156/
I read the government design paper you linked. Pretty thorough, with lots of lotteries. I have a couple questions. One, do the policy juries need consensus? Or is it majority or some other percentage? And two, are the decisions up or down decisions? Or some other kind of decision? I’m thinking that sometimes, an up or down decision, like we make on initiatives or ballot measures here in Oregon, is not a very good way to form policy. It’s often much more nuanced.
The nuance, refinement and deliberation happens in prior lottery bodies perhaps with super majorities or consensus, etc... i insist that the body that drafts and refines a bill NOT be the body that gets to make the final decision. Pride of authorship and information cascades or groupthink can distort good decision making. a new jury with fresh eyes needs to hear all the pro and con arguments and then decide.. But whether a final draft bill becomes a law or is rejected needs a final yes/no vote. So as not to unfairly favor the status quo, I assume this would be a majority vote... but the rules panel might also choose to adopt a rule that a vote with a margin of less than X % needs a second jury to take a look, and see if they make the same decision. One point is that with a lottery selected rules committee, this democracy would be subject to un-ending improvement, based on lived experience with the system.
Couldn’t find English errors this time. But I have to comment on the idea that our legislature doesn’t represent us because they don’t belong to all the demographic groups. You do mention this in the text, but it’s of course possible for an old rich white guy to represent the ideas of the majority of their district. There are also many real examples of people of power in minority demographic groups (for instance Clarence Thomas) who don’t represent their demographic group very well.
You are certainly correct, but there is much more to this (as I will discuss in later chapters) having to do with diversity of experience, cognitive style and latent knowledge. An individual man may do an excellent job working for the interests of women. But when extrapolated over an entire body of mostly men, the odds of optimal representation go down and down.
Ok, agreed, but what I was getting at was that it’s obvious that our Congress isn’t like us, and that they don’t represent us well. But it’s not imperative that our congressmen and women look like us or reside in our demographic groups to make good policy, especially in the international arena, which they should be focusing on. But good deliberation is necessary. (It's hard to write down my thoughts – this is a very deep and wide subject area, but hopefully you get the idea, and will be talking about this in subsequent chapters.)
Another question... With sortition, if you include a demographic component, do you lose the random nature of the process, and hence lose the statistical basis for its efficacy? I thought in polls, the whole idea is to be really random. Otherwise, a small sample can’t be representative. I think…