Here’s a different hypothesis that also accounts for opinions reverting in the direction of the original uneducated position. Suppose “uneducated” and “contrarian” opinion are two independent random (e.g. normal) variables with the same mean representing the truth (but maybe higher variance for “uneducated”); and suppose what you call “meta-contrarian” opinion is just the truth. Then if you start from “contrarian” it’s more likely that “meta-contrarian” opinion will be in the direction of “uneducated” than in the opposite direction, simply because “uneducated” contains nonzero information about where the truth is. I think you can also see this as a kind of regression to the mean.
I don’t see why we should expect the random variables to be based around “truth”. I’d believe in a common centerpoint, but I think it would be more usefully labeled “human-intuitive position” than “truth”.
It seems to me that uneducated-person opinion would be the “human-intuitive position”, and the educated person opinion would be… changed off from that, with a tendency to be somewhere in the direction of truth.
Although the uneducated-person opinion won’t always be constant across different times and cultures (although it varies; Death is Bad is probably a universal for 5-year-olds, Racism might be pretty universal outside of isolated groups with populations too small to have any racial diversity), so I don’t think it will usually be an inherent position.
I think steven0461′s statement makes some sense though if you talk about the average position among many different issues, and also if you only look at issues where things like evidence and human reasoning can tell you about the truth. I expect that the uneducated opinion will appear distributed randomly around the truth (although not absurdly far from it, when you consider the entirety of possibility-space), and the educated opinion will diverge from it in a way that will usually be towards what the evidence supports, but often overshooting, undershooting, or going far to the side. Likewise the 3rd-stage opinion should diverge from the educated opinion in a similar manner, except… by definition it will be in the rough direction of the original position, or we’d just call it a more radical version of the educated position.
However there seems to be a MAJOR potential pitfall in reasoning about where they’re located, since all the examples listed tend to align politically (roughly conservative/liberal/LessWrong-type). So trying to reason by looking at those examples and seeing which one is true, and then trying to derive a theory on the tendencies involved based on that, will tend to give you a theory which supports your position being right.
I think that this only works if positions are in one dimension. If they are in many dimensions then I suspect that the truth and the uneducated opinion are on the same side of the contrarian opinion as often as they are on opposite sides.
EDIT: I no longer think the above makes any sense. I’m tired, sorry!
I’m a little sad that I’ve integrated this pretty thoroughly into my epistemology because it’s a very good point and yet most people probably missed this comment.
Thank-you for commenting and bringing this to my attention. This also makes for a fantastic “shut down the contrarian” response when your meta-contrarianism is questioned.
Here’s a different hypothesis that also accounts for opinions reverting in the direction of the original uneducated position. Suppose “uneducated” and “contrarian” opinion are two independent random (e.g. normal) variables with the same mean representing the truth (but maybe higher variance for “uneducated”); and suppose what you call “meta-contrarian” opinion is just the truth. Then if you start from “contrarian” it’s more likely that “meta-contrarian” opinion will be in the direction of “uneducated” than in the opposite direction, simply because “uneducated” contains nonzero information about where the truth is. I think you can also see this as a kind of regression to the mean.
I don’t see why we should expect the random variables to be based around “truth”. I’d believe in a common centerpoint, but I think it would be more usefully labeled “human-intuitive position” than “truth”.
It seems to me that uneducated-person opinion would be the “human-intuitive position”, and the educated person opinion would be… changed off from that, with a tendency to be somewhere in the direction of truth. Although the uneducated-person opinion won’t always be constant across different times and cultures (although it varies; Death is Bad is probably a universal for 5-year-olds, Racism might be pretty universal outside of isolated groups with populations too small to have any racial diversity), so I don’t think it will usually be an inherent position.
I think steven0461′s statement makes some sense though if you talk about the average position among many different issues, and also if you only look at issues where things like evidence and human reasoning can tell you about the truth. I expect that the uneducated opinion will appear distributed randomly around the truth (although not absurdly far from it, when you consider the entirety of possibility-space), and the educated opinion will diverge from it in a way that will usually be towards what the evidence supports, but often overshooting, undershooting, or going far to the side. Likewise the 3rd-stage opinion should diverge from the educated opinion in a similar manner, except… by definition it will be in the rough direction of the original position, or we’d just call it a more radical version of the educated position.
However there seems to be a MAJOR potential pitfall in reasoning about where they’re located, since all the examples listed tend to align politically (roughly conservative/liberal/LessWrong-type). So trying to reason by looking at those examples and seeing which one is true, and then trying to derive a theory on the tendencies involved based on that, will tend to give you a theory which supports your position being right.
I think that this only works if positions are in one dimension. If they are in many dimensions then I suspect that the truth and the uneducated opinion are on the same side of the contrarian opinion as often as they are on opposite sides.
EDIT: I no longer think the above makes any sense. I’m tired, sorry!
I’m a little sad that I’ve integrated this pretty thoroughly into my epistemology because it’s a very good point and yet most people probably missed this comment.
Thank-you for commenting and bringing this to my attention. This also makes for a fantastic “shut down the contrarian” response when your meta-contrarianism is questioned.