Perhaps we could investigate the difference between those good posts, and the bad kind of posts we don’t want to have, and define the rules accordingly.
Seems to me that good things are: long-term experience with using some approach; statistics from sources that seem reliable; a theory that sounds plausible and explains the experience/statistics.
And bad things are: an edgy theory of everything from internet that the author feels strongly about, and is going to test it tomorrow but today we have a passionate article explaining and defending the theory.
I am not saying that this “good” is necessarily always good. People can still draw wrong conclusions from real data. But I think the “bad” things are pretty much guaranteed to have negative value.
In other words, I prefer the ethos “data first, opinion later”.
Here are some posts I would categorize as good based on these criteria:
Anti-social Punishment: Based on empirical literature, draws on author’s extensive familiarity with Slovak culture to help flesh out those studies.
The Virtue of Silence: Providing an original and logical take on several specific, publicly-relevant situations, then generalizing from it.
Ambiguous posts:
Intellectual Hipsters and Meta-Contrarianism: Draws on a little bit of empirical data (the LW survey) and theory (signalling). May stretch the substance a little thin, especially given that it asks you to question why you believe what you believe your most well-thought-out positions, in a way that’s kind of unfalsifiable—because are you just being a meta-meta-contrarian?
Why Our Kind Can’t Cooperate: Weaves together empirical evidence and Bayesian principles with anecdote and intuition pumps; having no single point, it’s hard to argue with or use as a reference.
The Nature of Offense: Sensible and necessary (because it deals with an in-community fracas that’s not going to receive a more formal treatment). But doesn’t engage with any external data or theory.
Edgy theories of everything:
Of Two Minds: Vague references to evo-psych, uses lots of LW- and cherry-picked examples, no attempt at addressing a counterargument.
Antimemes: No data, no concrete examples, conspiracy-ish.
I generally agree with your ordering of quality, but in my opinion both your “good” and “ambiguous” are acceptable.
And the “edgy” are indeed bad, because even if they happen to point to something real and important, they are written in an unhelpful way. (For example the “antimemes” sounds like something from SCP wiki, and is completely one-sided. The idea that some things are useful but ignored regardless, is interesting and potentially very important. But rational discussion would have to include the trade-offs involved, rather than start from the assumption that disagreeing is an obvious mistake. So, although it could be an inspiration to a great discussion, the article per se is not good.)
Perhaps we could investigate the difference between those good posts, and the bad kind of posts we don’t want to have, and define the rules accordingly.
Seems to me that good things are: long-term experience with using some approach; statistics from sources that seem reliable; a theory that sounds plausible and explains the experience/statistics.
And bad things are: an edgy theory of everything from internet that the author feels strongly about, and is going to test it tomorrow but today we have a passionate article explaining and defending the theory.
I am not saying that this “good” is necessarily always good. People can still draw wrong conclusions from real data. But I think the “bad” things are pretty much guaranteed to have negative value.
In other words, I prefer the ethos “data first, opinion later”.
Here are some posts I would categorize as good based on these criteria:
Anti-social Punishment: Based on empirical literature, draws on author’s extensive familiarity with Slovak culture to help flesh out those studies.
The Virtue of Silence: Providing an original and logical take on several specific, publicly-relevant situations, then generalizing from it.
Ambiguous posts:
Intellectual Hipsters and Meta-Contrarianism: Draws on a little bit of empirical data (the LW survey) and theory (signalling). May stretch the substance a little thin, especially given that it asks you to question why you believe what you believe your most well-thought-out positions, in a way that’s kind of unfalsifiable—because are you just being a meta-meta-contrarian?
Why Our Kind Can’t Cooperate: Weaves together empirical evidence and Bayesian principles with anecdote and intuition pumps; having no single point, it’s hard to argue with or use as a reference.
The Nature of Offense: Sensible and necessary (because it deals with an in-community fracas that’s not going to receive a more formal treatment). But doesn’t engage with any external data or theory.
Edgy theories of everything:
Of Two Minds: Vague references to evo-psych, uses lots of LW- and cherry-picked examples, no attempt at addressing a counterargument.
Antimemes: No data, no concrete examples, conspiracy-ish.
I wonder if we agree.
I generally agree with your ordering of quality, but in my opinion both your “good” and “ambiguous” are acceptable.
And the “edgy” are indeed bad, because even if they happen to point to something real and important, they are written in an unhelpful way. (For example the “antimemes” sounds like something from SCP wiki, and is completely one-sided. The idea that some things are useful but ignored regardless, is interesting and potentially very important. But rational discussion would have to include the trade-offs involved, rather than start from the assumption that disagreeing is an obvious mistake. So, although it could be an inspiration to a great discussion, the article per se is not good.)