I mean, yes, in a far mode, censorship creates fear and so on… but let’s come back to near mode and ask: “What is the worst consequence of stepping just a little on the wrong side of this uncertain line?”
Well, Eliezer would delete my comment or article, and that’s it. It does not really make my legs shake.
My guess is that “tobacco kills more people than guns and cars combined. Should we ?”, written literally like this, is acceptable. Probability estimate? It would be 98% in a different discussion on a different day, and perhaps 95% here and now because Eliezer may still be in the deleting mood. So what? If I am wrong, he will delete that comment, and perhaps also my comment for quoting it. And that’s all. Am I afraid? No. Actually, I would probably not even notice if that happened.
Generally, I also prefer precise rules to imprecise ones, but there are limits how precise one can be in topics like this. Trying to make the rules exact (to avoid all harm, but allow all harmless discussion) is a FAI-complete problem. Even the real-world laws often have imprecise parts. Also, the more precise rules, the greater pressure on moderators to follow them literally; but I would prefer them using their own judgement.
Trying to make the rules exact (to avoid all harm, but allow all harmless discussion) is a FAI-complete problem.
Eliezer should engage into any deliberate practice that involves solving FAI-complete problems and not shun away from putting thinking into them and leave vague rules because the problem is too hard.
When time passes and the above post is not censored uncertainty will decrease. Arguments about chilling effects due to uncertainty are probably systemically overstated because the about of uncertainty we have now is much higher than the average uncertainty over the life of the policy.
This kind of uncertainty about what is and is not acceptible, is perhaps the primary reason why such censorship policies are evil.
This is a huge exaggeration!
I mean, yes, in a far mode, censorship creates fear and so on… but let’s come back to near mode and ask: “What is the worst consequence of stepping just a little on the wrong side of this uncertain line?”
Well, Eliezer would delete my comment or article, and that’s it. It does not really make my legs shake.
My guess is that “tobacco kills more people than guns and cars combined. Should we ?”, written literally like this, is acceptable. Probability estimate? It would be 98% in a different discussion on a different day, and perhaps 95% here and now because Eliezer may still be in the deleting mood. So what? If I am wrong, he will delete that comment, and perhaps also my comment for quoting it. And that’s all. Am I afraid? No. Actually, I would probably not even notice if that happened.
Generally, I also prefer precise rules to imprecise ones, but there are limits how precise one can be in topics like this. Trying to make the rules exact (to avoid all harm, but allow all harmless discussion) is a FAI-complete problem. Even the real-world laws often have imprecise parts. Also, the more precise rules, the greater pressure on moderators to follow them literally; but I would prefer them using their own judgement.
Eliezer should engage into any deliberate practice that involves solving FAI-complete problems and not shun away from putting thinking into them and leave vague rules because the problem is too hard.
When time passes and the above post is not censored uncertainty will decrease. Arguments about chilling effects due to uncertainty are probably systemically overstated because the about of uncertainty we have now is much higher than the average uncertainty over the life of the policy.
Also the post in question is great.
P.S. fuck the police.