LessWrong orthodoxy includes a large number of propositions (over a hundred posts in just core sequences, at least one thesis per post)
The deductions that lead to each claim are largely independent (if post B was an obvious corollary of post A, it would have saved writer’s and readers’ time not to write it)
Reasoning is error-prone, especially when not formalized (this is a point made in the sequences; if it’s wrong then q.e.d.)
Even if each deduction is overwhelmingly likely (let’s say 99%) to be correct, it would be likely (63% in this case) that at least one out of a hundred would be incorrect
Because these are deductive chains of reasoning (they’re “the sequences”, not just “the set”), one false deduction can invalidate any number of conclusions which follow from it. The Principle of Explosion has been defeating brilliant people for millennia.
In other words, even if you believe that each item of LessWrong consensus is almost certain to be correct, you should still be doubtful that every item of LessWrong consensus is likely to be correct. And if there are significant errors, then how else will they be found and publicized other than via a controversial discussion?
I agree that there are errors in the “LW consensus.” I agree that a cost-effective mechanism for identifying those errors would be a valuable thing.
By your estimation, how many controversial discussions have occurred on LW in the last year? How many of them have contributed to identifying any of those errors?
Those are both good questions (as is the implicit point about cost-effectiveness or lack thereof); I’m afraid I’m not a heavy enough reader here to quickly give accurate answers.
I’m not looking to you for accurate answers, I’m trying to understand the model you’re operating on. If you tell me you think there have been a few controversial (in the sense you describe above) discussions and you think they’ve contributed to identifying errors, then it makes sense to me that you think having more such discussions is valuable. I may disagree, but it’s clear to me what we’re disagreeing about. If you tell me you don’t think we’ve had any such discussions, I can sort of understanding you believing that they would be valuable if we had them, but I would also conclude I don’t quite know what sorts of discussions you’re talking about. If you tell me you think we’ve had a few such discussions but they haven’t contributed anything, then I would be very confused and want to revisit my understanding of why you believe what you believe. Etc.
Here’s a case for the value of controversy.
LessWrong orthodoxy includes a large number of propositions (over a hundred posts in just core sequences, at least one thesis per post)
The deductions that lead to each claim are largely independent (if post B was an obvious corollary of post A, it would have saved writer’s and readers’ time not to write it)
Reasoning is error-prone, especially when not formalized (this is a point made in the sequences; if it’s wrong then q.e.d.)
Even if each deduction is overwhelmingly likely (let’s say 99%) to be correct, it would be likely (63% in this case) that at least one out of a hundred would be incorrect
Because these are deductive chains of reasoning (they’re “the sequences”, not just “the set”), one false deduction can invalidate any number of conclusions which follow from it. The Principle of Explosion has been defeating brilliant people for millennia.
In other words, even if you believe that each item of LessWrong consensus is almost certain to be correct, you should still be doubtful that every item of LessWrong consensus is likely to be correct. And if there are significant errors, then how else will they be found and publicized other than via a controversial discussion?
I agree that there are errors in the “LW consensus.”
I agree that a cost-effective mechanism for identifying those errors would be a valuable thing.
By your estimation, how many controversial discussions have occurred on LW in the last year?
How many of them have contributed to identifying any of those errors?
Those are both good questions (as is the implicit point about cost-effectiveness or lack thereof); I’m afraid I’m not a heavy enough reader here to quickly give accurate answers.
I’m not looking to you for accurate answers, I’m trying to understand the model you’re operating on.
If you tell me you think there have been a few controversial (in the sense you describe above) discussions and you think they’ve contributed to identifying errors, then it makes sense to me that you think having more such discussions is valuable. I may disagree, but it’s clear to me what we’re disagreeing about.
If you tell me you don’t think we’ve had any such discussions, I can sort of understanding you believing that they would be valuable if we had them, but I would also conclude I don’t quite know what sorts of discussions you’re talking about.
If you tell me you think we’ve had a few such discussions but they haven’t contributed anything, then I would be very confused and want to revisit my understanding of why you believe what you believe.
Etc.