I don’t think you should be optimizing to avoid extreme views, but in favor of those with the most robust models, who can also communicate them effectively to the desired audience. I agree that if we’re going to be trying anything resembling public outreach it’d be good to have multiple voices for a variety of reasons.
On the first half of the criteria I’d feel good about Paul, Buck, and Luke. On the second half I think Luke’s blog is a point of evidence in favor. I haven’t read Paul’s blog, and I don’t think that LessWrong comments are sufficiently representative for me to have a strong opinion on either Paul or Buck.
I don’t think you should be optimizing to avoid extreme views, but in favor of those with the most robust models, who can also communicate them effectively to the desired audience. I agree that if we’re going to be trying anything resembling public outreach it’d be good to have multiple voices for a variety of reasons.
On the first half of the criteria I’d feel good about Paul, Buck, and Luke. On the second half I think Luke’s blog is a point of evidence in favor. I haven’t read Paul’s blog, and I don’t think that LessWrong comments are sufficiently representative for me to have a strong opinion on either Paul or Buck.