Far from being innately evil aaronsw appears to be acting just like any reasonably socially competent human with some debate skills can be expected to act when they wish to persuade people of something. It just so happens that doing so violates norms against bullshit, undesired forms of rhetoric and the use of arguments as soldiers without applying those same arguments to his own position.

Forget "innately evil". In fact, forget about the author entirely. What matters is that the post and the reasoning contained therein is below the standard I would like to see on lesswrong. Posts like it need to be weeded out to make room for better posts. This includes room for better reasoned criticisms of SIAI or lesswrong, if people are sufficiently interested (whether authored by aaronsw or someone new).

Hm. Maybe you're right that I'm giving him too much credit just because he's presenting a view unpopular on LW. (Although, come to think of it, having a double standard that favors unpopular conclusions might actually be a good idea.) In any case, it looks like he rewrote his post.

If you sincerely believe that the optimal use of a FAQ on an organisations website is to argue against your own conclusions then I suggest that you have a fundamentally broken model of how the world---and people---work. It would be an attempt at counter-signalling that would fail abysmally. I'd actually feel vicarious embarrassment just reading it.

I think the optimal use of an FAQ is to give informed and persuasive answers to the questions it poses, and that an informed and persuasive answer will acknowledge, steel-man, and carefully refute opposing positions.

I'm not sure why everyone seems to think the answers to the questions in an FAQ should be short. FAQs are indexed by question, so it's easy for someone to click on just those questions that interest them and ignore the rest. lukeprog:

the linear format is not ideal for analyzing such a complex thing as AI risk

...

What we need is a modular presentation of the evidence and the arguments, so that those who accept physicalism, near-term AI, and the orthogonality thesis can jump right to the sections on why various AI boxing methods may not work, while those who aren't sure what to think of AI timelines can jump to those articles, and those who accept most of the concern for AI risk but think there's no reason to assert humane values over arbitrary machine values can jump to the article on that subject.

I even suggested creating a question-and-answer site as a supplement to lukeprog's proposed wiki.

I don't fault SI much for having short answers in the current FAQ, but it seems to me that FAQs are ideal tools for presenting longer answers relative to other media.

One option is for each question in the FAQ to have a page dedicated to answering it in depth. Then the main FAQ page could give a one-paragraph summary of SI's response along with a link to the longer answer. Maybe this would achieve the benefits of both a long and and a short FAQ?