these are four of the seven most important themes on the site in terms of immediate advice about what to do
What are the other three? And shouldn’t there be an explanation why they are excluded from this outside view analysis? (EDIT: See Mass_Driver’s explanation here.)
let’s call it ‘Omega’ instead of God
Please call it something else? Using ‘Omega’ seems unnecessarily confusing given that there’s already a convention for using that name to denote a powerful and trustworthy (but not necessarily Friendly) entity in decision theory problems.
I seem to have worded the bit about “four out of seven” poorly—it’s just meant to be a sort of confidence interval, but people seem to think I’m jealously hoarding my pile of three extra advice themes which I (for some reason) refuse to share with you all. I don’t know what they are. If I had three more themes, I’d have listed them, and then said I had “seven out of twelve” or something like that. It’s precisely because what’s salient for me may not be so for others that I’m trying to be humble about how many of the top themes I’ve managed to identify.
I don’t agree with the list of “most prominent themes in terms of short-term behavioral advice” mention in the original post, but I also don’t think that it is completely unfounded:
55 upvotes on a comment of someone donating the the current balance of his bank account.
15 upvotes on a request for more discussions of general cognitive enhancing tools such as Adderall and N-Back.
42 upvotes on a post that claims that you are a lousy parent if you don’t sign up your kids for cryonics.
Eliezer telling people to put hope into cryonics and advanced nanotechnology rather than “Noble Lies”.
Right now, as I’m writing this, someone coming across the LW home page would have some grounds to conclude that LW is advising:
reading interesting books or articles with quotable material
attending meetups
introspective exercises on why we reject some actions
watching your thoughts and words
brainstorming exercises
measuring your aversions
writing (or maybe reading) horoscopes
(I’m omitting posts which appear to be purely informational.)
Over the course of the next few weeks, this list will change until new content has entirely replaced the old; at that point if you asked again the question “what is LW advising” you’d see something different, maybe with substantial overlap with the above list, maybe not.
So that is one procedure to (attempt to) determine what are LW’s major themes of short-term advice.
My point is that different procedures may yield different results, for different readerships.
Cryonics comes up every so often, but may or may not be perceived as a major theme—depending on how you read LW.
ETA: if you’re going to count all-time upvotes, then it would make more sense for me to do a systematic survey: rank all posts by number of upvotes, possibly normalize by how long ago the material has been posted (more recent material has had less time to accumulate upvotes), extract from each post what advice it gives if applicable. What seems to be going on for both you and the OP is that you rank as “major” the things that have struck you the most. They may have struck you the most precisely because they were most unconventional, in which case you will come to unsound conclusions.
This doesn’t prove anything, but I thought it was interesting. You can conduct your own searches, what results do you anticipate on a site like lesswrong if it cares most strongly about rationality and much less about topics like AI and cryonics?
What are the other three? And shouldn’t there be an explanation why they are excluded from this outside view analysis? (EDIT: See Mass_Driver’s explanation here.)
Please call it something else? Using ‘Omega’ seems unnecessarily confusing given that there’s already a convention for using that name to denote a powerful and trustworthy (but not necessarily Friendly) entity in decision theory problems.
An excellent point. Can you suggest a better title? I could call it the “Singularity” story, but that would be a bit unfair as well.
“Sysop”.
Question seconded. I’d also like to ask how you came to your conclusions as to the “themes of short-term advice” encountered on LW.
What is most salient or most available for you may not be so for everyone else or even most of the readership.
I seem to have worded the bit about “four out of seven” poorly—it’s just meant to be a sort of confidence interval, but people seem to think I’m jealously hoarding my pile of three extra advice themes which I (for some reason) refuse to share with you all. I don’t know what they are. If I had three more themes, I’d have listed them, and then said I had “seven out of twelve” or something like that. It’s precisely because what’s salient for me may not be so for others that I’m trying to be humble about how many of the top themes I’ve managed to identify.
I don’t agree with the list of “most prominent themes in terms of short-term behavioral advice” mention in the original post, but I also don’t think that it is completely unfounded:
55 upvotes on a comment of someone donating the the current balance of his bank account.
15 upvotes on a request for more discussions of general cognitive enhancing tools such as Adderall and N-Back.
42 upvotes on a post that claims that you are a lousy parent if you don’t sign up your kids for cryonics.
Eliezer telling people to put hope into cryonics and advanced nanotechnology rather than “Noble Lies”.
Let me try to clarify what I mean.
Right now, as I’m writing this, someone coming across the LW home page would have some grounds to conclude that LW is advising:
reading interesting books or articles with quotable material
attending meetups
introspective exercises on why we reject some actions
watching your thoughts and words
brainstorming exercises
measuring your aversions
writing (or maybe reading) horoscopes
(I’m omitting posts which appear to be purely informational.)
Over the course of the next few weeks, this list will change until new content has entirely replaced the old; at that point if you asked again the question “what is LW advising” you’d see something different, maybe with substantial overlap with the above list, maybe not.
So that is one procedure to (attempt to) determine what are LW’s major themes of short-term advice.
My point is that different procedures may yield different results, for different readerships.
Cryonics comes up every so often, but may or may not be perceived as a major theme—depending on how you read LW.
ETA: if you’re going to count all-time upvotes, then it would make more sense for me to do a systematic survey: rank all posts by number of upvotes, possibly normalize by how long ago the material has been posted (more recent material has had less time to accumulate upvotes), extract from each post what advice it gives if applicable. What seems to be going on for both you and the OP is that you rank as “major” the things that have struck you the most. They may have struck you the most precisely because they were most unconventional, in which case you will come to unsound conclusions.
site:lesswrong.com “artificial intelligence” = 30,700 results
site:lesswrong.com rationality = 13,500 results
site:lesswrong.com “Singularity” = 32,000 results
site:lesswrong.com bias = 5,230 results
site:lesswrong.com “cryonics” = 1,680 results
site:lesswrong.com bayes = 1,660 results
site:lesswrong.com “evolutionary psychology” = 804 results
site:lesswrong.com “Bayes’ theorem” = 689 results
This doesn’t prove anything, but I thought it was interesting. You can conduct your own searches, what results do you anticipate on a site like lesswrong if it cares most strongly about rationality and much less about topics like AI and cryonics?
Thought this was because of the logo at the top of the page, so searched for “Singularity Institute for Artificial Intelligence” and got:
site:lesswrong.com “Singularity Institute for Artificial Intelligence” = 111,000 results
So something’s weird. Also, if you move “site:lesswrong.com″ to the right side you get 116,000 instead.
Google’s result counter is an estimate, and not a very good one. It’s within 2 or 3 orders of magnitude… usually.
You’re right.
Or maybe those result counts don’t measure what you think they measure.
What is most salient or most available for you may not be so for everyone else or even most of the readership.