I find it inconsistent to censor things to protect sensitive people who think about AI but not people who are sensitive to all the other things that are discussed here.
To the extent there is censorship of dangerous information on LW, the danger is to the future of mankind rather then to the (very real and I don’t mean to minimize this) feelings of readers.
One could make the argument that anything that harms the mission of lesswrong’s sponsoring organizations is to the detriment of mankind. I’m not opposed to that argument, but googling censorship of lesswrong did not turn up anything I considered to be particularly dangerous. Maybe that just means that the censorship is more effective than I would have predicted, or is indicative or a lack of imagination on my part.
I’d say that “censorship” (things that could be classified or pattern-matched to this word) happens less than once in a year. That could actually contribute to why people speak so much about it; if it happened every day, it would be boring.
From my memory, this is “censored”:
inventing scenarios about Pascal’s mugging by AI
debating, even hypothetically, harm towards specific people or organization
replying to a downvoted post (automatically penalized by −5 karma)
And the options 2 and 3 are just common sense, and could happen on any website. Thus, most talk about “censorship” on LW focuses on the option 1.
(By the way, if you learned about the “basilisk” on RationalWiki, here is a little thing I just noticed today: The RW article has a screenshot of dozens of deleted comments, which you will obviously associate with the incident. Please note that the “basilisk” incident happened in 2010, and the screenshot is from 2012. So this is not the censorship of the original debate. It is probably a censorship of some “why did you remove this comment two years ago? let’s talk about it forever and ever” meta-threads that were quite frequent and IMHO quite annoying at some time.)
Also, when a comment or article is removed, at least the message about the removal stays there. There is no meta-censorship (trying to hide the fact that censorship happened). If you don’t see messages about removed comments at some place, it means no comments were removed there.
By meta-censorship I meant things like removing the content from the website without a trace, so unless you look at the google cache, you have no idea that anything happened, and unless someone quickly makes a backup, you have no proof that it happened.
Leaving the notices “this comment was removed” on the page is precisely what allowed RW to make a nice screenshot about LW censorship. LW itself provided evidence that some comments were deleted. Providing a hyperlink instead of screenshot would probably give the same information.
Also, I am mentioning basilisk now, and I have above 95% confidence that this comment will not be deleted. (One of the reasons is that it doesn’t get into details; it doesn’t try to restart the whole debate. Another reason is that don’t start a new thread.)
There’s not a lot of actual censorship of dangerous information “for the future of mankind”. Or at least, I rate that as fairly unlikely, given that when the scientific groundwork for a breakthrough has been laid, multiple people usually invent it in parallel, close to each-other in time. Which means that unless you can get everyone who researches dangerous-level AI into LW, censoring on LW won’t really help, it will just ensure that someone less scrupulous publishes first.
“Three may keep a secret, if two of them are dead.”
Conspiracy is hard. If you don’t have actual legal force backing you up, it’s nearly impossible to keep information from spreading out of control—and even legal force is by no means a sure thing. The existence of the Groom Lake air station, for example, was suspected for decades before publicly available satellite images made it pointless to keep up even the pretense of secrecy.
For an extragovernmental example, consider mystery religions. These aren’t too uncommon: they’re not as popular as they once were, but new or unusual religions still often try to elide the deepest teachings of their faiths, either for cultural/spiritual reasons (e.g. Gardnerian Wicca) or because they sound as crazy as six generations of wolverines raised on horse tranquilizers and back issues of Weird Tales (e.g. Scientology).
Now, where’s it gotten them? Well, Gardnerian Wiccans will still tell you they’re drinking from a vast and unplumbed well of secret truths, but it’s trivially easy to find dozens of different Books of Shadows (some from less restrictive breakaway lineages, some from people who just broke their oaths) that agree on the broad strokes and many of the details of the Gardnerian mysteries. (Also many others that bear almost no resemblance beyond the name and some version of the Lesser Banishing Ritual of the Pentagram, but never mind that.) As to Scientology, Operation Clambake (xenu.net) had blown that wide open years before South Park popularized the basic outline of what’s charmingly known as “space opera”; these days it takes about ten minutes to fire up a browser and pull down a more-or-less complete set of doctrinal PDFs by way of your favorite nautical euphemism. Less if it’s well seeded.
“But these are just weird minority religions,” you say? “Knowing this stuff doesn’t actually harm my spiritual well-being, because I only care about the fivefold kisses when my SO’s involved and there’s no such thing as body thetans”? Sure, but the whole point of a mystery religion is selecting for conviction. Typically they’re gated by an initiation period measured in years and thousands of dollars, not to mention some truly hair-raising oaths; I don’t find it plausible that science broadly defined can do much better.
You are clearly right that conspiracy is hard. And yet, it is not impossible. Plenty of major events are caused by conspiracies, from the assassination of Julius Caesar to the recent coup in Thailand. In addition, to truly prevent a conspiracy, it is often necessary to do more than merely reveal it; if the conspirators have plausible deniability, then revealing (but not thwarting) the conspiracy can actually strengthen the plotters hands, as they can now co-ordinate more easily with outside supporters.
Successful conspiracies, like any other social organization, need incentive compatibility. Yes, it’s easy to find out the secrets of the Scientology cult. Not so easy to find out the secret recipe for Coca Cola, though.
I find it inconsistent to censor things to protect sensitive people who think about AI but not people who are sensitive to all the other things that are discussed here.
To the extent there is censorship of dangerous information on LW, the danger is to the future of mankind rather then to the (very real and I don’t mean to minimize this) feelings of readers.
Have you asked the people who are able to censor information on LW, or do you just assume this to be the case?
Do the people in charge of LW censor information that are neither dangerous nor spam?
I infer it’s the case from being a regular reader of LW. I don’t know if LW censors other types of information in part because spam is not a well defined category.
dangerous information on LW, the danger is to the future of mankind
I think that would be far overstating the importance of this forum. If Eliezer/MIRI have some dark secrets (or whatever they consider to be dangerous knowledge), they surely didn’t make it to LW.
To the extent there is censorship of dangerous information on LW, the danger is to the future of mankind rather then to the (very real and I don’t mean to minimize this) feelings of readers.
One could make the argument that anything that harms the mission of lesswrong’s sponsoring organizations is to the detriment of mankind. I’m not opposed to that argument, but googling censorship of lesswrong did not turn up anything I considered to be particularly dangerous. Maybe that just means that the censorship is more effective than I would have predicted, or is indicative or a lack of imagination on my part.
I’d say that “censorship” (things that could be classified or pattern-matched to this word) happens less than once in a year. That could actually contribute to why people speak so much about it; if it happened every day, it would be boring.
From my memory, this is “censored”:
inventing scenarios about Pascal’s mugging by AI
debating, even hypothetically, harm towards specific people or organization
replying to a downvoted post (automatically penalized by −5 karma)
And the options 2 and 3 are just common sense, and could happen on any website. Thus, most talk about “censorship” on LW focuses on the option 1.
(By the way, if you learned about the “basilisk” on RationalWiki, here is a little thing I just noticed today: The RW article has a screenshot of dozens of deleted comments, which you will obviously associate with the incident. Please note that the “basilisk” incident happened in 2010, and the screenshot is from 2012. So this is not the censorship of the original debate. It is probably a censorship of some “why did you remove this comment two years ago? let’s talk about it forever and ever” meta-threads that were quite frequent and IMHO quite annoying at some time.)
Also, when a comment or article is removed, at least the message about the removal stays there. There is no meta-censorship (trying to hide the fact that censorship happened). If you don’t see messages about removed comments at some place, it means no comments were removed there.
And yet earlier in your post you’re talking about some posts in 2012 about censorship in 2010 being deleted. Smells like meta-censorship to me.
By meta-censorship I meant things like removing the content from the website without a trace, so unless you look at the google cache, you have no idea that anything happened, and unless someone quickly makes a backup, you have no proof that it happened.
Leaving the notices “this comment was removed” on the page is precisely what allowed RW to make a nice screenshot about LW censorship. LW itself provided evidence that some comments were deleted. Providing a hyperlink instead of screenshot would probably give the same information.
Also, I am mentioning basilisk now, and I have above 95% confidence that this comment will not be deleted. (One of the reasons is that it doesn’t get into details; it doesn’t try to restart the whole debate. Another reason is that don’t start a new thread.)
There’s not a lot of actual censorship of dangerous information “for the future of mankind”. Or at least, I rate that as fairly unlikely, given that when the scientific groundwork for a breakthrough has been laid, multiple people usually invent it in parallel, close to each-other in time. Which means that unless you can get everyone who researches dangerous-level AI into LW, censoring on LW won’t really help, it will just ensure that someone less scrupulous publishes first.
“Three may keep a secret, if two of them are dead.”
Conspiracy is hard. If you don’t have actual legal force backing you up, it’s nearly impossible to keep information from spreading out of control—and even legal force is by no means a sure thing. The existence of the Groom Lake air station, for example, was suspected for decades before publicly available satellite images made it pointless to keep up even the pretense of secrecy.
For an extragovernmental example, consider mystery religions. These aren’t too uncommon: they’re not as popular as they once were, but new or unusual religions still often try to elide the deepest teachings of their faiths, either for cultural/spiritual reasons (e.g. Gardnerian Wicca) or because they sound as crazy as six generations of wolverines raised on horse tranquilizers and back issues of Weird Tales (e.g. Scientology).
Now, where’s it gotten them? Well, Gardnerian Wiccans will still tell you they’re drinking from a vast and unplumbed well of secret truths, but it’s trivially easy to find dozens of different Books of Shadows (some from less restrictive breakaway lineages, some from people who just broke their oaths) that agree on the broad strokes and many of the details of the Gardnerian mysteries. (Also many others that bear almost no resemblance beyond the name and some version of the Lesser Banishing Ritual of the Pentagram, but never mind that.) As to Scientology, Operation Clambake (xenu.net) had blown that wide open years before South Park popularized the basic outline of what’s charmingly known as “space opera”; these days it takes about ten minutes to fire up a browser and pull down a more-or-less complete set of doctrinal PDFs by way of your favorite nautical euphemism. Less if it’s well seeded.
“But these are just weird minority religions,” you say? “Knowing this stuff doesn’t actually harm my spiritual well-being, because I only care about the fivefold kisses when my SO’s involved and there’s no such thing as body thetans”? Sure, but the whole point of a mystery religion is selecting for conviction. Typically they’re gated by an initiation period measured in years and thousands of dollars, not to mention some truly hair-raising oaths; I don’t find it plausible that science broadly defined can do much better.
So I’m the only one here who actually took a hair-raising oath before making an account?
You’re not allowed to talk about the oath! Why am I the only one who seems able to keep it?
Because there are different factions at work, you naked ape.
Nah, I hear we traditionally save that for after you earn your 10,000th karma point and take the Mark of Bayes.
You probably need to get those 10K karma points from Main.
You are clearly right that conspiracy is hard. And yet, it is not impossible. Plenty of major events are caused by conspiracies, from the assassination of Julius Caesar to the recent coup in Thailand. In addition, to truly prevent a conspiracy, it is often necessary to do more than merely reveal it; if the conspirators have plausible deniability, then revealing (but not thwarting) the conspiracy can actually strengthen the plotters hands, as they can now co-ordinate more easily with outside supporters.
Successful conspiracies, like any other social organization, need incentive compatibility. Yes, it’s easy to find out the secrets of the Scientology cult. Not so easy to find out the secret recipe for Coca Cola, though.
Have you asked the people who are able to censor information on LW, or do you just assume this to be the case?
Do the people in charge of LW censor information that are neither dangerous nor spam?
I infer it’s the case from being a regular reader of LW. I don’t know if LW censors other types of information in part because spam is not a well defined category.
I think that would be far overstating the importance of this forum. If Eliezer/MIRI have some dark secrets (or whatever they consider to be dangerous knowledge), they surely didn’t make it to LW.