= 27d567bc2d48d9f40e9ccc20ea370b15)
wedrifid comments on Scientific Self-Help: The State of Our Knowledge - Less Wrong
You are viewing a comment permalink. View the original post to see all comments and the full post content.
= 783df68a0f980790206b9ea87794c5b6)
Loading…= b715d02e85f7b3b4ae65ca3d3e450868)
Subscribe to RSS Feed
Powered by Reddit
= f037147d6e6c911a85753b9abdedda8d)
You are viewing a comment permalink. View the original post to see all comments and the full post content.
Comments (493)
Oooh, PU basilisks. Where? Show me!
Are you sure you want me to do that on a public forum? I do not want to have my account deleted for posting dangerous stuff.
Then message me. The concept of a PU basilisk seems unlikely to me but still somewhat intriguing. The closest things I can imagine are in the form of disillusionment with ideals.
It would be interesting to have a collection of basilisks somewhere, like an ammo dump for a mimetic war.
Absolutely! Maybe not on lesswrong, that might make some people cry. But I'd love to have a list somewhere else. And this can be considered an open request to send any basilisk spotted in the wild to me personally for examination.
I've yet to see a basilisk that was remotely intimidating to me. And would like to be able to further improve my resistance to exposure to new 'basilisks' while they are framed as basilisks so I am even less likely to be vulnerable to them in the wild.
A superintelligence would almost certainly be able to construct sentences that could hack my brain and damage it. Some humans could if they were able to put me in a suitable social or physical environment and ensure ongoing exposure (and environment and exposure are far more important than the abstract concepts conveyed). But things like "Roko's Basilisk" are just cute. You can tame them and keep them as pets. :)
My perspective on this is very similar to yours. If you were sent any interesting PU basilisks, would you please forward them to me?
Being careful not to criticise what was sent to me (and by so doing discourage others) I don't think what I was sent fits the term 'basilisk'. Instead I got some well thought out considerations and potential pitfalls that people may fall into along a PUA journey. In fact I think people would appreciate them being spoken publicly. Rather than "things that will kill you just by looking at them" they are things that you are better off looking at so you can avoid falling into them. Obviously the exceptional case is the pessimistic person who is looking for excuses not to try - which is not an uncommon mindset.
I would not repost them here (that would be discourteous) but I suggest that if the author did post his thoughts publicly they would be well received (ie. would get an 8+karma rating if the comment was not buried too deeply to gain exposure.)
I violently agree with all of this. Have you seen any basilisk-like ideas besides roko's? Roko's at least looks like a real basilisk until you think about it. Everything else I've seen doesn't come close to living up to the name.
'Violent agreement' seems to have been adopted for use in situations in which the participants have been arguing aggressively only to discover that they agree on the substantive issues. For a term that hasn't been hijacked as jargon I go with "vehemently". It has a more visceral feel to it too. :)
Roko's is the most interesting I've seen too. Although for some people a combination of Pascal's Wager and certain religious doctrines about children not being held accountable for their beliefs until a certain age would do it. Once again it is the ability to apply abstract reasoning while at the same time the naivety and weakness in following the rational conclusion correctly that would cause the problem.
Anyone have a really solid working definition for 'basilisk' as we use it here?
Am I supposed to be able to see it from just this? Assuming it's not the kind of thing that would hurt LW posters can you explain? Otherwise, pm it?
One interesting idea is that it seems plausible to create basilisks that only effect your memetic/cognitively different enemies- perhaps the only way to avoid the harm of the basilisk is to deconvert from your religion/ideology. A basilisk that only worked on, say, religious fundamentalists would be a really powerful weapon (I'm not suggesting that the basilisk be capable of killing anyone, necessarily).
A quick google on "do young children go to hell?" led me to this, excerpt:
(The scripture quotes in question were totally reaching by the way. But that's the whole point of theology.)
Knowledge or concepts, the comprehension of which will cause one significant disulitly.
That is over-broad unfortunately. The concept needs to be distinguished from "ugly truths" where the disutility comes from an emotional reaction to how far from ideal the world is.
I think I know what wedrifid is getting at, but I don't think Pascal's Wager would do it. Pascal's Wager argues that one should act as if one believes in God because the costs are low and the potential benefits (Heaven) are high.
But in order to get to the particular failure state at which I think wedrifid is hinting, you can't just be betting on God -- you have to be absolutely certain that Heaven exists and that its joys outweigh on every axis everything that Earth has to offer. Most people, no matter what they say, are not that certain, which is why we don't routinely slaughter infants in order to ensure their blameless souls entry into Heaven. (Similar logic has been invoked to rationalize murders--such as innocent deaths at witch trials--but in these cases, as a justification pasted on after the fact rather than an honest motive towards murder.)
No, absolute certainty is definitely not required. The cost is increased so a proportionate increase in
probability*payoffis required. But this is still all dwarfed by the arbitrarily large payoffs inherent in religious questions. The whole point of 'afterlife' focussed doctrine is to encourage the flock to discount all 'earthly' matters as trivial compared to eternal questions.No, that would not be a rational reason to refrain from the slaughter. The difference between 90% sure and absolutely certain isn't really much of a big deal when you have the chance of flipping the sign bit of an arbitrarily large disulility payoff (Hell). A 0.05% hunch would be more than enough.
Rational agents that really have arbitrarily large utility payoffs floating around in their utility function will inevitably do things that look insane to us.
I am sure that this is possible, but wonder why it has not been done yet - or at least appeared on my radar. Might be one of the more darker arts, and a very interesting one!
Creating basilisks is hard- as evidenced by the fact that we have no recorded instance of one ever existing.
No. It would be better to first develop defenses against them. Basilisks seem to only affect people of a certain mental capacity able to understand and process them. If you look up the Charles Langan interview, or his writings, or this Ted/Unabomba guy you see how really bright people can go wrong.
I would hate LW to contribute to that.
I want LWers and myself to not only have a realistic view of reality, but also be able to life in it and be happy and productive.
I'm not sure how you develop defenses to Basilisks without know what they are. Unless we get lucky and there is a fully general countermeasure.
I was just talking about collecting them though- it's another question entirely whether or not the list should be public. One doesn't usually leave ammo dumps unlocked.
And, probably more importantly, without certain other mental capacities that allow them to handle information appropriately.
If you tell wedrifid privately, then you have to promise to tell me.
I have a few minor basilisks (not from PU alone, but from combining PU with psychometrics or feminism). Nothing so bad that I think it would make people want to ban me, but it might be disconcerting and depressing for many people, and some of it I'm still thinking through.
Did the two annihilate each other, destroying swathes of your cerebral cortex?
Or I'll tell you. :)