LessWrong 2.0 Reader

and makes it satisfy the desires of all the things which ought be happy

Some things don't endorse themselves being happy as an important consideration.

Believing QI is the same as a Bayesian update on the event "I will become immortal".

Imagine you are a prediction market trader, and a genie appears. You ask the genie "will I become immortal" and the genie answers "yes" and then disappears.

Would you buy shares on a Taiwan war happening?

If the answer is yes, the same thing should apply if a genie told you QI is true (unless the prediction market already priced QI in). No weird anthropics math necessary!

Merging states would be far superior incentive-wise, as the influence of every signatory would increase and there would be essentially no expected net shift in the election results.

Suppose I am an Okalahoman who will always vote for the farther right candidate or a Californian who will always vote for the farther left candidate. Wouldn't I only support this if my odds on this increasing the chance of the farther right (resp. farther left) candidate winning was at least 50%? Wouldn't most people fall in one of these two categories?

I think this issue has been discussed at length and repeatedly on LW, leading to a weak consensus that at least strong moral realism isn't true.

Can anyone supply links to some other good posts on the topic?

I think that if they:
- Find a feature that is necessary for defection in all of the cases of defection in a diverse and realistic set of model organisms
- Locate that feature with very little optimization on the defection cases
I'd be fairly confident (>~80%) that it would catch any particular deployment-time defection attempt from an ASL-4 schemer. This would cut risk from the deployment by >5x when paired with compelling incrimination and a shutdown. This is a substantial risk reduction, and I might even buy this safety case if there weren't better options for this threat model (e.g. untrusted monitoring). My main concerns are:
- It seems unlikely we would get a feature that meets the above two criteria
- The feature might have too many false positives
- The possibility that the model organisms were sabotaging your eval by coordinating to route their defection attempts through a particular feature (this seems hard at ASL-4 though (but this claim would need an eval)).

See [V1.1]

This made me unreasonably happy. Thank you :D

links 11/07/2024: https://roamresearch.com/#/app/srcpublic/page/11-07-2024

• on Donna Karan
  • https://www.vogue.com/slideshow/donna-karan-seven-career-highlights-cold-shoulder
  • https://www.vogue.com/article/donna-karan-vintage
• https://du42p.r.a.d.sendibm1.com/mk/mr/sh/1f8JAEjGcfF85pENVqcuM6hh5D/tiVMw3KFimvC?fbclid=IwY2xjawGZnpRleHRuA2FlbQIxMQABHdHszEMzS3x1Xda2tTh60KMighJEJKDe30rsduQmFydSSyfTpK8mwG50vg_aem_IjjUQWdiwZf2AdgdLo9azA opinion on how hormonal contraception should be done differently -- I'm intrigued but I haven't yet checked these claims out
• http://esr.ibiblio.org/?p=8720 Eric S. Raymond on "user stories" done right and wrong
• https://endpts.com/biotech-industry-worries-over-potential-for-rfk-jr-ally-as-fda-pick/ Casey Means has been floated as the new pick for FDA head; apparently she's expressed concerns about vaccines and over-medication on the Joe Rogan podcast and has written a book about how most chronic diseases can be prevented by healthy lifestyles (which probably overstates the case)
• https://www.slowboring.com/p/the-tyranny-of-climate-targets Matt Yglesias on why a lot of aggressive climate targets are impossible to actually meet.
  • why do people try anyway? if it's "cheap talk", why is there so much costly, substantive follow-through? incentive misalignment, I suppose?
• https://www.gordian.bio/blog/the-in-vivo-screening-revolution/ Martin Borch Jensen on in-vivo screening 

Note that you can make untrusted monitoring work without the synthetic inputs too, you just have to do the honeypots during deployment instead of before deployment.

I am an American who knows what Estonia is, and I found the joke hilarious.