{"items": [{"author": "James", "source_link": "https://www.facebook.com/jefftk/posts/775488981742?comment_id=775497689292", "anchor": "fb-775497689292", "service": "fb", "text": "How likely do you think it is that cryonics is proven effective, becomes massively popular, and then excellent technical and legal infrastructure becomes commonplace?

How likely do you think it is that we have a positive singularity and reach a post-scarcity economy where many of these things become non-issues?

This is the same issue that Nate Silver had with Trump. When you're conditioning on the hospital not refusing access, you're giving some mass to the world where cryonics is ubiquitous.", "timestamp": "1458190882"}, {"author": "Ben", "source_link": "https://www.facebook.com/jefftk/posts/775488981742?comment_id=775498682302", "anchor": "fb-775498682302", "service": "fb", "text": "I saw this approach used to estimate the probability terrorists could obtain a nuclear weapon based on looking at all the individual stages of the process. It said (page 14) a 1 in 3.4 billion chance with generous 1 in 3 odds for terrorists o accomplish each stage of the process. This always struck me as seemling too low http://politicalscience.osu.edu/fac.../jmueller/apsachgo.pdf", "timestamp": "1458191798"}, {"author": "Frederic", "source_link": "https://www.facebook.com/jefftk/posts/775488981742?comment_id=775526741072", "anchor": "fb-775526741072", "service": "fb", "text": "Back when I was doing predictions for the Good Judgement Project this is something that the top forecasters would use all the time. I don't recall it being thought inaccurate and the superforecasters were all pretty sharp cookies who were empirically good at making predictions.", "timestamp": "1458218295"}, {"author": "Jeff Kaufman", "source_link": "https://www.facebook.com/jefftk/posts/775488981742?comment_id=775526741072&reply_comment_id=775529021502", "anchor": "fb-775526741072_775529021502", "service": "fb", "text": "→ It could be that the technique only works well for experienced calibrated people, and attempts to apply it my non experts give worse results than just \"give your best guess without a model\".

(Though that's not my expectation.)", "timestamp": "1458219795"}, {"author": "Frederic", "source_link": "https://www.facebook.com/jefftk/posts/775488981742?comment_id=775528178192", "anchor": "fb-775528178192", "service": "fb", "text": "Your base rates for \"all people die\" and \"society falls apart\" are quite alarming!", "timestamp": "1458219133"}, {"author": "Jeff Kaufman", "source_link": "https://www.facebook.com/jefftk/posts/775488981742?comment_id=775528178192&reply_comment_id=775528871802", "anchor": "fb-775528178192_775528871802", "service": "fb", "text": "→ Yeah, pretty alarming!", "timestamp": "1458219640"}, {"author": "Mark", "source_link": "https://www.facebook.com/jefftk/posts/775488981742?comment_id=775531940652", "anchor": "fb-775531940652", "service": "fb", "text": "For the nate silver thing, remember there were 17 republican candidates. If they all had equal likelihood, that's 6% each. Given how unprecedented a trump nomination would be 2% seems reasonable to me, even in retrospect.", "timestamp": "1458221430"}, {"author": "Frederic", "source_link": "https://www.facebook.com/jefftk/posts/775488981742?comment_id=775535004512", "anchor": "fb-775535004512", "service": "fb", "text": "Ok having looked at your numbers a little more closely -- I have a few quibbles. I think the risk of global extinction or collapse is lower than what you predict. I also note the \"other\" in your predictions -- that implies a cumulative 75% chance of cryonics failing for reasons you haven't thought of which seems high. On the technological side you assign a pretty high probability to the freezing process failing to work properly/ failing to capture sufficient data and also to cryonic revival not being technologically feasible-- these seem hard to differentiate for me.", "timestamp": "1458222630"}, {"author": "Jeff Kaufman", "source_link": "https://www.facebook.com/jefftk/posts/775488981742?comment_id=775535004512&reply_comment_id=775555518402", "anchor": "fb-775535004512_775555518402", "service": "fb", "text": "→ The numbers in http://www.jefftk.com/p/breaking-down-cryonics-probabilities were my first crack at this, and I updated some of the numbers later on in http://www.jefftk.com/p/more-cryonics-probability-estimates

\"I think the risk of global extinction or collapse is lower than what you predict\"

My current guess is that this is a combined chance of ~40% over the timescale of \"now until whenever cryonics works if it does work\". There are a lot of things to be nervous about: pandemics, nuclear disasters, bioengineering, asteroids, extreme catastrophic climate change, war escalation, etc.

\"On the technological side you assign a pretty high probability to the freezing process failing to work properly/ failing to capture sufficient data and also to cryonic revival not being technologically feasible-- these seem hard to differentiate for me\"

The difference I'm trying to capture is whether the information gets preserved at all. For example, if the cryoprotectant profusion isn't good enough you might end up with ice crystals in the middle of the brain, totally disrupting the freezing and losing the information for good. On the other hand, you might get a perfect preservation where you could even extract locally the values of individual neuron connections, but Moore's law stops or something and doing this for the whole brain is just too large a scale for people to ever manage.

(I think the information not being properly preserved by the current process is a very large issue at this point, much more than when I first looked into this. In order to win the small mammal brain preservation prize [1] the winning team followed a protocol that's much better than what cryonics organizations currently do, and preservations done with the current process weren't good enough. Additionally I've seen several [2][3][4] neuroscientists arguing that this method is insufficient.)

[1] http://www.brainpreservation.org/small-mammal-announcement/

[2] http://lesswrong.com/.../neil_degrasse_tyson_on.../6krm

[3] https://www.facebook.com/.../permalink/10153146707115333/...

[4] http://scienceblogs.com/.../how-can-you-protect-a-brain.../", "timestamp": "1458230515"}, {"author": "Todd", "source_link": "https://www.facebook.com/jefftk/posts/775488981742?comment_id=775535792932", "anchor": "fb-775535792932", "service": "fb", "text": "Has Yudkowsky provided a breakdown of exactly what makes your post \"absolutely ridiculous and egregious\" somewhere else? It's clear that you made a good faith effort, and it's probably unnecessary for him to be that harsh regardless, but he certainly shouldn't do it without backing it up by going into detail. It's not as if this is the sort of thing that's obvious to most people.", "timestamp": "1458223196"}, {"author": "Jeff Kaufman", "source_link": "https://www.facebook.com/jefftk/posts/775488981742?comment_id=775535792932&reply_comment_id=775539066372", "anchor": "fb-775535792932_775539066372", "service": "fb", "text": "→ He talked about this some in the comments here http://lesswrong.com/.../more_cryonics_probability.../ but my reading of his FB post is that it's intended to stand alone.", "timestamp": "1458224651"}, {"author": "Jim", "source_link": "https://www.facebook.com/jefftk/posts/775488981742?comment_id=775724589582", "anchor": "fb-775724589582", "service": "fb", "text": "I think some amount of outside view is helpful here. Compare to the other broken-down probability estimates you cite in the newer post. Your estimate is more than two orders of magnitude away from the median and more than one order of magnitude away from the next most extreme. That's a pretty serious red flag!", "timestamp": "1458311278"}, {"author": "David Chudzicki", "source_link": "https://www.facebook.com/jefftk/posts/775488981742?comment_id=775724589582&reply_comment_id=775725527702", "anchor": "fb-775724589582_775725527702", "service": "fb", "text": "→ Not convinced that set of people is an effective \"outside view\".", "timestamp": "1458311698"}, {"author": "Jim", "source_link": "https://www.facebook.com/jefftk/posts/775488981742?comment_id=775724589582&reply_comment_id=775732708312", "anchor": "fb-775724589582_775732708312", "service": "fb", "text": "→ Maybe, but those were the only people whose probability estimates were cited.", "timestamp": "1458313326"}, {"author": "Jim", "source_link": "https://www.facebook.com/jefftk/posts/775488981742?comment_id=775732044642", "anchor": "fb-775732044642", "service": "fb", "text": "I think your estimate does have some amount of failing to condition on the previous things. For example, you have:

The technology is never developed to extract the information - 0.5
The technology is never developed to run people in simulation - 0.4
It is too expensive to extract your brain's information - 0.3

But these are linked to a common factor of general tech progress which makes them correlated; eg, most of the cases where the technology is developed to extract the information, it's molecular nanotech, which solves the other two.

Here's another correlated-looking group of events in your list:

Some law is passed that prohibits cryonics before you die - 0.1
Some time after you die cryonics is outlawed - 0.1
No one is interested in your brain's information - 0.3
Running people in simulation is outlawed - 0.4

And another:

All cryonics companies go out of business - 0.2
The cryonics company you chose goes out of business - 0.3
Your cryonics company screws something up and you are defrosted - 0.3

This is supposed to be conditioned on cryonics not being outlawed and society not falling apart. A company's cryopreserved people getting defrosted is very likely to cause them to go out of business. (There's also a missing disjunctive path around the second one: cryonics companies have indicated willingness to take over maintenance of each others' cryopreserved patients if they go out of business.)

Somewhat more nitpicky, you have:

Some law is passed that prohibits cryonics before you die - 0.1

This feeds into a probability which feeds into a cost-benefit analysis, but it would mean you stop paying, so it subtracts as much from the cost side as it does from the benefit side. (The older breakdown had two more items with this property in it contributing a 21% combined failure chance.)", "timestamp": "1458313261"}, {"author": "Eliezer", "source_link": "https://www.facebook.com/jefftk/posts/775488981742?comment_id=775732044642&reply_comment_id=775755332972", "anchor": "fb-775732044642_775755332972", "service": "fb", "text": "→ But he said he did condition his estimates, so obviously there's no problem", "timestamp": "1458321008"}, {"author": "Jim", "source_link": "https://www.facebook.com/jefftk/posts/775488981742?comment_id=775732044642&reply_comment_id=775756121392", "anchor": "fb-775732044642_775756121392", "service": "fb", "text": "→ Indeed he did. I presented these groups of estimates as evidence that he didn't look hard enough for correlations that conditional probabilities need to account for.", "timestamp": "1458321547"}, {"author": "David Chudzicki", "source_link": "https://www.facebook.com/jefftk/posts/775488981742?comment_id=775782453622", "anchor": "fb-775782453622", "service": "fb", "text": "One possible disjunctive path (which branches after everything about whether it's possible to run you in simulation) could involve creating a trust to fund getting your data out and running you in simulation. Although I don't think people do this in practice.", "timestamp": "1458330384"}, {"author": "Jim", "source_link": "https://www.facebook.com/jefftk/posts/775488981742?comment_id=775782453622&reply_comment_id=775786275962", "anchor": "fb-775782453622_775786275962", "service": "fb", "text": "→ That trust fund is *much* less likely to survive than your remains are. In particular, you'd run into problems with the rule against perpetuities.", "timestamp": "1458331690"}, {"author": "David Chudzicki", "source_link": "https://www.facebook.com/jefftk/posts/775488981742?comment_id=775782453622&reply_comment_id=775786844822", "anchor": "fb-775782453622_775786844822", "service": "fb", "text": "→ Oh well.", "timestamp": "1458332027"}, {"author": "Jeff Kaufman", "source_link": "https://www.facebook.com/jefftk/posts/775488981742?comment_id=775782453622&reply_comment_id=775788960582", "anchor": "fb-775782453622_775788960582", "service": "fb", "text": "→ @Jim: You might be able to get around that rule by also freezing a zygote: http://www.jefftk.com/p/perpetuities-via-frozen-embryos", "timestamp": "1458332865"}, {"author": "Jeff Kaufman", "source_link": "https://www.facebook.com/jefftk/posts/775488981742?comment_id=775782453622&reply_comment_id=775789529442", "anchor": "fb-775782453622_775789529442", "service": "fb", "text": "→ legal analysis: http://www.jefftk.com/mccrimmon2000.pdf", "timestamp": "1458333117"}, {"author": "Mark", "source_link": "https://plus.google.com/106199220175508895485", "anchor": "gp-1462326018990", "service": "gp", "text": "I think some of your probabilities aren't really accurate conditional probabilities, i.e. you aren't updating \nenough.\n
\n
\nYou've given a 50% chance of \"The technology is never developed to extract the information,\" given that society lasts forever in a good enough state to keep your body successfully preserved, and that the required information is there and possible to access. \n
\n
\nThis seems absurdly low to me, I think most of that 50% of failure is based on technological progress stopping, which we've already mostly ruled out.\n
\n
\nAgain, conditional on cryonics working and being used in a functioning society, you have very high probabilities that reviving you is too expensive/ people don't bother/ etc. This seems like the same problem as the 6 stages of Trump example, you're failing to imagine just how trivial these problems would become in a society \nwhere all the previous hurdles have been passed.", "timestamp": 1462326018}]}