Percentages are probably not the best way to elicit well-calibrated guesses about very probable or very improbable events. (The difference between 1⁄1,000 and 1⁄1,000,000 is a lot bigger in reality than it looks, when you put them both between 0 and 1 on a scale of 0 to 100.)
Computing P(Many Worlds) requires assuming that the phrase “Many Worlds” refers to a specific set of concrete predictions about the nature universe, which admit the possibility of truth or falsity. I tend to disagree with that presumption.
P(Anti-Agathics) seems, from the name, not to be intended to include cryonics, but does seem to include it in the actual text. I predict paradoxical answers in which people give P(Cryonics) > P(Anti-Agathics), even though cryonics is a way of allowing a person alive today to reach the age of 1000 years.
P(Simulation) may or may not actually be a well-defined question. If, as some people are surely visualizing while answering it, there are aliens somewhere hovering over a computer terminal with us running on it, certainly the answer is ‘yes’. Whatever the reality, it seems likely to be a lot stranger than that. Eliezer’s own “Finale of the Ultimate Meta Mega Crossover” describes a scenario (admittedly fanciful) in which one would be hard pressed to answer the “simulation” question with a simple yes or no.
Percentages are probably not the best way to elicit well-calibrated guesses about very probable or very improbable events. (The difference between 1⁄1,000 and 1⁄1,000,000 is a lot bigger in reality than it looks, when you put them both between 0 and 1 on a scale of 0 to 100.)
None of the questions in the survey sound to me like ones where one could easily get more than 99% sure of (outside an argument)
Eliezer’s own “Finale of the Ultimate Meta Mega Crossover” describes a scenario (admittedly fanciful) in which one would be hard pressed to answer the “simulation” question with a simple yes or no.
Surveyed!
Thanks for putting this together.
Perceived flaws:
Percentages are probably not the best way to elicit well-calibrated guesses about very probable or very improbable events. (The difference between 1⁄1,000 and 1⁄1,000,000 is a lot bigger in reality than it looks, when you put them both between 0 and 1 on a scale of 0 to 100.)
Computing P(Many Worlds) requires assuming that the phrase “Many Worlds” refers to a specific set of concrete predictions about the nature universe, which admit the possibility of truth or falsity. I tend to disagree with that presumption.
P(Anti-Agathics) seems, from the name, not to be intended to include cryonics, but does seem to include it in the actual text. I predict paradoxical answers in which people give P(Cryonics) > P(Anti-Agathics), even though cryonics is a way of allowing a person alive today to reach the age of 1000 years.
P(Simulation) may or may not actually be a well-defined question. If, as some people are surely visualizing while answering it, there are aliens somewhere hovering over a computer terminal with us running on it, certainly the answer is ‘yes’. Whatever the reality, it seems likely to be a lot stranger than that. Eliezer’s own “Finale of the Ultimate Meta Mega Crossover” describes a scenario (admittedly fanciful) in which one would be hard pressed to answer the “simulation” question with a simple yes or no.
None of the questions in the survey sound to me like ones where one could easily get more than 99% sure of (outside an argument)
Thanks for the spoiler. ;-|