Anthropics Does Not Work LIke That
post by Pavitra · 2011-08-12T18:52:20.375Z · LW · GW · Legacy · 17 commentsContents
17 comments
People around here seem to think that a recent series of near-misses, such as not destroying the world in the Cold War, is evidence in favor of quantum immortality.
This fails to appreciate that the anthropic selection bias has no limit on how far back it can make things retroactively seem to happen. If, as has been suggested, a majority of the Everett branches from our 1950 destroyed the world, then it is equally true that a majority of the Everett branches from our 1750 in which there is someone still alive in 2010 failed to contain probably-world-destroying technology.
The existence of x-risk near-miss events should be taken as evidence against quantum immortality.
17 comments
Comments sorted by top scores.
comment by Vladimir_Nesov · 2011-08-12T19:38:20.302Z · LW(p) · GW(p)
See Preference For (Many) Future Worlds. "Quantum immortality" is not really a hypothesis that constrains anticipation.
comment by JGWeissman · 2011-08-12T19:06:00.484Z · LW(p) · GW(p)
The development of nuclear weapons that could wipe out humanity is, as you say, evidence against quantum immortality.
Given the development of nuclear weapons, narrowly avoided nuclear wars are evidence in favor of quantum immortality.
The balance of all this evidence is not clear to me. It depends on how strong each piece is, on how much branch measure from 1750 has humans developing nuclear weapons, and how much from 1950 has incidents that could spark a nuclear war.
comment by timtyler · 2011-08-12T19:56:46.319Z · LW(p) · GW(p)
Wikipedia says:
Quantum immortality refers to the subjective experience of surviving quantum suicide regardless of the odds.
You can have evidence against scientific hypotheses - but not against subjective experiences.
So: the notion of "evidence against quantum immortality" does not seem to make very much sense.
If you are trying to say that "quantum immortality" refers to a scientific hypothesis, perhaps you should say which hypothesis you mean.
Replies from: Vladimir_Nesov, JGWeissman, JoshuaZ↑ comment by Vladimir_Nesov · 2011-08-12T21:29:06.876Z · LW(p) · GW(p)
I see three problems with this comment (even though on some level I seem to agree with its intended message):
- Appeal to dictionary definition (essentially), paying close attention to details in authoritative wording. In this case, the wording misses the point, because the relevant features of the problem are not yet visible at this level of detail.
- Stress on the word "scientific" instead of more relevant features of proper/improper hypotheses. See Scientific Evidence, Legal Evidence, Rational Evidence.
- Dismissing subjective experience as domain of proper hypotheses. Even congitive scientists have tools now to perform scientific investigation of subjective experience, and rational hypotheses about subjective experience that are not scientific abound.
↑ comment by JGWeissman · 2011-08-12T21:01:36.657Z · LW(p) · GW(p)
It seems pretty clear to me that "evidence against quantum immortality" refers to evidence against the theory that predicts the subjective experience referred to by "quantum immortality".
It would be good to consider the possibility that someone may use a word somewhat differently than how it is defined on Wikipedia, before telling them that their argument is just incoherent.
Replies from: wedrifid↑ comment by wedrifid · 2011-08-12T21:06:55.204Z · LW(p) · GW(p)
It seems pretty clear to me that "evidence against quantum immortality" refers to evidence against the theory that predicts the subjective experience referred to by "quantum immortality".
That would mean "evidence against quantum physics". (And I'm actually not even sure it would count as evidence against quantum mechanics.)
It would be good to consider the possibility that someone may use a word somewhat differently than how it is defined on Wikipedia, before telling them that their argument is just incoherent.
Tim's reply is not based on semantics.
↑ comment by JoshuaZ · 2011-08-12T20:32:10.391Z · LW(p) · GW(p)
You can have evidence against scientific hypotheses - but not against subjective experiences.
So: the notion of "evidence against quantum immortality" does not seem to make very much sense.
In that case, by conservation of expected evidence the idea of "evidence for quantum immortality" makes about as much sense. In context then, this is just as much of a problem with Pavitra's objection to people interpreting the current state as evidence for.
comment by gwern · 2011-08-12T19:37:56.792Z · LW(p) · GW(p)
If, as has been suggested, a majority of the Everett branches from our 1950 destroyed the world, then it is equally true that a majority of the Everett branches from our 1750 in which there is someone still alive in 2010 failed to contain probably-world-destroying technology.
Equally true? So in other words, the Anthropic Shadow is indefinitely long? I wonder what Nick Bostrom would make of that assertion.
comment by James_Miller · 2011-08-12T23:14:16.941Z · LW(p) · GW(p)
This isn't true but what if I wrote and you believed that "I survived an event that had a 1-10^(-50) probability of killing me." Would this be evidence for you of quantum immortality?
Replies from: Desrtopa, shminux↑ comment by Desrtopa · 2011-08-13T23:23:10.140Z · LW(p) · GW(p)
Unless I'm laboring under a significant misunderstanding, quantum immortality does not predict that one should observe other people surviving against steep odds, unless one's own survival is contingent on it. So it wouldn't favor quantum immortality over no quantum immortality.
↑ comment by Shmi (shminux) · 2011-08-13T00:04:26.658Z · LW(p) · GW(p)
You (well, half of you) have certainly survived an event that kills 99.999999...% of participants, your own conception. Is it an evidence for or against quantum immortality?
Replies from: wedrifid, James_Miller↑ comment by wedrifid · 2011-08-13T10:42:45.916Z · LW(p) · GW(p)
You (well, half of you) have certainly survived an event that kills 99.999999...% of participants, your own conception. Is it an evidence for or against quantum immortality?
Neither. It is the kind of anthropic thinking that would apply even in a classical universe and applies just as much to a quantum one.
↑ comment by James_Miller · 2011-08-13T01:56:13.773Z · LW(p) · GW(p)
Would it matter if I knew about the idea of quantum immortality before surviving the event?
comment by wedrifid · 2011-08-12T22:16:41.077Z · LW(p) · GW(p)
If, as has been suggested, a majority of the Everett branches from our 1950 destroyed the world, then it is equally true that a majority of the Everett branches from our 1750 in which there is someone still alive in 2010 failed to contain probably-world-destroying technology.
I would strengthen this somewhat for emphasis. Obviously it is far more likely in the 1750 case than the 1950 case via a conjunction.
comment by Shmi (shminux) · 2011-08-12T19:24:52.280Z · LW(p) · GW(p)
There is no testable mathematical model of either Everett's branches, or quantum immortality, so it is pointless to argue about the evidence for or against, unless you are a tenured professor who is paid to do so.
Forget testability, there is not even a way to set up reasonable priors and make a Bayesian estimate. The Everett-based models are not less wrong, they are not even wrong.
Replies from: orthonormal↑ comment by orthonormal · 2011-08-13T18:52:37.463Z · LW(p) · GW(p)
Eliezer's take on those objections:
Decoherence is Falsifiable and Testable
Quantum immortality is quite another can of worms, and there's not as much of a LW consensus on the way anthropic probabilities work.
comment by DanielLC · 2011-08-12T19:04:48.919Z · LW(p) · GW(p)
Also, it can change what time we're in. After all, according to timeless physics, the past is just another Everett branch.
It's unlikely that we're in 2010 with world-destroying technology regardless of quantum immortality. Are you sure it's evidence against?