Why Sleeping Beauty Is Lost inwards Time (by Pradeep Mutalik),has joined those who debate that both widespread answers to the Sleeping Beauty problem, \(P=1/2\) as well as \(P=1/3\), are right as well as they respond 2 unlike questions. The halfers (such as your humble correspondent) exercise assist almost the "background" of events patch the thirders (which includes fans of the radical anthropic regulation as well as Boltzmann Brains) don't.
Sleeping beauty thirders' rudimentary Markoff chain error partly dedicated to an declaration yesteryear swain halfer Bob Walters. H5N1 fundamental signal is that whenever nosotros order that "some belongings \(A\) holds for/in the part of spacetime nosotros just inhabit", this probability \(P(A)\) must e'er live identified amongst the probability of the evolution\[
P(BB\to A)
\] from the "Big Bang" (which is my somewhat arbitrary get upwards for the "last trial nosotros tin live 100% sure about") to the present. The development operator also depends on the information almost how much fourth dimension the Universe evolves as well as where the part where \(A\) is evaluated sits.
But the implicit supposition that there's some "universal initial state" is necessary because it's the fact that\[
P(BB) = 1
\] that is the source of the dominion that the nitty-gritty of the probabilities of all the mutually exclusive outcomes must e'er live 100%. This basic dominion needed for the consistency of probabilities e'er arises because some "certainty almost a minute inwards the past" evolves into an "uncertainty" at a after time. The probabilities are e'er slices of a "pie" that had to live baked before nosotros could await at the pieces.
The laws of Nature – including e.g. the constabulary that a fair money lands heads-or-tails at 50-50 odds – are the primary laws that create well-defined numbers, as well as those are e'er probabilities of an before state's development to a after state. On the other hand, all of our "reconstructions" of the yesteryear are examples of a "reverse engineering" that requires Bayesian inference. And at that spot can't live whatever additional "simply yet quantitatively exact" laws that could allow you lot to assign probabilities of an before province from the noesis of a after state.
The thirders mostly ignore the logical arrow of time, inwards i agency or another, which "allows" them to destination upwards amongst all the "wonderful" answers such as \(P=1/3\) or the claim that nosotros should live afraid of beingness Boltzmann Brains.
In the correct, halfers' probability calculus, it's obvious why \(P=1/2\). There was the Big Bang whose probability was 100%. Afterwards, a money was tossed as well as the 100% pie was divided to 2 slices, heads (50%) as well as tails (50%). When the Sleeping Beauty is woken up, she doesn't larn whatever novel information – it was guaranteed that she would live awakened at to the lowest degree once, as well as that's the solely thing that she observes – as well as thus these 50%-50% slices remain. At most, she may enquire what is the twenty-four hr menstruation assuming that the money is showing tails. She may split the piece 50% for tails to 2 smaller slices, inwards the symmetric instance 25% for tails-Monday as well as 25% for tails-Tuesday.
But solely a 50% piece was "reserved" for tails when the money was tossed as well as she can't "inflate" the size of this piece proportionally to the set out of awakenings because the awakenings occurred after the money toss. For her, on Mon or Tuesday, to modify the 50-to-50 sectionalisation of the pie that took house on Dominicus would hateful to modify her past. But i just cannot modify the past! So whatever details almost the experiment (the twenty-four hr menstruation that the Sleeping Beauty experiences) are asked after the money toss just has to operate amongst the pieces of the pie that were previously reserved for "tails".
One may compute to a greater extent than complicated probabilities that are ratios whose numerators as good as denominators are smaller-than-100-percent pieces of the master pie, 100%. But i tin never assume that the pie grows bigger than 100%. Thirders constantly commit this fallacy because they basically assume that they have got 300% of a pie to operate amongst if at that spot are 3 possible (coin_state, day) combinations. You just can't ever start amongst a 300% full probability. One may estimate probabilities yesteryear a measurement, inwards the frequentist way, as \(N/N_{\rm total}\), but these are e'er just measurements of something that withal require the noesis of the laws of Nature to deduce something interesting – as well as the laws of Nature e'er have got 100% as the denominator for probabilities.
Moreover, if you lot desire to determine a probability as \(P=N/N_{\rm total}\) inwards the frequentist way, you lot must truly stair out some \(N,N_{\rm total}\). The Sleeping Beauty doesn't stair out whatever \(N,N_{\rm total}\), as well as thus she just can't role whatever "frequentist" formula for the probability. It's that simple! Thirders are "imagining" some über-observer who picks some random ensemble inwards a whole spacetime as well as defines "probabilities" as some relative sizes of subsets of this set. But that's never the pregnant of a probability. \(N/N_{\rm total}\) is solely interpreted as the "approximately measured probability", inwards the frequentist way, if i has a command over the initial conditions, repeats them \(N_{\rm total}\) times, as well as counts how many times, \(N\), a belongings holds. The Sleeping Beauty doesn't have got a command over the initial weather condition of the experiment – she can't laid upwards it \(N_{\rm total}\) times (e.g. because her retention is erased) – as well as thus she just cannot role the frequentist formula for the probability. Instead, she has to role Bayesian inference as well as assume the laws of Nature (the fairness of the coin).
Whenever person else has the command over the "preparation of the experiment", you lot just can't translate the ratio \(N/N_{\rm total}\) as the probability. Readers of newspapers inwards Democratic People's South Korea can't translate the fraction of articles proverb that "North Korea is the most prosperous dry soil inwards the world" as the probability that Democratic People's South Korea is the most prosperous dry soil inwards the world. They may have got read it 100,000 times inwards the newspapers but they were non preparing the ensemble of \(N_{\rm total}\) experiments yesteryear themselves which makes it illegitimate to assume that the ratios are representative of the actual probabilities.
If nosotros convey Matulik's explanation of thirders' thinking seriously as well as if nosotros add together some erstwhile linguistic communication of your humble correspondent, thirders typically imagine that all copies or events involving the Sleeping Beauty (or Boltzmann Brains) inwards the whole spacetime define the "ensemble", the denominator inwards the fractions that quantify the probabilities, but it's just never possible.
If all 3 awakening events inwards the whole spacetime (or all lives of the Boltzmann Brains inwards the spacetime) defined the denominator, as well as 100% would live the nitty-gritty of their probabilities, it would include the supposition that all these statements "I am inwards the spacetime locus" are mutually exclusive amongst each other as well as they have got an extra predetermined metaphysical constabulary (e.g. the claim that they're as likely) that determines what their probabilities truly are.
But that's impossible inwards Nature because it's the laws of physics that truly dictate the right relationships betwixt all these probabilities. So at that spot just cannot live another constabulary on peak of them that would impose additional relationships betwixt the probabilities. You may either believe that at that spot be laws of Nature, or you lot may believe that all awakenings (or all lives of the Boltzmann Brains) are as likely, but you lot just can't believe both because the spousal human relationship is logically contradictory.
Let me offering you lot a random novel occupation to brand the point.
Cloning babes
Imagine that on Monday, at that spot is i babe. Every evening, every babe's retention from the twenty-four hr menstruation is erased as well as at midnight, she's cloned into 2 copies. So at that spot are 2 babes on Tuesday, 4 babes on Wednesday, ... as well as 64 babes on Sunday. When a infant is awakened, she's asked what is the probability that it's Monday. On the next week, the babes are allowed to last (although I originally had to a greater extent than drastic plans) but they're assured that the week-long experiment is over.
I have got eliminated all randomness, coins, as well as dice.
It seems obvious to me that the thirders volition respond \(P=1/127\). In total, at that spot are 1+2+...+64 arrangements "which babe, which day", as well as solely the maiden off i one of them lives on Monday. For an anthropic, Boltzmann-Brain-loving thirder, all these options are as likely, because they define the "experiments" as well as the set out of elements inwards this laid determines the denominator from which they calculate their "probabilities".
Halfers know that this "equal probability" supposition is inconsistent amongst whatever laws of physics.
When a infant is created yesteryear splitting her "mother" yesteryear cloning, the piece of her probability pie is beingness divided to halves, too. If the infant has no other information or bias concerning the "current date", it's truly right for her to assume that the probability of Mon is \(P=1/7\). The probability that you're a exceptional infant who lives on Th is 50% of the probability that you're "her cloning mother" who lives on Wednesday. The probability pie for the babe's spacetime identity has to live cutting from the pre-given slices.
The departure betwixt the halfers' (now, seventhers') right respond as well as the thirders' (now, one-hundred-twenty-seventhers') wrong respond grows exponentially large if you lot prolong the experiment. If the babes are allowed to alive as well as split for 30,000 days (or infinitely many days), thirders would order that the probability it's the maiden off twenty-four hr menstruation is \(P=1/2^{30,000}\) or so: they could basically (or strictly) eliminate the gamble that it's the maiden off twenty-four hr menstruation – or any other early on day. Because they solely bargain amongst the inverse proportionality, halfers volition non eliminate it.
There can't live a probability calculus consistent amongst the laws of physics that would allow you lot to eliminate the probability that "we're closed to the maiden off generations" because this reasoning is acausal. (This reasoning would also live catastrophic for scientific discipline because it would allow you lot to attempt out that cosmology, geology, development of species, or whatever "historical" natural scientific discipline is wrong because nosotros can't perchance live finitely closed to the of import early on events.) The declaration "we're at most the \(X\)-th generation" has a probability that just cannot live influenced yesteryear the assumptions almost the glorious futurity of the mankind: whatever such influence would live acausal (the futurity affects the past). The laws of physics are chronologically causal (the yesteryear affects the future) as well as if you lot assumed both methods to determine the probabilities, you lot would have got the equivalent of closed time-like curves – the paradoxes that ruin naive science-fiction movies amongst fourth dimension travel.
A modification of the Sleeping Beauty problem
Let me supply to the Sleeping Beauty occupation as well as modify it inwards i to a greater extent than way. Even if the money lands tails, the Sleeping Beauty volition solely live interviewed i time – on Mon or on Tuesday. The twenty-four hr menstruation of the interview is decided yesteryear some other money toss right away after the maiden off i when the maiden off i gives tails (on Sunday).
So when the Sleeping Beauty wakes up, she tin run into that there's no interview, as well as right away determine that the maiden off money landed tails. However, when the interview begins, she is uncertain whether the maiden off money toss was "heads" or "tails" again. In that case, what is her correctly calculated subjective probability of heads? Her one-day retention is withal erased on Mon night.
With this twist, the results of halfers as well as thirders acquire sort of interchanged.
Now, when she sees that the interview starts, the Sleeping Beauty does have a novel information, as well as that matters. Up to the minute when she learned whether the interview started, she had learned no novel (nonzero) information since Dominicus when she was told almost the rules of the game.
But as presently as an interview \(I\) starts, she is seeing an trial that was predicted to convey house amongst the probability \(P(I|H)=1\) inwards the instance of heads, as well as \(P(I|T)=1/2\) inwards the instance of tails. Because the "heads theory" made a to a greater extent than self-confident prediction of the observation that was just made (by a factor of two), its relative odds must have got increased yesteryear a factor of 2 relatively to the "tails theory" that was uncertain whether an interview should begin.
So i time the interview begins, a halfer decides – yesteryear Bayes' theorem – that the heads-to-tails odds are 2-to-1 as well as \(P(H)=2/3\).
What almost the thirders' answer? Well, thirders depict a spacetime diagram as well as imagine that the same week-long experiment takes house on many weeks. The set out of interviews on the heads weeks is exactly the same as the set out of interviews on the tails weeks, as well as that's why the naively spacetime-frequentist logic of the thirders leads them to order \(P(H)=1/2\). Nominally, thirders acquire out "halfers" but that shouldn't confuse you: they are withal as stupid.
Why is it stupid? Because inwards this modified experiment, the logic of the master Sleeping Beauty occupation got reverted. In the master problem, the Sleeping Beauty learned no nontrivial information when the interview started but thirders behaved as if some information emerged. Now, it's the other agency around: the Sleeping Beauty clearly did larn some novel information when the interview started – because it didn't have got to start – but the thirders seem to completely ignore the information.
The set out of heads interviews as well as tails interviews may live the same if the same experiment is repeated many times (for many weeks). But the symmetry betwixt "heads" as well as "tails" is clearly broken. It is non difficult to run into that the same wrong thinking of the thirders may atomic number 82 them to ignore whatever bear witness or all evidence inwards the world. If it doesn't bear on their heads-tails odds if they larn almost some trial that was predicted to live much to a greater extent than probable yesteryear i hypothesis than some other hypothesis, they may ignore whatever other evidence, too. I believe that for whatever variety of bear witness \(E\) inwards whatever situation, you lot could laid upwards analogous modifications of the Sleeping Beauty occupation that shows that the thirders ignore the bear witness \(E\).
For thirders, whenever the probability of a hypothesis \(H\) is suppressed yesteryear a factor of \(K\) yesteryear the bear witness but they desire \(H\) to live right, anyway, they may compensate this fact yesteryear thinking almost the possibility that \(H\) is right, after all, \(K\) times. In that way, they increment the "number of events inwards the spacetime that follow from \(H\)" yesteryear a factor of \(K\), as well as \(H\) looks as skillful as it did before.
In 2012, ATLAS as good as CMS announced the 5-sigma regain of the Higgs boson, i.e. nominally a 99.9999% probability that a novel Higgs-like particle around \(125\GeV\) exists. H5N1 thirder may holler back almost the development of the information inwards a Higgs-less Universe 1 1000000 times (and solely i time almost the Higgs-ful Universe). The ensemble (the denominator of his "version of a probability") is filled amongst lots of Higgs-less histories, as well as thus the Higgs vs no-Higgs ratio stays the same.
I guess that everyone at to the lowest degree amongst some traces of a encephalon understands why this approach to the Higgs bear witness is illogical or biased. Whether the Higgs exists or non was discovered billions of years agone as well as because of basic causality (cause precedes its effects) you lot just can't bear on the odds of this yesteryear determination inwards the Universe – patch the details of your "measurements of the probability" would plainly influence the result.
But thirders seem to misunderstand why their approach to the Sleeping Beauty occupation is isomorphic to the lightheaded error (or scam) that "allowed" the guy to ignore the Higgs evidence. The frequentist measuring of the probabilities is solely acceptable if the experimenter's lineament standards are imposed. The experimenter has to have got the command over the training of the initial province as well as the physical care for must live actively repeated \(N_{\rm total}\) times. When the ensemble of \(N_{\rm total}\) events is obtained differently, this \(N_{\rm total}\) can't live considered to live a denominator of whatever notion of a probability.
Because every awakening of the Sleeping Beauty follows after a money toss that is the solely controllable maiden off of the experiment, it's the set out of money tosses (and non the set out of awakenings) that must deed as \(N_{\rm total}\).
Supreme Court odds
Mutalik offers a funnily stupid "analogy" explaining why he says that both halfers as well as thirders are right. Up to recently, six as well as 3 the U.S. of A. Supreme Court judges were Harvard as well as Yale Law School alumni, respectively, as well as thus the thirders' reasoning gave (accidentally the same) numerical value \(P=1/3\) for the probability that a randomly chosen electrical flow Supreme Court justice is a Yale alumnus. On the other hand, Mutalik claims that the halfers would count the odds that some random Americans acquire to Yale – they would imitate the past, Mutalik thinks – which is why the halfers would conclude that \(P\ll 1/3\) because at that spot are many schools etc.
But this is ludicrous. When nosotros verbalize almost a "judge randomly chosen from the electrical flow Supreme Court" as well as we're given the composition, nosotros are clearly given totally unlike input information than inwards the different occupation inwards which the members of the Supreme Court are unknown as well as have got to live "simulated" yesteryear next some random students' motility through the the U.S. of A. schools institutions. The results are unlike because the problems are self-evidently different. Halfers would for sure concord that if a justice is uniformly randomly chosen from the specified Supreme Court, nosotros would acquire \(P=1/3\) for Yale.
On the other hand, inwards the Sleeping Beauty problem, at that spot is no "contrast" betwixt these 2 real unlike problems. The final information that the Sleeping Beauty truly knows is the information almost the invention of the experiment she was told almost on Sunday. Whether it's "analogous" to the information almost demographics of the the U.S. of A. or the composition of the electrical flow Supreme Court can't live said clearly. Analogies tin live constructed inwards many unlike ways. However, the signal is that inwards the Sleeping Beauty problem, it's completely unambiguous what she's truly allowed to assume almost the coin, herself, as well as the people around her. There is no ambiguity inwards the occupation that would live analogous to the ambiguity inwards the Supreme Court example!
What may await analogous betwixt the thirders' flawed Sleeping Beauty thinking as well as the \(P=1/3\) Yale thinking is that the thirders define their ensembles inwards the "future" relatively to where the halfers "cut the pie". But it's just wrong to brand assumptions almost her probabilities to live inwards i awakening or some other – just because she isn't getting whatever information almost those as well as at that spot is no constabulary of physics that would brand them as probable etc. In the Supreme Court case, the random fellow member of the courtroom is obtained yesteryear running a (pseudo)random generator, some trial that produces set out 1-9 amongst equal probabilities, as well as those determine which justice nosotros pick. But inwards the Sleeping Beauty case, at that spot is just no trial inwards which a random generator yields a set out betwixt 1-3 that would determine which (coin_state, day) combination she is just experiencing. So there's just no rational footing for her to conclude that the 3 combinations are as likely. The actual rational justification exists for \(P(H)=P(T)\) because of the toss of the fair money on Sunday. And inwards the instance of tails, at that spot may also live a rational footing for \(P(T,Mon)=P(T,Tues)\) because the 2 days "feel the same" inwards the given history of the Universe.
But the egalitarian assumptions of the style\[
P(H,Mon) = P(T,Tues)
\] comparison the probabilities of loci inwards 2 different possible histories are e'er wrong – this variety of egalitarianism is always an illustration of mixing apples as well as oranges. The pie is sliced to alternative histories inwards the past, during the money toss inwards this case, as well as no extra "quotas" on the equality of unlike probabilities across the spacetime and/or across its alternative possible histories tin ever live correct.
No comments:
Post a Comment