Proof of Immortality III

[jt512]

- I'd say that ~H does not imply that all of physics is wrong -- it implies that Newtonian Physics and Relativity don't have all the answers. I'd also say that Quantum Mechanics seems to be going in my direction -- i.e., consciousness is becoming a big deal.

You'd say that because you don't know anything about quantum mechanics.

 

[John Jones]

You'd say that because you don't know anything about quantum mechanics.

All woo-slingers eventually glom onto quantum something or other.

 

[JayUtah]

All woo-slingers eventually glom onto quantum something or other.

...as taught by spiritualist speakers to lay audiences. Good heavens, we don't even start teaching it to physics majors in college until their second year, and only after the substantial mathematics prerequisites are solidly mastered. And we don't expect many to get it right the first time.

 

[John Jones]

...as taught by spiritualist speakers to lay audiences. Good heavens, we don't even start teaching it to physics majors in college until their second year, and only after the substantial mathematics prerequisites are solidly mastered. And we don't expect many to get it right the first time.

That's about right. It took me months to solve the simplest Schroedinger wave equation with any sort of understanding.

 

[The Sparrow]

That's about right. It took me months to solve the simplest Schroedinger wave equation with any sort of understanding.

There is a Sam Harris vs Deepak Chopra debate on youtube somewhere, and a key moment when Sam Harris says we don't know everything yet. At this, Deepak loudly exclaims Thankyou Thankyou. In other words, woo slingers will constantly find a gap into which they can wedge their beliefs because they absolutely positively have not 100% with absolute certainty been disproven.

If their is even a 1/1,000,000,000 of a chance there is room for a woo explanation, that's as good as 100% certainty their woo is true to them.

 

[JayUtah]

If their is even a 1/1,000,000,000 of a chance there is room for a woo explanation, that's as good as 100% certainty their woo is true to them.

That's undoubtedly what Jabba is doing here. Only he's misusing Bayesian methods to suggest that the tiny fragment of uncertainty that it can be is tantamount to objective proof that it must be. He's frankly said so in so many words.

 

[JoeMorgue]

Woo Slingers love "quantum" because to the layperson it sounds like "Hey this sciencey stuff says that weird, unusual, or highly improbable things can happen, so you can just claim anything weird, unusual, or highly improbably is 'quantum' and those big mean skeptics can't say anything because it their precious science saying that."

It's near perfect for use with the "Foot in the Door" tactic that Jabba is using.

 

[Loss Leader]

- Remember, if I'm right about 7,000,000,000/∞ (or even 7,000,000,000/1X10^-80), I've already won.

That's a big "if."

I'd say that ~H does not imply that all of physics is wrong -- it implies that Newtonian Physics and Relativity don't have all the answers.

Your ~H is completely borked. You have no way of separating your preferred scenario from the infinite number of other possible things that are ~H. Give me a good reason to suspect that you have an immortal soul rather than that everything in the universe necessarily had to come about given the starting conditions? If we live in a clockwork universe, then the odds of you coming into existence always were 1. Why is your ~H any less stupid than mine?

 

[jt512]

jt,
- The following is how I calculated the likelihood of E given ~H:

11.3. Re P(E|~H):
11.3.1. The probability (“likelihood”) of E given ~H, involves several specific hypothetical possibilities.
11.3.1.1. That only some of us have but one finite life.
11.3.1.2. That we each have numerous finite lives.
11.3.1.3. That only some of us have numerous finite lives.
11.3.1.4. That we each have an infinity of finite lives.
11.3.1.5. That only some of us have an infinity of finite lives.
11.3.1.6. That we each have an infinite life.
11.3.1.7. That only some of us have an infinite life.
11.3.1.8. That time isn’t what we think it is (to be explained).
11.3.1.9. Some other explanation.

11.3.2. Now I must estimate (roughly) the prior probability (rounded off to three decimal places) of each more specific possibility (hypothesis), given ~H.
11.3.2.1. That only some of us have but one finite life: .000
11.3.2.2. That we each have numerous finite lives: .002.
11.3.2.3. That only some of us have numerous finite lives: .000.
11.3.2.4. That we each have an infinity of finite lives; .002
11.3.2.5. That only some of us have an infinity of finite lives: 000.
11.3.2.6. That we each have an infinite life: .002
11.3.2.7. That only some of us have an infinite life: .000
11.3.2.8. That time isn’t what we think it is (to be explained): .002
11.3.2.9. Some other explanation: .002

11.3.3. And now, I must estimate the likelihood of my own current existence given the different specific hypotheses under ~Hs.
11.3.3.1. That only some of us have but one finite life: .10.
11.3.3.2. That we each have numerous finite lives: .10.
11.3.3.3. That only some of us have numerous finite lives: .25.
11.3.3.4. That we each have an infinity of finite lives; 1.00
11.3.3.5. That only some of us have an infinity of finite lives: .50.
11.3.3.6. That we each have an infinite life: 1.00
11.3.3.7. That only some of us have an infinite life: .50
11.3.3.8. That time isn’t what we think it is (to be explained): .50
11.3.3.9. Some other explanation: .50
11.3.4. And now, I must multiply each of the probabilities of ~H above by the likelihoods of my current existence, given each specific hypothesis, and add up their products. And, the total likelihood of my current existence given ~H:
11.3.4.1. P(E|~H) = (0*.5) + (.002*.10) + (0*.25) + (.002*1.0) + (0*.5) + (1*..002) + (0*.5) + (.002*.5) + (.002*.5), or
11.3.4.2. P(E|~H) = 0 + .0002 + 0 +.002 + 0 + .002 + 0 + .001 + .001, or
11.3.4.3. P(E|~H) = .0052. And,
11.3.5. P(H|E) = 0*.99/(0*.99 + .0052*.01) = (0/.000052) = 0.
11.3.6. P(H|E) = 0.
11.3.7. IOW, given my current existence, the posterior probability that I will have one, finite life is zero…

Alas, after stating that I didn't want to know how you calculated P(E|~H), I read your explanation anyway. Although the following correction doesn't matter, because (1) none of the numbers enters into the calculation of the posterior, since your P(E|H)=0, and thus the 0 probability of your posterior is a foregone conclusion, and (2) all your numbers are just made up, I feel pathologically pedagogically compelled to point out that the prior probabilities 11.3.2.1–11.3.2.9 need to be normalized to sum to 1. After doing so, P(E|~H)=.62 (not .0052). Thus under your pet hypothesis, or ensemble of hypotheses, you had a 62% chance (strangely enough) of having come into existence.

 

[Mojo]

That's a big "if."

And it's also wrong, because it relies on the Texas sharpshooter fallacy, the prosecutor's fallacy, equivocation between very large numbers and infinity, and (as you note in the rest of the post) a false dilemma.

 

[Jabba]

Alas, after stating that I didn't want to know how you calculated P(E|~H), I read your explanation anyway. Although the following correction doesn't matter, because (1) none of the numbers enters into the calculation of the posterior, since your P(E|H)=0, and thus the 0 probability of your posterior is a foregone conclusion,

jt,
- Again, correct.

and (2) all your numbers are just made up,

- They are, in fact, estimates -- that's how Bayes works. Each of us is free to project our own estimates, and to judge the plausibility of the other estimates.

I feel pathologically pedagogically compelled to point out that the prior probabilities 11.3.2.1–11.3.2.9 need to be normalized to sum to 1. After doing so, P(E|~H)=.62 (not .0052). Thus under your pet hypothesis, or ensemble of hypotheses, you had a 62% chance (strangely enough) of having come into existence.

- Unfortunately, they need to be normalized to the sum of .01, rather than 1.

 

[jt512]

- They are, in fact, estimates -- that's how Bayes works. Each of us is free to project our own estimates, and to judge the plausibility of the other estimates.

They are not estimates. In statistics, an estimate is an inference about an unobservable quantity. Your likelihoods are not inferences. Statistical inferences are derived from data and statistical models. Your likelihoods are not implied by any statistical models; they are just numbers you've made up.

Unfortunately, they need to be normalized to the sum of .01, rather than 1.

Fortunately or unfortunately, they (i.e., the prior weights) need to be normalized to 1. Your "estimate" of P(E|~H) is too low by a factor of 119: a factor of 100 due to failure to normalize and a factor of 19 due to an arithmetic error.

Until now, I had failed to perceive your smugness and your arrogance. Now that I have, adios.

 

[jt512]

Alas, after stating that I didn't want to know how you calculated P(E|~H), I read your explanation anyway. Although the following correction doesn't matter, because (1) none of the numbers enters into the calculation of the posterior, since your P(E|H)=0, and thus the 0 probability of your posterior is a foregone conclusion

jt,
- Again, correct.

You just admitted that you haven't been doing Bayesian inference all along, but rather, just rationalizing a foregone conclusion. P(E|H)=0 means that the evidence is impossible under H; therefore, H has been refuted; not made statistically improbable, but logically refuted. In other words your argument is really just:

If H, then not E.
E.
Therefore, ~H.

The argument boils down to "I exist; therefore, I have an immortal soul."

This ends the thread, right?

 

[caveman1917]

You don't have to normalize it.

P(H|E) / P(~H|E) = P(H) / P(~H) * P(E|H) / P(E|~H)

Any normalization constant would cancel because of the division anyway. Yes, a probability function is defined as summing to 1, that doesn't mean it's always necessary though, specifically when we're only interested in relative probabilities.

 

[jt512]

You don't have to normalize it.

P(H|E) / P(~H|E) = P(H) / P(~H) * P(E|H) / P(E|~H)

Any normalization constant would cancel because of the division anyway. Yes, a probability function is defined as summing to 1, that doesn't mean it's always necessary though, specifically when we're only interested in relative probabilities.

No. The normalization I'm talking about does not cancel out. P(E|~H) should be .62, not .0052.

 

[Jabba]

No. The normalization I'm talking about does not cancel out. P(E|~H) should be .62, not .0052.

jt,
- Showing my ignorance, and lack of recent statistical study, I now agree with .62 rather than .0052.
- I'll have to think more about the need to normalize -- but for now, I agree with you on that also...

 

[Jabba]

You just admitted that you haven't been doing Bayesian inference all along, but rather, just rationalizing a foregone conclusion. P(E|H)=0 means that the evidence is impossible under H; therefore, H has been refuted; not made statistically improbable, but logically refuted. In other words your argument is really just:

If H, then not E.
E.
Therefore, ~H.

The argument boils down to "I exist; therefore, I have an immortal soul."

This ends the thread, right?

jt,
- I tried to explain my reasoning back in #2132:
- Here, I'm pretty sure you're wrong.
- If an hypothesis excludes the possibility of a particular event, and the event occurs, you can throw out that hypothesis. You don't have to worry about Bayes.
- If an hypothesis almost excludes a particular event, and the event occurs, you can't just throw out that hypothesis -- you need to weigh the likelihood of said event, given the original hypothesis, against the prior probability of the complement of the original hypothesis...
- That's where Bayes comes in. In our case, OOFLam only almost excludes the event -- the zero is the result of rounding off to three digits. - Though, I probably should have used 1X10-80in the formula instead of ∞.

 

[JoeMorgue]

- If an hypothesis excludes the possibility of a particular event, and the event occurs, you can throw out that hypothesis. You don't have to worry about Bayes.

Jabba do us all a favor and just stare at this sentence for a long, long time.

 

[jsfisher]

Jabba,

You are the subject of a statistics exercise. There you are, an existing human being with all the characteristics that go along with that. Would you agree we could estimate the probability of you having blue eyes?

Certainly, we could.

Now, if we became aware of your ancestry, say for example, you were of Swedish descent, we could apply Bayes Theorem to this additional knowledge to refine our probability estimate of you having blue eyes, right?

We all agree, right?

Now, you are the subject of another statistics exercise. There you are, an existing human being with all the characteristics that go along with that. Would you agree we could estimate the probability of you existing?

What would that probability estimate be?

 

[Loss Leader]

- That's where Bayes comes in. In our case, OOFLam only almost excludes the event -- the zero is the result of rounding off to three digits. Though, I probably should have used 1X10-80in the formula instead of ∞.[/I]

Zero is not the same as very small.