Proof of Immortality, the IV league

[Jedi]

I'm saying that the conditional probability of A given B, P(A|B), is the probability of A under the assumption that B is true.

we're both saying exactly the same thing. you're assuming that B is true. B, in this case, is that God exists and can lead people to water, but the point I'm making is that you're assuming that's true in the data. I'm ok with that in the prior. but assuming that it's true in the data has been the issue, and the only issue, I've had with your approach from pages back. recall, you wrote:

Our data consist of 2 explorers, both of whom got lost, but found water and lived. Thus n=2 and y=2. Plugging those values into the above expression and carrying out the integration, we find that the (marginal) likelihood under H1, L(Y|H1) = .34013.

(I can't include your equation, so i'll write it out)
so the marginal likelihood under the God assumption uses a binomial:

p^y(1-p)^(n-y)

and this is where you include the data you're using, i.e., the two explorers. and the way you include them is to say that y=2 and n=2. n being the sample size of 2 which is fine. but p is the probability of God directing them to water and surviving, and you're squaring it, which means you're treating both explorers' anecdotes as evidence that the God hypothesis is true. the (1-p)^(n-y) is the probability that they survived without God. and since

n=y=2
(1-p)^(2-2) = 1

so this term drops out. you're saying there is no evidence that anyone may have survived outside the God hypothesis as you are accepting the explorers anecdote that God saved them. continuing on...

so including the prior, your p(live|God)=34% (that's correct, I checked). you then divide this by the chance that these two found water by random chance, or p^2=.02^2=.0004 and conclude that God is 850 times more helpful! what if we had 3 explorers? well then p(live|God)=0.255 and the p(live|not God)=0.02^3=.000008. so now we can conclude that God is 32,875 times more helpful than chance alone!

given that your assuming all your data confirms the God hypothesis, the more data you have the more it's confirmed. this is the reason why I keep saying that you're assuming the thing you're trying to show.
and since they're all anecdotes, there are reasons other than God as to why these people survived. many of them. and I can't see any way you can confirm that they survived due to God. you're taking their word for it and you're not including contrary evidence (to say the least).

this is all a very tedious way of asking the same question I've asked before:
why are you assuming that the data you collected of the explorers anecdotes about surviving due to God are true? how do you justify this given it's an anecdote?

*done in haste so please excuse typos, etc.

 

[aleCcowaN]

For those keeping score at home: "OOFLam" is Jabba's strange shorthand term for "only one finite lifetime at most" - that is to say, the idea that you don't get reincarnated and there is no afterlife.

Jabba wants to make an argument for the existence of immortal souls. This "argument", thus far, has consisted of saying that his existence is highly unlikely given OOFLam, because, if he is not an eternal, immortal soul, then any one of a functionally infinite number of other people could have existed instead.

But he exists anyway. So, since his existence is guaranteed if he is an immortal soul, but highly unlikely if he is not, he claims that his existence is evidence for immortality.

Of course, as has already been pointed out, "cheating guarantees a royal flush" is not equivalent to "a given royal flush is evidence of cheating", so his entire argument collapses on its face even without getting into all the nitty-gritty that posters like JayUtah have been dealing with - the fact that his "formula" begs the question, his use of special pleading to deal with the issue of infinite souls, and so on.

Everything about near-death experiences and anecdotes is just a sideshow attempt to try and dredge up whatever hopeless "evidence" he can to support the idea of souls. His call for support from other members is just an appeal to popularity.

Yes, it really is that shallow.

Yes, it really is that silly.

Yes, it really has been this exact same thing for the entire run of the thread.

Let me add these Jabba-shaped threads exist just because the ideas discussed are supposed to be the truth because tradition says so and the believers confirm them independently just because they want them to be true.

And this thread will go on forever (meaning the closing of the forum or the demise of the believer and their transition into eternity, that is, archive.org). And that happens because keeping the thread alive sustains the illusion of the superstitions being discussed might be true. Between replying the most nonsensical **** and not replying at all, **** is the choice because not replying is an admission of all of it being wrong.

So, there's nothing that can make some participants blush because they have an higher goal: keeping the illusion their superstition is a foothold in their climbing to eternity.

Remember: if someone fears death and the annihilation that comes with it, it suffices with posting here some rehash of past mistakes every now an then. As that fears invade periodically their minds, the posts will come ritualistically and they'll contain just a succinct combination of addressing one poster of the opposed field, the repetition of a past mistake and a humble question to some poster regarding something inconsequential to the whole debate. That's the trinity-trifecta of choice for those who believe the universe was created for them to be born in order to enjoy the eternal life as free loaders (that why their sky daddies are almighty, just to not make them look like they're asking to much of life)

 

[jt512]

we're both saying exactly the same thing. you're assuming that B is true. B, in this case, is that God exists and can lead people to water, but the point I'm making is that you're assuming that's true in the data.

I'm not assuming that that is true "in the data." Our data is simply that we have two people who each got lost and each found water and returned. I don't actually use the information that they claim to have been helped by God in the calculation.

and this is where you include the data you're using, i.e., the two explorers. and the way you include them is to say that y=2 and n=2. n being the sample size of 2 which is fine. but p is the probability of God directing them to water and surviving, and you're squaring it, which means you're treating both explorers' anecdotes as evidence that the God hypothesis is true.

I'm not treating their return as evidence of God helping them. I'm separately computing the probability of the observed data under H1 and and H2, and taking the ratio to see which hypothesis better predicts the data.

you are accepting the explorers anecdote that God saved them

As I said, I didn't actually use the fact that they said God saved them in the computation. I only used the fact that they both returned.

so including the prior, your p(live|God)=34% (that's correct, I checked). you then divide this by the chance that these two found water by random chance, or p^2=.02^2=.0004 and conclude that God is 850 times more helpful!

No. I conclude that the data are more consistent with the God hypothesis than with the random hypothesis by a factor of 850.

what if we had 3 explorers? well then p(live|God)=0.255 and the p(live|not God)=0.02^3=.000008. so now we can conclude that God is 32,875 times more helpful than chance alone!

No. We would conclude that the probability of 3 out of 3 lost explorers surviving is 32,000 times more likely under the God hypothesis than under the random hypothesis, where each explorer has only a 2% chance of returning.

and I can't see any way you can confirm that they survived due to God.

We haven't confirmed anything. We've only showed that the data are more consistent with one hypothesis than another.

 

[Fudbucker]

Only One Finite Life At Most, the straw-man hypothesis Jabba has set up in a false dilemma between it and what he aims to prove in this thread: that there is life after death. He thinks that by showing "OOFlam" to be statistically unlikely, he will have proven immortality by default.

The NDE anecdotes that have been collected so far only establish that a non-trivial percentage of people report a set of very similar spiritual (at least to them) experiences near the moment of death. Science has tried to explain these with various hypotheses (oxygen deprivation, ketamine-type chemicals that produce OOB feelings, etc.).

I think it's interesting that so many people report having these experiences. I think it's surprising that, if materialism is true, many people report highly spiritual experiences near death. It's not what I would have predicted. Surprising results have to be explained. But the anecdotes aren't good evidence for survival-after-death. They only establish the fact that people report these experiences.

Awhile back, there was a big NDE study with signs in places people couldn't see. What would have happened in a person had reported an NDE and reported exactly right what the hidden sign said? That would have been a hugely significant piece of anecdotal evidence, and if more than one person had reported what the various signs were, it would have been covered extensively. But nobody was able to tell what was on the signs, so the survival-after-death hypothesis took a bit of a hit when that study didn't produce any good results.

 

[Jabba]

I've only been involved in the epistemology side of things. What is "OOFlam"?

Fudbucker,

- Thanks for stopping by.

- "OOFLam" is the hypothesis that we each have only one finite life (at most).

- I've been claiming that
1. Anecdotes can constitute evidence re scientific hypotheses.
2. Some anecdotal NDEs constitute evidence of immateriality.
3. Immateriality supports the possibility that OOFLam is wrong.
4. Given OOFLam, the likelihood of my current existence is virtually zero. And,
5. Consequently, according to Bayesian statistics, OOFLam has to be wrong...

- You are one of the few participants that appear to agree with me re #1.
- I'm hoping that I can interest you in #2 and the rest.

 

[Jedi]

I'm not assuming that that is true "in the data." Our data is simply that we have two people who each got lost and each found water and returned. I don't actually use the information that they claim to have been helped by God in the calculation.

you are using the binomial, where, as far as I can tell:

p=survived due to God
(1-p) = survived not due to God
n=sample size
y=number of 'successes' (number of people who survived due to God)

and you're saying n=y=2, so

p^2 (1-p)^(2-2)
= (survived due to God)^2 (did not survive due to God)^0
= (survived due to God)^2

in your example, the (1-p) term will always be 1 (i.e., drop out). if you're not assuming all your anecdotes are proof of God, why is the (1-p) term in the model at all?

 

[Loss Leader]

Remember: if someone fears death and the annihilation that comes with it, it suffices with posting here some rehash of past mistakes every now an then.

In college, I wrote a paper speculating that Vladimir and Estragon were waiting for Godot as a method of avoiding death. I got a high grade on the paper, but the professor wrote, "I think you've let your argument run away from you." I wish I still had that paper. I'd turn it into a meme.

Thanks for "introducing" me to Laura Bell Bundy.

Then my day has been a success.

 

[jt512]

you are using the binomial, where, as far as I can tell:

p=survived due to God
(1-p) = survived not due to God
n=sample size
y=number of 'successes' (number of people who survived due to God)

and you're saying n=y=2, so

p^2 (1-p)^(2-2)
= (survived due to God)^2 (did not survive due to God)^0
= (survived due to God)^2

in your example, the (1-p) term will always be 1 (i.e., drop out). if you're not assuming all your anecdotes are proof of God, why is the (1-p) term in the model at all?

The random variable Y_i = 1 if explorer i returned, and 0 otherwise. It doesn't matter why the explorer did or did not return. Why is not part of the data, only if is part of the data. For the last time, I am only using the number of explorers and whether each returned or not; and I am seeing which hypothesis better predicts those data.

 

[JayUtah]

Anecdotes can constitute evidence re scientific hypotheses.

Equivocation. "Can constitute" obscures much important detail that affects your argument. "Evidence re scientific hypotheses" equivocates your original claim: you said originally that anecdotes support a scientific hypothesis. Do not keep shifting the goalposts to make it seem as of others have agreed with you.

Some anecdotal NDEs constitute evidence of immateriality.

You haven't provided any that do. You've simply quote dumbly from a book full of essentially undocumented ghost stories whose validity and accurate reporting you haven't verified.

Further, you keep trying to sneak in a lesser claim. You are here to prove immortality.

Immateriality supports the possibility that OOFLam is wrong.

No. There can be immateriality without immortality. You said you were going to prove immortality. You don't get to claim victory by proving a broader class of circumstance that just happens to include the one you were originally after.

Given OOFLam, the likelihood of my current existence is virtually zero.

Not a given -- just some crap you made up and are trying to foist on everyone.

Consequently, according to Bayesian statistics, OOFLam has to be wrong.

Asked and answered. Your Bayesian formulation "confirms" every hypothesis it's given regardless of what the hypothesis may actually be. How does that constitute a proof? You were told what was wrong with that formulation, explicitly, by one of the people you've explicitly called out to help you. Explain why you don't respect that explanation.

You are one of the few participants that appear to agree with me re #1.

Equivocation. Here you posited that NDE anecdotes were evidence of life after death. Elsewhere you asked merely whether anecdotes were evidence. Don't try to mix the two.

I'm hoping that I can interest you in #2 and the rest.

What part of "But the anecdotes aren't good evidence for survival-after-death. They only establish the fact that people report these experiences." is unclear?

 

[Jabba]

The NDE anecdotes that have been collected so far only establish that a non-trivial percentage of people report a set of very similar spiritual (at least to them) experiences near the moment of death. Science has tried to explain these with various hypotheses (oxygen deprivation, ketamine-type chemicals that produce OOB feelings, etc.).

I think it's interesting that so many people report having these experiences. I think it's surprising that, if materialism is true, many people report highly spiritual experiences near death. It's not what I would have predicted. Surprising results have to be explained. But the anecdotes aren't good evidence for survival-after-death. They only establish the fact that people report these experiences.
Awhile back, there was a big NDE study with signs in places people couldn't see. What would have happened in a person had reported an NDE and reported exactly right what the hidden sign said? That would have been a hugely significant piece of anecdotal evidence, and if more than one person had reported what the various signs were, it would have been covered extensively. But nobody was able to tell what was on the signs, so the survival-after-death hypothesis took a bit of a hit when that study didn't produce any good results.

Fudbucker,
- I hadn't read your last post before I posted mine.
- Tomorrow, I'll try to address your hilighted position.

 

[Resume]

I suggest that only an after-death experience would constitute proof for immortality.

Anyone ever come back with one of those?

 

[JayUtah]

Tomorrow, I'll try to address your hilighted position.

Or, if you were honest, you could retract the insinuation that he agrees with you. That makes two people who have repudiated your desperate attempt to drag them into the rhetorical swamp that is your argument.

 

[Jedi]

The random variable Y_i = 1 if explorer i returned, and 0 otherwise. It doesn't matter why the explorer did or did not return. Why is not part of the data, only if is part of the data. For the last time, I am only using the number of explorers and whether each returned or not; and I am seeing which hypothesis better predicts those data.

yes, y_i is an indicator of whether they survived. i'm talking about p.

under H1, the hypothesis that God guided them, you are integrating the function on the interval [0.2, 1] wrt p. this is the interval that you yourself identified as likely probabilities that God guided them. so p is the prob of God guiding them. the model is then

(1/0.98) p^y (1-p)^(n-y)

integrated over the values of p in which you identified as likely values for the prob that God was guiding them. the values you are integrating over is how you condition on H1, the God hypothesis. for the 'random chance' hypothesis, you set p=0.20, the value you chose to indicate random chance rather than God.

you then assigned both survivors to the 'success' category, p, and no one to the 'failure' (not survived given God guided them) category. you are assuming all survivors are due to God guiding them.

 

[John Jones]

Fudbucker,

- Thanks for stopping by.

- "OOFLam" is the hypothesis that we each have only one finite life (at most).

- I've been claiming that
1. Anecdotes can constitute evidence re scientific hypotheses.
2. Some anecdotal NDEs constitute evidence of immateriality.
3. Immateriality supports the possibility that OOFLam is wrong.
4. Given OOFLam, the likelihood of my current existence is virtually zero. And,
5. Consequently, according to Bayesian statistics, OOFLam has to be wrong...

- You are one of the few participants that appear to agree with me re #1.
- I'm hoping that I can interest you in #2 and the rest.

Stop grovelling for support. It's pathetic.

 

[jt512]

yes, y_i is an indicator of whether they survived. i'm talking about p.

under H1, the hypothesis that God guided them, you are integrating the function on the interval [0.2, 1] wrt p. this is the interval that you yourself identified as likely probabilities that God guided them. so p is the prob of God guiding them. the model is then

(1/0.98) p^y (1-p)^(n-y)

integrated over the values of p in which you identified as likely values for the prob that God was guiding them. the values you are integrating over is how you condition on H1, the God hypothesis. for the 'random chance' hypothesis, you set p=0.20, the value you chose to indicate random chance rather than God.

you then assigned both survivors to the 'success' category, p, and no one to the 'failure' (not survived given God guided them) category. you are assuming all survivors are due to God guiding them.

No matter how many times you repeat that, it will still be wrong. I don't know how to explain things any better than I already have—more than once. At this point, I would suggest that you study up on Bayesian hypothesis testing using Bayes factors, and return to the conversation once your understanding is better.

 

[Jedi]

No matter how many times you repeat that, it will still be wrong. I don't know how to explain things any better than I already have—more than once. At this point, I would suggest that you study up on Bayesian hypothesis testing using Bayes factors, and return to the conversation once your understanding is better.

In your likelihood ratio test, define p. Note that it takes on the values [.2, 1] for p(D|H1) and .02 for p(D|H2).

 

[Jedi]

No matter how many times you repeat that, it will still be wrong. I don't know how to explain things any better than I already have—more than once. At this point, I would suggest that you study up on Bayesian hypothesis testing using Bayes factors, and return to the conversation once your understanding is better.

maybe i can make my point if i simplify the problem.

you wrote that the p(H1) lies in the interval [.02,1]. let's use a point estimate rather than an interval so we can avoid integrals. how about p(H1)=0.8 (why not?).

you stated the probability of surviving given random chance is 0.02. you then correctly calculated the probability that 2 explorers survived due to random chance as

0.02^2 = 0.0004

analogously, you can calculate the probability that 2 explorers survived due to God's guiding them as

0.8^2 = 0.64

do you agree?

 

[jt512]

maybe i can make my point if i simplify the problem.

you wrote that the p(H1) lies in the interval [.02,1].

"p(H1)" looks like it means the "probability of H1" What I said was that under H1, The Y_i's | p are iid bernoulli(p) random variables, and that p has a uniform[.02, 1] distribution.

let's use a point estimate rather than an interval so we can avoid integrals. how about p(H1)=0.8 (why not?).

If you want to use p=.8 under H1, that's fine with me.

you stated the probability of surviving given random chance is 0.02. you then correctly calculated the probability that 2 explorers survived due to random chance as

0.02^2 = 0.0004

No. See my comments below regarding H1. They apply here (ie, to H2) as well.

analogously, you can calculate the probability that 2 explorers survived due to God's guiding them as

0.8^2 = 0.64

do you agree?

I said our data D consist of two explorers, both of whom got lost, found water, and lived. Your version of model H1 (p=.8) would predict that the probability of the observed data would be .64. Again, this is the probability of the observed data, given that the model is true, P(D|H1). But that is different than "the probability that 2 explorers survived due to God's guiding them." The probability that the 2 explorers survived due to God guiding them is the posterior probability of the hypothesis, that is, P(H1|D). The posterior probability depends not only the likelihood, P(D|H1), but the prior probability of the hypothesis, P(H1), which we have not even touched on.

To see that "the probability that 2 explorers survived due to God's guiding them" depends on the prior probability of the hypothesis, consider what "the probability that 2 explorers survived due to God's guiding them" is if the prior probability P(H1) = 0. Then regardless of how probable the data were under H1, "the probability that 2 explorers survived due to God's guiding them" would be 0, because the probability that God guided them would be 0.

 

[aleCcowaN]

In college, I wrote a paper speculating that Vladimir and Estragon were waiting for Godot as a method of avoiding death. I got a high grade on the paper, but the professor wrote, "I think you've let your argument run away from you." I wish I still had that paper. I'd turn it into a meme.

Point well taken.

But when someone says with momentousness something like "some anecdotal NDEs constitute evidence of immateriality" and he doesn't catch the confessional meaning of his words...

 

[Squeegee Beckenheim]

I think it's interesting that so many people report having these experiences. I think it's surprising that, if materialism is true, many people report highly spiritual experiences near death. It's not what I would have predicted.

I think it becomes less surprising and more in line with what is predictable when you take into account the fact that what people see when they have NDEs is shaped by their culture - in other words, only people raised in Christian environments have visions of angles.

Awhile back, there was a big NDE study with signs in places people couldn't see. What would have happened in a person had reported an NDE and reported exactly right what the hidden sign said? That would have been a hugely significant piece of anecdotal evidence, and if more than one person had reported what the various signs were, it would have been covered extensively. But nobody was able to tell what was on the signs, so the survival-after-death hypothesis took a bit of a hit when that study didn't produce any good results.

Yes, Jabba cited that study himself, so he's well aware of it.