Proof of Immortality, the IV league

[MRC_Hans]

And now you have gone completely off the rails. First you say that the probability of getting a royal flush randomly is .003%, and therefore for the BF for cheated vs random is 33,000; and then you completely contradict yourself, saying that the Bayes factor could be 1, for comparing those very same hypotheses using the very same data.

This may or may not be correct for the Royal Flush scenario, but never the less, this is one big weakness in most of these cases: There are very rarely only two plausible hypotheses. Even if you make the choice binary by making it [my favorite hypothesis] vs [all other hypotheses] you would need to figure out some probability for each and every alternative, one of which will very often be "something we didn't think of".

Hans

 

[Hokulele]

Well, that is jt512's point. Bayes has a very specific, and limited usage, and most people are trying to make it do something it was never intended to do (determine which out of all possible hypotheses is the correct one).

So although Jabba's initial usage of Bayes was superficially correct, the numbers he is trying to plug into it are wildly, laughably, insupportable.

 

[MarkCorrigan]

Jabba, I'm new to this discussion so you haven't seen me say anything before and I'm not going to state who it is I'm agreeing with so far but I wonder if you could respond to a question really quickly?

Could you, in your own words, describe the Texas Sharpshooter fallacy? It's been mentioned a few times and I wonder if you could give me your understanding of it. Thanks in advance.

 

[aleCcowaN]

...

p(cheat|royal) = 1/.003% * p(cheat) = 33,333 * p(cheat)

...

So far, so good.





Still good.




Still good.





Now you start to go off the rails.





...

Can you both cut that screeching concert for symphonic chalk and blackboard off?

I will post back next Thursday -the beginning of the last long weekend before Xmas here-.

 

[MRC_Hans]

Well, that is jt512's point. Bayes has a very specific, and limited usage, and most people are trying to make it do something it was never intended to do (determine which out of all possible hypotheses is the correct one).

So although Jabba's initial usage of Bayes was superficially correct, the numbers he is trying to plug into it are wildly, laughably, insupportable.

Agreed.

Hans

 

[Nonpareil]

Now you start to go off the rails.

You say this, but you haven't actually explained how.

And now you have gone completely off the rails. First you say that the probability of getting a royal flush randomly is .003%, and therefore for the BF for cheated vs random is 33,000; and then you completely contradict yourself, saying that the Bayes factor could be 1, for comparing those very same hypotheses using the very same data.

Yes, I'm aware. That's the point. Manipulating the hypothesis can give you any Bayes factor you want, because the probability of the result given the hypothesis is not dependent on how probable the actual hypothesis is. p(shuffled this way) is very small, but p(royal flush|shuffled this way) is one.

My point is that Bayesian formulas are easily manipulated. In order to have any actual weight as evidence, the priors must be clearly defined, and the calculations must be made with great care.

You can get any Bayes factor you want. What counts is being able to use it. You have to be able to compare the actual probabilities of the hypotheses themselves, not just how likely each one is to result in the given outcome. You must be able to show how likely it is that God actually reached down and guided these lost souls to water, not just say "if he did, they would have a much greater chance of surviving".

Otherwise, all you end up with is a big, fat zero.

This is Jabba's fundamental issue with Bayes throughout this thread. He is saying that, since the probability of his existence is one given that he has an immortal soul, and the probability of his existence would be much less than that if he didn't, the former hypothesis must be true. The Bayes factor is very high, after all.

But he can't establish that p(immortal soul) itself is anything other than zero, so the Bayes factor never enters the equation.

He has no evidence.

He has no argument.

All he has is "I'd be less likely to exist if I weren't special, so I must be special."

A big, fat zero.

 

[Jabba]

Jabba, I'm new to this discussion so you haven't seen me say anything before and I'm not going to state who it is I'm agreeing with so far but I wonder if you could respond to a question really quickly?

Could you, in your own words, describe the Texas Sharpshooter fallacy? It's been mentioned a few times and I wonder if you could give me your understanding of it. Thanks in advance.

Mark,
- Yeah. Saying what your target was after you shoot.
- I think that this is the one real issue in my claim -- i.e. is my current existence an appropriate E for P(E|H)?
- I tried to get members here to focus on that issue, but they preferred to focus on the prior probabilities of H and ~H, leading us to the issues of anecdotal evidence and what exactly a "self" is anyway.

- You're more than welcome.

 

[Nonpareil]

I think that this is the one real issue in my claim -- i.e. is my current existence an appropriate E for P(E|H)?

No, it isn't. This has long since been established. It has been explained to you many times.

Claiming your specific existence as evidence is the Texas sharpshooter fallacy.

I tried to get members here to focus on that issue, but they preferred to focus on the prior probabilities of H and ~H, leading us to the issues of anecdotal evidence and what exactly a "self" is anyway.

That was you, Jabba.

 

[Mojo]

Mark,
- Yeah. Saying what your target was after you shoot.
- I think that this is the one real issue in my claim -- i.e. is my current existence an appropriate E for P(E|H)?
- I tried to get members here to focus on that issue, but they preferred to focus on the prior probabilities of H and ~H, leading us to the issues of anecdotal evidence and what exactly a "self" is anyway.

Jabba,
- your memory is playing tricks on you again. Many members have focused on the Texas sharpshooter issue; you have resolutely ignored what they have said about it.

Now, back to the issue of whether you are the One True Jabba, or merely one of the infinite number of potential selves that could have occupied your body. Do you have any evidence to offer on this matter? It is crucial to your argument, after all.

 

[JayUtah]

I tried to get members here to focus on that issue, but they preferred to focus on the prior probabilities of H and ~H, leading us to the issues of anecdotal evidence and what exactly a "self" is anyway.

No, that is most certainly not what happened. Your formulation of P(E|H) requires a denominator that is a very large number. You have tried several rationales to specify that large number, most recently the concept of "potential selves" which exist abstractly, so you claim, in an infinite number. As most of your critics cannot see a distinction between this concept and the soul in your P(E|~H), they challenged you to provide evidence. You attempted to oblige by once more citing near-death experiences as proof that "something supernatural" was happening. It was you who, in the midst of that presentation, switched to discussing P(H) instead of P(E|H) over the objection of several members, including myself.

It is highly dishonest and insulting of you to accuse your critics of being the ones to constantly change the subject, especially following Loss Leader's quite appropriate dressing-down for your violation of your own terms of debate and your own most recent shift from P(E) back to P(E|H). Do not blame your critics for your own sloth and incompetence.

 

[MarkCorrigan]

Mark,
- Yeah. Saying what your target was after you shoot.


- You're more than welcome.

A little simplified but correct.

So do you see why you're committing it?

 

[JayUtah]

Yeah. Saying what your target was after you shoot.

And do you see how declaring that the specific thing a system must produce is the thing it has already produced is exactly the Texas sharpshooter's fallacy?

I think that this is the one real issue in my claim -- i.e. is my current existence an appropriate E for P(E|H)?

It is not. But that is not the only real issue with your proof. You have many times been given lists of everything that is wrong with your proof and you simply bounce from one to the other. By my reckoning you have been fairly cornered on every single flaw in your theory at least once. You simply don't care; you change subjects and then blame that misdirection on your critics.

 

[Jabba]

A little simplified but correct.

So do you see why you're committing it?

Mark,
- I think that I know why it appears that I'm committing it.
- Hopefully, this will communicate. It's like claiming that the lottery was rigged because somebody (anybody) won it. If that doesn't communicate, I'll try harder.

 

[The Sparrow]

Oh no. Another fringe reset, back to the lottery one.

 

[JayUtah]

If that doesn't communicate, I'll try harder.

Your lottery analogies never work because the emergence of consciousness is not like a lottery. You need to stop thinking of becoming conscious as "winning," because it implies way too many things that just aren't in any way part of the scientific model.

 

[Nonpareil]

Mark,
- I think that I know why it appears that I'm committing it.
- Hopefully, this will communicate. It's like claiming that the lottery was rigged because somebody (anybody) won it. If that doesn't communicate, I'll try harder.

For once, that's exactly right.

That is exactly what your argument does. First, you claim that a lottery exists (JayUtah and other posters have been dealing with the fact that there really isn't one, but that's beside the point for now). Then you claim that you won it, and that therefore it must be rigged in your favor.

It's the Texas sharpshooter fallacy. Point blank, period. There is no getting around it. It is a fallacy, and your "argument" collapses as a result.

 

[jt512]

You say this, but you haven't actually explained how.

I explained it in the next paragraph, which you just quoted.

Yes, I'm aware. That's the point. Manipulating the hypothesis can give you any Bayes factor you want, because the probability of the result given the hypothesis is not dependent on how probable the actual hypothesis is.

WTF!? Your reasoning is so bizarre, I don't even know how to correct it. It makes so little sense, it doesn't even rise to the level of "wrong." If you change the hypotheses you're comparing, of course you'll get a different Bayes factor. By definition, it is a different Bayes factor. But that's not "manipulation." It's comparing different models.

My point is that Bayesian formulas are easily manipulated.

If they are, you sure haven't shown it.

In order to have any actual weight as evidence, the priors must be clearly defined....

No, no, no, no, no! The Bayes factor does not depend on the priors. I can't believe you still don't understand this. The whole point of computing the Bayes factor in the first place is because the Bayes factor, or "weight of the evidence," is independent of the priors.

This is Jabba's fundamental issue with Bayes throughout this thread. He is saying that, since the probability of his existence is one given that he has an immortal soul, and the probability of his existence would be much less than that if he didn't, the former hypothesis must be true. The Bayes factor is very high, after all.

As I have explained before, the fundamental issue with Jabba's argument is that under both hypotheses, he can only observe the data (E), that he exists, if, in fact, he exists. Thus he is conditioning the data on the data itself. That is, his Bayes factor is

P(E|E, H1) / P(E|E, H2) = 1 / 1 = 1 ,

which, ironically, is the same mistake you made above.

He has no evidence.

Correct conclusion. Wrong reason for it.

 

[Jabba]

...
He is saying that, since the probability of his existence is one given that he has an immortal soul, and the probability of his existence would be much less than that if he didn't, the former hypothesis must be true. The Bayes factor is very high, after all...

jt,
- I'm saying that if H (OOFLam) is wrong, the likelihood of my current existence is MUCH greater than it is if H is right, and that the difference between the two more than makes up for the difference between the prior probabilities of H and ~H.

 

[Mojo]

- I'm saying that if H (OOFLam) is wrong, the likelihood of my current existence is MUCH greater than it is if H is right, and that the difference between the two more than makes up for the difference between the prior probabilities of H and ~H.

But in order to arrive at this conclusion you have had to take the ludicrous position that if you have an immortal soul, immortal souls don't exist, and if you don't they do.

 

[jt512]

jt,
- I'm saying that if H (OOFLam) is wrong, the likelihood of my current existence is MUCH greater than it is if H is right, and that the difference between the two more than makes up for the difference between the prior probabilities of H and ~H.

I know that's what you are saying, and I have explained repeatedly, including in the post you just replied to, why you are wrong.