The fine tuning argument

[westprog]

The number of trials is crucial in determining a frequency of occurance. the more you toss teh coin, the better the ferquency. And a run of ten is less certain than a run of a hundred, and much less certain than a run of ten thousand. The larger the data set the more likely the inference that it is a biased coin.

Which is irrelevant to the fine tuning argument, you can not state probabilities on the unknown.

One can't state absolute probabilities - but absolute probabilities are a fiction anyway. One can make assumptions, and calculate from those assumptions the likely outcomes.

 

[Yoink]

I wonder if those who think that there's something to this "fine tuning" argument would answer a question for me: let us assume that we discover that our universe's parameters are in one possible configuration out of exactly one million equally possible configurations (I know this is unlikely, but for the sake of clarity, let's just assume it).

Let us further assume that we discover that this universe is the only universe that ever was or will be created. And, finally, let us also assume that we discover that in only one of these million possible sets of parameters could something like human life have evolved. So, we're the one-in-a-million shot.

Now, would such discoveries, in your view, strongly support the "fine tuning" hypothesis? And (and this is the really important part, for me, of my question): to what extent? In other words: if the odds against a universe in which anything like human life could emerge are 999,999-to-1, does that make the odds of there being a "Fine Tuner" also 999,000-to-1? How, exactly, do we derive the "probability" of a "tuner" from the "improbability" of the event?

 

[westprog]

Since Malerin is content to be silent on the issue -- which we expected, given his history -- would you care to explain how one would go about deciding that?

Firstly one can look at the likely nature (given current knowledge) of the universe if the fundamental constants had different values. This is doable, I believe. Thus we can establish whether this particular universe is unique or special. If most of the other universes appear to be similar, then it tends one towards a belief that there is some reason for it.

Then it's a matter of what explanation one prefers - multiverse, creator, simulation, hidden laws of physics or anything else.

 

[westprog]

I wonder if those who think that there's something to this "fine tuning" argument would answer a question for me: let us assume that we discover that our universe's parameters are in one possible configuration out of exactly one million equally possible configurations (I know this is unlikely, but for the sake of clarity, let's just assume it).

Let us further assume that we discover that this universe is the only universe that ever was or will be created. And, finally, let us also assume that we discover that in only one of these million possible sets of parameters could something like human life have evolved. So, we're the one-in-a-million shot.

Now, would such discoveries, in your view, strongly support the "fine tuning" hypothesis? And (and this is the really important part, for me, of my question): to what extent? In other words: if the odds against a universe in which anything like human life could emerge are 999,999-to-1, does that make the odds of there being a "Fine Tuner" also 999,000-to-1? How, exactly, do we derive the "probability" of a "tuner" from the "improbability" of the event?

But it's not possible to allocate probabilities for the myriad of explanations which can be plugged in. How can one say that a simulation is more or less likely than a multiverse?

 

[yy2bggggs]

I easily got a max run length of 4, several times:

Sure... that's fairly likely.

And this was my last run through:

Sure... it can happen. Keep in mind, though, that for this "gambler's fallacy" property, I'm also counting the runs. A sequence of 55 is also fairly small, so we're not really talking insane odds here (though 1 in 20 is 5%, and in this context, it falls just into the realm of what is typically considered statistically significant)--the main point, rather, is that there are such properties which suggest intent--things that are characteristic of the actual entities we know float around in the real world.

If you pit weighted coins against typical people off the street generating random numbers, then both the most "special" sequences, and the least "special" ones, are coin flips (least special would be fair coins... most special, heavily weighted). Humans pulled off of the street would tend to develop sequences that follow particular patterns.

This applies to the FT argument for God--note that the parameters we're talking about that "make life possible" are, not entirely coincidentally, also the ones that maximize the variety of objects.

 

[Yoink]

But it's not possible to allocate probabilities for the myriad of explanations which can be plugged in. How can one say that a simulation is more or less likely than a multiverse?

I'm simply asking for an explanation of the understanding of "probability" in a given hypothetical case. Whether or not the hypothetical is realistic (of course it isn't) has no effect whatsoever on the mathematics of probability. We have many people in this thread arguing that if our universe is "improbable" we should be able to infer something about the "probability" of a "Fine Tuner" from that "improbability." Well, I've given a hypothetical case of very precise "improbability" and I'd be interested to see exactly what level of "probability" that implies for a "Fine Tuner"--and how that result is derived.

 

[rocketdodger]

Firstly one can look at the likely nature (given current knowledge) of the universe if the fundamental constants had different values.

Um, no, that isn't good enough.

You have to generate some kind of distribution for the values of those constants.

And therein lies the problem for FT proponents -- how in the world do you propose to come up with a distribution for an event you know nothing about the cause of?

You can't just say, for example, that the gravitational constant could be any real number without some kind of evidence that it could be any real number.

We know a coin toss is binary. We know a six sided die can take on values 1 - 6. We know a lottery can take on some finite number of values. We know a 32 bit word in computer memory can take on a finite number of values. We know the spatial coordinates of a particle can be any real number, ignoring possible planck discretization, given a coordinate frame.

We do not know which values any universal constant might have been, and even if we did, we wouldn't know the probability it might have been what it might have been.

This seems pretty obvious to me. So obvious, in fact, that I interpret omission of this caveat in any FT argument as deliberate dishonesty. That is why I have about as little respect for Malerin as I did for Kleinman -- it is clear that there is zero evidence that our universal constants could be "any value at all," yet Malerin insists on continuing to rely upon exactly this assumption in all his arguments.

So the question is, are you going to act the same way, or are you going to let reason prevail and admit that the FT argument is currently a dead end because we just don't have enough data?

 

[rocketdodger]

And, I should add, when we do have enough data it is highly likely that the data will defeat the FT hypothesis anyway.

Because the only way to confirm any ideas we have about the possible distribution of the values of universal constants will be to observe other universes where those values are different -- and once other universes are observed the multiverse hypothesis is implicitly confirmed.

 

[Yoink]

And, I should add, when we do have enough data it is highly likely that the data will defeat the FT hypothesis anyway.

Because the only way to confirm any ideas we have about the possible distribution of the values of universal constants will be to observe other universes where those values are different -- and once other universes are observed the multiverse hypothesis is implicitly confirmed.

Ian Hacking rather famously argued that a multiverse would make no difference to the argument (he called it the "inverse gambler's fallacy"). I think that on this point he's cuckoo for coconuts, but I just want to point out that the argument won't go away even if a multiverse is confirmed.

 

[Dancing David]

One can't state absolute probabilities - but absolute probabilities are a fiction anyway. One can make assumptions, and calculate from those assumptions the likely outcomes.

I did not say absolute, I said that without frequency you can not observe the actual probability of an event.

You can not say if the universe has a bias when you can not say how many sides the die has.

 

[Dancing David]

Firstly one can look at the likely nature (given current knowledge) of the universe if the fundamental constants had different values. This is doable, I believe. Thus we can establish whether this particular universe is unique or special. If most of the other universes appear to be similar, then it tends one towards a belief that there is some reason for it.

Then it's a matter of what explanation one prefers - multiverse, creator, simulation, hidden laws of physics or anything else.

Yes but we can not know what the factors of change in the constants are 1000, 100, 10, 1, .1, .01, .001, etc...

So you can't sya anything about what might or might not be.

 

[Cynic]

David, forgive me, but you must be somewhat, dillusioned and misplaced.

Did i note recite a reference? A rhetorical question of course, but one inexorably due, since what is generally-considered to provide a proof, is by such an action. So you are making yourself look more and more foolish when you confabulate the extremities of your assertions impying that i have not given any proof, or cannot back it up.

Citing a reference in this case would be tantamount to arguing from authority, since you cannot show how any of your arguments work.

 

[Beth]

I wonder if those who think that there's something to this "fine tuning" argument would answer a question for me: let us assume that we discover that our universe's parameters are in one possible configuration out of exactly one million equally possible configurations (I know this is unlikely, but for the sake of clarity, let's just assume it.

Let us further assume that we discover that this universe is the only universe that ever was or will be created. And, finally, let us also assume that we discover that in only one of these million possible sets of parameters could something like human life have evolved. So, we're the one-in-a-million shot.

Now, would such discoveries, in your view, strongly support the "fine tuning" hypothesis? And (and this is the really important part, for me, of my question): to what extent? In other words: if the odds against a universe in which anything like human life could emerge are 999,999-to-1, does that make the odds of there being a "Fine Tuner" also 999,000-to-1? How, exactly, do we derive the "probability" of a "tuner" from the "improbability" of the event?

If the 'tuner' is the only other explanation possible, then the probability of the tuner is one minus the probability of getting those constants by random chance alone - using your numbers, it would be 0.999999. A fairly convincing probability IMO. However, if it isn't the only other hypothesis, then the combined probability of the other hypotheses will sum to .999999.

 

[Yoink]

If the 'tuner' is the only other explanation possible, then the probability of the tuner is one minus the probability of getting those constants by random chance alone - using your numbers, it would be 0.99999. A fairly convincing probability IMO. However, if it isn't the only other hypothesis, then the combined probability of the other hypotheses will sum to .99999.

Thanks for the answer. Is .99999 a typo, though? Why isn't it .999999?

 

[Beth]

Thanks for the answer. Is .99999 a typo, though? Why isn't it .999999?

Yes, I fixed it. Thanks.

 

[fls]

I'm simply asking for an explanation of the understanding of "probability" in a given hypothetical case. Whether or not the hypothetical is realistic (of course it isn't) has no effect whatsoever on the mathematics of probability. We have many people in this thread arguing that if our universe is "improbable" we should be able to infer something about the "probability" of a "Fine Tuner" from that "improbability." Well, I've given a hypothetical case of very precise "improbability" and I'd be interested to see exactly what level of "probability" that implies for a "Fine Tuner"--and how that result is derived.

Ah, yes. The Statistical Hypothesis Inference Testing fallacy as outlined by Jacob Cohen in "The Earth is Round (p<0.05)".

Linda

 

[Yoink]

Yes, I fixed it. Thanks.

O.K., So, in the absence of any other explanations, an outcome the odds of which are 1-in-1,000,000 offer .999999 probability of being deliberately 'guided' by some controlling force.

O.K.

So, say I roll a single die. I get a 4. The odds of my getting a 4 were 1-in-6. So from a single die-roll we can infer odds of 5/6 that there is a mystical controlling intelligence that "selected" that rather unlikely outcome. Correct?

If I roll it again and get a 6, the odds against that sequence are 35/36! The odds of a mystical controlling intelligence selecting that outcome are now staggeringly high! One more roll of the die--let alone two!--and God has been pretty definitively proven!

But hey--I don't even need die. Let me create a "wheel of fortune" type wheel with very long circumference. Let me then place one million divisions around the circumference and then give the wheel a spin. No matter what the outcome, the odds of getting that particular outcome are one-in-a-million. That means that simply by spinning this wheel once, no matter what result I get, and no matter what random fluctuations influence the outcome, I will have proven that the result was "selected" by a "fine tuner"--and I will have done so to a confidence level of .999999--or, in other words, virtual certainty.

I think this little thought experiment shows the problem with the "Fine tuner" hypothesis pretty clearly, don't you?

 

[fls]

I wonder if those who think that there's something to this "fine tuning" argument would answer a question for me: let us assume that we discover that our universe's parameters are in one possible configuration out of exactly one million equally possible configurations (I know this is unlikely, but for the sake of clarity, let's just assume it).

Let us further assume that we discover that this universe is the only universe that ever was or will be created. And, finally, let us also assume that we discover that in only one of these million possible sets of parameters could something like human life have evolved. So, we're the one-in-a-million shot.

Now, would such discoveries, in your view, strongly support the "fine tuning" hypothesis? And (and this is the really important part, for me, of my question): to what extent? In other words: if the odds against a universe in which anything like human life could emerge are 999,999-to-1, does that make the odds of there being a "Fine Tuner" also 999,000-to-1? How, exactly, do we derive the "probability" of a "tuner" from the "improbability" of the event?

You can't.

The two bits of information are independent, so one is not the complement of the other.

Linda

 

[Yoink]

You can't.

The two bits of information are independent, so one is not the complement of the other.

Linda

I agree absolutely, but I wanted to see if the proponents of the FTA had some sort of argument to get us from "improbability of random occurrence" to "probability of 'tuned' outcome." So far it seems not.

 

[fls]

O.K., So, in the absence of any other explanations, an outcome the odds of which are 1-in-1,000,000 offer .999999 probability of being deliberately 'guided' by some controlling force.

O.K.

So, say I roll a single die. I get a 4. The odds of my getting a 4 were 1-in-6. So from a single die-roll we can infer odds of 5/6 that there is a mystical controlling intelligence that "selected" that rather unlikely outcome. Correct?

If I roll it again and get a 6, the odds against that sequence are 35/36! The odds of a mystical controlling intelligence selecting that outcome are now staggeringly high! One more roll of the die--let alone two!--and God has been pretty definitively proven!

But hey--I don't even need die. Let me create a "wheel of fortune" type wheel with very long circumference. Let me then place one million divisions around the circumference and then give the wheel a spin. No matter what the outcome, the odds of getting that particular outcome are one-in-a-million. That means that simply by spinning this wheel once, no matter what result I get, and no matter what random fluctuations influence the outcome, I will have proven that the result was "selected" by a "fine tuner"--and I will have done so to a confidence level of .999999--or, in other words, virtual certainty.

I think this little thought experiment shows the problem with the "Fine tuner" hypothesis pretty clearly, don't you?

It illustrates that the set-up was incorrect. You are essentially creating a set which contains all the possible outcomes and their frequency, in the presence of a fine-tuner. What you really wish to create is a set of all possible causes and their frequency in the presence of a particular sequence of die rolls. Unfortunately, that information is the very information that is in question.

Linda