Re: GOD=G_uv BAYESIAN P=99.9 %

From: The Ghost In The Machine (ewill_at_aurigae.athghost7038suus.net)
Date: 07/29/04 

Date: Thu, 29 Jul 2004 16:01:01 GMT

In sci.physics.relativity, George Hammond
<research137@hotmail.com>
 wrote
on Wed, 28 Jul 2004 21:52:23 GMT
<reVNc.205545$XM6.185249@attbi_s53>:
>
> GOD=G_uv BAYESIAN P=99.9 %
>
> NOTE: Stephen Unwin PhD (physics/quantum gravity) is author of
> _The Probability of God_ (Crown Forum, 2003) in which he
> uses Bayesian probability theory to calculate that the
> probability is P=67% that God exists.
> A 1 page review of this book by Michael Shermer appears
> in the current (July) issue of Scientific American on
> page 46.
>
> In that SCIAM review the formula used is given as:
>
>
> P(bef) * D
> P(aft) = ----------------------------
> P(bef) * D + 100 - P(bef)
>
> where P(bef) = the prior probability (percent)
> P(aft) = the posterior probability (decimal)
> D = the liklihood ratio
>
> for a succession of evidences (see below) each P(aft) is used as
> a P(bef) in the next round of calculations.
>
> I have programmed this formula in GWBASIC to save you using a
> hand calculator:
>
> 100 CLS
> 110 PB=50
> 120 INPUT "HOW MANY INPUT DATA";N
> 130 PRINT
> 140 DIM D(N)
> 150 FOR I=1 TO N
> 160 INPUT "DATA ";D(I)
> 170 NEXT I
> 180 PRINT
> 190 PRINT "P( 0 )= 50.00000 %"
> 200 FOR K=1 TO N
> 210 PA=PB*D(K)/((PB*D(K))+100-PB)
> 220 PRINT "P(";K;")=";PA*100
> 230 PB=PA*100
> 240 NEXT K
>
> For instance, to confirm Unwin's Result for the existence
> of God just input 6 when it asks "how many data", and then
> punch in 10 .5 .1 2 1 2 in answer to the 6 input
> queries. You will find the answer comes up 66.7%
> (see SciAm article for more info)
>
>
> At any rate, I give below six "liklihood ratios" for the six pieces
> of evidence I feel summarize the argument for GOD=G_uv, they
> turn out to be D= 2, 10, 2, 2, 10, 2
> If you punch these six numbers into the program you get:
>
> P=99.9 % that GOD=G_uv is TRUE
>
>
>=========================
>
> In other words, it is an
>
> ABSOLUTE DEAD CERTAINTY
>
> that GOD=G_uv is correct
>
>=========================

[rest snipped]

A C version of the above program can be straightforwardly coded:

---- 8< --gh.c-- >8 -----
#include <stdio.h>
#include <stdlib.h>

int main(int argc, char **argv)
{
        double pa;
        double pb;
        int n;
        double * d;
        int i;
        
        pb = 50;

        if(argc < 2) {
                fprintf(stderr,"Usage: %s i1 i2 ... in\n", argv[0]);
                fprintf(stderr,"where each datum is a preseed \n");
                return 1;
        }

        n = argc - 1;

        d = (double *) calloc(sizeof(double), n);

        for(i=0;i<n;i++)
        {
                d[i] = strtod(argv[i+1], 0);
        }

        printf("p(0) = %6.4f%%\n", pb);

        for(i=0;i<n;i++)
        {
                pa = pb*d[i]/((pb*d[i])+100-pb);
                printf("p(%d) = %6.4f%%\n", i+1, pa*100.0);
                pb = pa*100;
        }

        return 0;
}
---- 8< -------- >8 -----

I'll leave Java to another post or poster, although the conversion
shouldn't be difficult.

$ cc gh.c -o gh
$ ./gh 10 .5 .1 2 1 2
p(0) = 50.0000%
p(1) = 90.9091%
p(2) = 83.3333%
p(3) = 33.3333%
p(4) = 50.0000%
p(5) = 50.0000%
p(6) = 66.6667%
$ ./gh 2 10 2 2 10 2
p(0) = 50.0000%
p(1) = 66.6667%
p(2) = 95.2381%
p(3) = 97.5610%
p(4) = 98.7654%
p(5) = 99.8752%
p(6) = 99.9375%

Now -- this program shows the results as described, but interpreting
the results I leave to the reader; I don't see this as proof of much
of anything beyond George being a BASIC programmer. :-) 

-- 
#191, ewill3@earthlink.net
It's still legal to go .sigless.

Relevant Pages

  • Re: GOD=G_uv BAYESIAN P=99.9 %
    ... In sci.physics.relativity, George Hammond ... > In that SCIAM review the formula used is given as: ... > for a succession of evidences each Pis used as ...
    (sci.physics.relativity)
  • Re: GOD=G_uv BAYESIAN P=99.9 %
    ... In sci.physics.relativity, George Hammond ... > In that SCIAM review the formula used is given as: ... > for a succession of evidences each Pis used as ...
    (sci.physics)
Loading