Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss
Re: Trampolining Armadillos Threaten Civil Aviation
23 views
Skip to first unread message
Message has been deleted
nigel
Jun 3, 2012, 6:22:51 AM6/3/12
to
nigel wrote:
> x-no-archive: yes
>
> Seen the movie 'Mars Attacks'? As if assaulted by a deadly dose of 'Only
> You', my half braincell has exploded and is splattered all over the
> inside of my skull.
>
> Take a cube.
>
> Populate it 'randomly' with points generated from the 100,000 6/49
> pseudo-lottery draws generated by unix and given to me for my
> 'randomness' study. (I'm not going to divulge the algorithm I used, but
> it's the same one that reveals the cyclical nature of data from
> random.org.)
>
> Now populate it 'randomly' with an equal number of points generated from
> the 100,000 6/49 pseudo-lottery draws generated by python (Mersenne
> Twister), also given to me for my 'randomness' study.
>
> Now take genuine tumble-drier-and-ping-pong-ball 6/49 lottery historical
> data and generate points for the cube in the same way.
>
> For each genuine lottery point, the nearest unix point is almost always
> closer than the nearest python point.
>
> Despite my lightly-made claims that there are different types of
> 'randomness', I'm way out of my depth here.
>
> Evil Nigel
>
It's been suggested that the python points are somehow constrained to
the nodes of a lattice whereas the unix and
ping-pong-ball-in-tumble-drier points have completely free rein within
the cube.
While that would explain the effect, that raises more questions than it
answers. The unix random number generation was almost certainly by a
PRNG algorithm, and by established randomness criteria inferior to
python's Mersenne Twister.
"Only yooo-oooo-ooooo-ooo-ooo-oooo"
Splat!
Evil Nigel
> x-no-archive: yes
>
> Seen the movie 'Mars Attacks'? As if assaulted by a deadly dose of 'Only
> You', my half braincell has exploded and is splattered all over the
> inside of my skull.
>
> Take a cube.
>
> Populate it 'randomly' with points generated from the 100,000 6/49
> pseudo-lottery draws generated by unix and given to me for my
> 'randomness' study. (I'm not going to divulge the algorithm I used, but
> it's the same one that reveals the cyclical nature of data from
> random.org.)
>
> Now populate it 'randomly' with an equal number of points generated from
> the 100,000 6/49 pseudo-lottery draws generated by python (Mersenne
> Twister), also given to me for my 'randomness' study.
>
> Now take genuine tumble-drier-and-ping-pong-ball 6/49 lottery historical
> data and generate points for the cube in the same way.
>
> For each genuine lottery point, the nearest unix point is almost always
> closer than the nearest python point.
>
> Despite my lightly-made claims that there are different types of
> 'randomness', I'm way out of my depth here.
>
> Evil Nigel
>
It's been suggested that the python points are somehow constrained to
the nodes of a lattice whereas the unix and
ping-pong-ball-in-tumble-drier points have completely free rein within
the cube.
While that would explain the effect, that raises more questions than it
answers. The unix random number generation was almost certainly by a
PRNG algorithm, and by established randomness criteria inferior to
python's Mersenne Twister.
"Only yooo-oooo-ooooo-ooo-ooo-oooo"
Splat!
Evil Nigel
nigel
Jun 6, 2012, 6:29:29 PM6/6/12
to
draws. Both have a unix point closer than the nearest python point.
Funny old world, innit!
Evil Nigel
nigel
Jun 12, 2012, 4:50:34 AM6/12/12
to
cube. The nearest unix point and the nearest python point were virtually
equidistant, the unix point shaving it by the width of a cigarette paper.
Evil Nigel
nigel
Jul 28, 2012, 4:21:41 AM7/28/12
to
I stuck another branch into my randomness investigation.
If you use random.org to generate sets (combos) of pseudo-lottery draws,
the results are significantly different to the results produced by
ping-pong balls and tumble-dryers.
However random.org offers a facility to "Print the sets in random
order". If the sets (combos) are additionally randomised in this way,
the results are a lot closer to the sort of results ping-pong balls in a
tumble-dryer would produce.
I hypothesise that the extra randomisation is destroying the inherent
periodicity of data from random.org, but proving it is waaaay beyond my
capabilities.
Evil Nigel
If you use random.org to generate sets (combos) of pseudo-lottery draws,
the results are significantly different to the results produced by
ping-pong balls and tumble-dryers.
However random.org offers a facility to "Print the sets in random
order". If the sets (combos) are additionally randomised in this way,
the results are a lot closer to the sort of results ping-pong balls in a
tumble-dryer would produce.
I hypothesise that the extra randomisation is destroying the inherent
periodicity of data from random.org, but proving it is waaaay beyond my
capabilities.
Evil Nigel
nigel
Aug 15, 2012, 1:59:48 PM8/15/12
to
How about:
The probability that a) sets of data from random.org and b) sets of data
from random.org printed in random order come from the same population
is, according to the latest figures from my trial, p<0.003
Evil Nigel
0 new messages