A Modest Proposal
tapewo...@my-dejanews.com
I was thinking about the "little cheats" post made a while back... would it
be possible to do an analysis of the Big Brother archives that calculated the
average difference in total pips thrown by each player, or something to that
effect? If there was no substantial (=statistically significant) difference
then that would debunk the existence of some sort of "little cheats" system,
wouldn't it? Anyone with more statistics knowledge want to correct &
implement this?
CR
-----== Posted via Deja News, The Leader in Internet Discussion ==-----
http://www.dejanews.com/ Now offering spam-free web-based newsreading
Vladimir Kuznetsov
It was pointed on this group already, but I will repeat it. The overall
evenness of the computer dice does not guarantee yet that it is
honest.
Let's consider all positions when you desperately need say, 5-5
to survive. How often did you get 5-5. It should be statistically
around 1/36. If it is not then something is wrong.
In all other respects dice can be very even and random.
vlad
In article <6n0stm$q5d$1...@nnrp1.dejanews.com>,
Patti Beadles
Not only is it possible to go through the Big Brother archives, but
it's certainly easy enough to do so. All of the information is
online, and anyone who wants to do this (including you) is welcome to
do so.
-Patti
--
Patti Beadles |
pat...@netcom.com/pat...@gammon.com |
http://www.gammon.com/ | Try to relax
or just yell, "Hey, Patti!" | and enjoy the crisis
Vince Mounts
This seemed as good a time as any to offer up some software I wrote related
to BigBrother recorded matches.
I have found that for the Win95/98 OS there was really no choice for a
viewer of Big brother files other than to 1) download and doctor up some c++
source or 2) buy a viewer that was out there. So instead I wrote one myself.
For now you can get it at ftp://ftp.mindspring.com/users/vmounts/bgsetup.zip
Just download it and when you run setup follow all of its instructions. The
download size is pretty hefty at 1.5 mb simply because it was written in
VisualBasic and there is need of support files. At some time in the future I
will offer a separate download for those who are sure they have the vb5
runtime files and eventually add a page to my home page about the program.
But for now just use the URL above in your browser. Anyone who wishes to
link to the file from their site should be okay to assume that the file will
stay put at its current location (and has my permission to link lol). Its
only been tested on 3 systems but should work okay for everyone with
1)win95/98 2)8mb ram 3)486/pentium 4)5mb hard drive (takes up less than that
but needs some extra during the install).
Drop me a line if you link to it from your site so i know for sure not to
move it around anywhere.
And before you flame me for spam it is freeware not shareware so enjoy. (I
did some intial calcs and figured with such a low demand for such a viewer
and the enormous economy of scale and whatnot I should probably charge about
$549.87 but figured what the hell I'll make it free :)
-----
Vince Mounts (a.k.a einniv)
E-Mail: vmo...@mindspring.com
Home Page URL: http://vmounts.home.mindspring.com
Patti Beadles wrote in message ...
Peter Fankhauser
Vladimir Kuznetsov wrote:
> It was pointed on this group already, but I will repeat it. The overall
> evenness of the computer dice does not guarantee yet that it is
> honest.
>
> Let's consider all positions when you desperately need say, 5-5
> to survive. How often did you get 5-5. It should be statistically
> around 1/36. If it is not then something is wrong.
>
> In all other respects dice can be very even and random.
>
> vlad
I've applied such a test about 2 years ago.
From my posting (1996/07/11) digged up at dejanews....
> I've applied Stephen Turner's test to all positions in the BigBrother
> Matches where a player needs to enter against a 4pt board
> (adding 36/20 to the counter on enter) or against a 5pt board
> (adding 36/11).
> There where 39060 such positions and the counter added up to 38365.4.
> This gives a 11/36*0.98 chance to enter against 5pt board, and
> 20/36*0.98 against a 4pt board ;-)
etc. etc.
0.98 appears close enough to 1. But then, maybe this sample is not representative,
since it has been calculated from matches between players who have not complained
about rigged dice on this forum yet ;)
As far as I remember, I performed a similar test on shots vs. hits - with similar
results,
but I don't remember whether I bothered posting this too.
funk