Re: [Bug-gnubg] Re: How fast can you cheat??

[Roy A. Crabtree]

The interesting thing to note is that GNUBG also cheats.

The engine is scrupulously honest and there is NO hidden trapdoor in the code.

What I am denoting is that the NNP "learns" the attributes of the specific roll generator you use.  And does it probably better than a human can.

It will also learn the average hidden "tell" in the pattern of a human player; and the full DB in all likelihood  (I so assert) has several folded state transition spaces covering the panoply of the subtypes of human play available. (given enough games rolled in).

It would be interesting to get a few world class players to own up to what "tells" they use when they only have the actual order of play present (no visual or timing tells):

You can tell alot about each player by their wn pattern;

and a true expert will be able to infer more about one player by how they react to another than you might believe.

So does the DB state space in a trained GNUBG database.

Basically, _everything_ goes into it.

And whether you fold the state space down from the infinite manifold caleed reality
into a mere 100,000 trained transition nodes

or down into only a few ....

_everything_ goes into it.

And _everything_ in it that -can_ come out __will__

when it is aligned with the "goal" or "purpose" of the overall engine involved:

   to win the game of backgammon.

So, if you find a state within the space of any NNP game playing mechanism

that seems to pick a combination that actually is NOT the optimal according to

compeltely fair dice rolls and probabiltiy theory

   and you _will_ if you look closely (I assert)

then:

  take it on faith that

   the program involved has found aparticular _weakness_ inherent in the play of either

     the human game platers loading the NNP state space or

       a hidden pattern in the PRNG series being used.

that is _not_ present in strictly fair uniform probabiltiy series.

Even random.org/com notes that their random series are "improved" (made "more" effectively or "intuitively" random) by removing or reducing large series of improbable rolls (such as 0's and 1's in a row).

And thatis, in reality, NOT random.

And I will guarantee that the engine in GNUBG will "earn" that and _use_ it to win at a higher rate than predicted by probabiltiy theory.

So:  GNUBG " cheats" _honestly_ by

    extracting information about the actual PRNG or input series given

   than is available to a humanplayer 

    who has NOT had a chance to see the entire DB load

    nor to run out a series of Mersenne twisters long enough to _see_

      hidden "patterns" inthe Mersenne series that

          the GNUBG negine _can_ sense and eventually _tabulate_ in the DB.

Not explciitly, not clearly in the open coded as such, but _inherent_ in the eigenstate coded in the DB.

Can a human cheat the same way?  

To a possibly more limited extent, by spotting either explicitly as a coded pattern or demi-pattern, ir explicitly by a "tell" or timing or marker or sequence start point, or  _implicitly_ by coalescently learning to make the move i9ntuitively without ever actually being able to say why, and possibly not even say _when_ before the actual moment occurs.

that's the neat thing about an indefinite state space as available within a living semiotic system:

It is really difficult to predict what is actually in it, and whether it can "beat" the :"cheat" that the GNUBG database gives the engine.

On Thu, Aug 20, 2009 at 15:53, Michael Petch <address@hidden> wrote:

On 20/08/09 9:26 AM, in article
4a8d6b09$0$504$address@hidden, "address@hidden"
<address@hidden> wrote:
>
> This is an interesting challenge.  Perhaps someone who knows gnubg better
> can explain how exactly the dice are generated from the Mersenne Twister.
> Given particular seeds, do we get the same sequence of dice rolls?  Or
> is it more complicated than that?

The answer is YES but with with a caveat. If you use the same seed you will
get the same dice results everytime. There is nothing special in the way
this is handled.

The oddity is in how the seed gets set when you use start a new match
through the GUI. If you start a new match, the seed is randomly set before
the match is played. You can see the seed it chose by starting a match, then
going to Settings/Options/Dice and looking at the "seed". If you set the
seed through the GUI (in options), and start a new match and then go back
and look at the seed it will change. I think what the GUI lacks in start
match is the option to use the current seed (instead of the one currently
set). I believe it was done this way so that for a casual player, each match
was different. I'm CC'ing the mailing list with this post/feature request..

Anyway, in current releases if you wish to use the GUI and set your own seed
and get the same dice rolls for an entire match do this:

A) Click "New" icon.
B) Enable "Manual Dice" checkbox
C) set any other player options that you wish.
D) Click "Ok"
E) The dice selection screen will appear. Don't select a roll, just hit
"Close"
E) Go to Options/Settings/Dice tab and
   a) Set the random number generator to Mersenne Twister (or which ever
      seeded PRNG you like)
   b) set the seed to some value (3101)
   c) Click "OK"

[At this point the seed is set]

F) Click anywhere on the board and you should get a message along the lines
of "A new match" has been started.
G) Play the match

Now do this process again and from the start with the same seed. The rolls
generated should be the same.

(There is  a shorter way to set the seed from the "command panel" using the
"set seed" command for power users).


Michael





_______________________________________________
Bug-gnubg mailing list
address@hidden
http://lists.gnu.org/mailman/listinfo/bug-gnubg

-- 
Use Reply-To: & thread your email 
after the first: or it may take a while, as
I get 2000+ emails per day.
--

Roy A. Crabtree
UNC '76 gaa.lifer#  11086

(mail, residence)
Roy A. Crabtree
3322 Wheeler Road SE
Oak Hill Apartments #T-4
Washington, DC 20032-4166
202-562-1909 US no voicemail 
    (try after 2100EST)

(secondary mail)
Roy A. Crabtree

USPS POB 58097
Washington, DC 20034-8097
703-318-2106 
(msgs only, use my name)
(best effort next day M-F pickup)

[When you hear/read/see/feel what a y*ehudi plays/writes/sculpts/holds]
[(n)either violinist {Menuhin} (n)or writer {"The Y*ehudi Principle"} (n)or molder (n)or older]

[you must strive/think/look/sense all of it, or you will miss the meanings of it all]

address@hidden Forwards only to:
address@hidden

address@hidden CC: auto to ^

http://musings-roy-crabtree.blogspot.com [& others]
http://www.authorsden.com/royacrabtree

http://skyscraper.fortunecity.com/activex/720/resume/full.doc
--
(c) RAC/IP, ARE,PRO,PAST
(Copyright) Roy Andrew Crabtree/In Perpetuity

    All Rights/Reserved Explicitly
    Public Reuse Only
    Profits Always Safe Traded