Re: [Bug-gnubg] Dice Rolls

[Thomas A. Moulton]

The best was to resolve this is to follow your gut:

If you feel the rolls that the program generates are at fault then
put it in manual dice mode and roll your own!

But! Be HONEST

If you play a few games like that and still lose badly then look
for another cause.

How GOOD is gnubg playing? Is it set to World Class? or Beginner?
Put the dice back on automatic and set the playing strength to something 
weaker.

Here's a case and point:

Where I work we make Factory Automation equipment and to test one of our new
systems we build a slot car track and each lane could be manually 
controlled or
it could playback a recorded lap (throttle speed) and while the head of 
the department
had the record of the fastest lap at 4.82 sec the PLC (computer) had a 
stored lap
of about 5.2 seconds.

Because the PLC could perform that lap with no error every time in a 
race of any length
nobody could beat the PLC because even the slightest errors by a human 
could be costly.
(fall off the track at worst)

The same goes for playing backgammon, gnubg (World Class) will make the 
perfect move
every time, so making a GOOD move every time is not really enough.

Tom

On 08/17/2013 12:11 AM, JD wrote:
> Yes, correct.
> All I am saying is that TO ME, it is sufficiently significant, because
> I will certainly NOT be playing millions or billions of games and note
> the rolls for each given board state.
> Up to now (as of typing this message), I have played about 10 games,
> and out pf 10 games, of 7 points each, I won only one game, even then,
> marginally. I lost 9 games, and each one with a huge disadvantage in
> the number of my pieces remaining on the board.
> Certainly, I did not set out to lose to myself.
> I would have hoped that the engine would not suddenly roll dice that
> would provide doubles to player 0 in the most amazingly opportune time,
> to to roll dice that would hit my piece exactly when it became uncovered.
> NO such dice would roll for player 1. NONE!!!!
> In 10 games, each game with 5 or 7 points.
>
> I am sorry, this may not be statistically significant to mathematicians.
> It is extremely significant to me, as I do not have the time to put in
> 100,000,000 games.
>
> Best regards,
>
> JD
>
> _______________________________________________
> Bug-gnubg mailing list
> address@hidden
> https://lists.gnu.org/mailman/listinfo/bug-gnubg