RE: [Bug-gnubg] Feature Request: Differential reporting of rolloutresults

[Massimiliano Maini]

address@hidden
wrote on 27/06/2009 18:01:38:

> I'll just come back on this, since it bugs me. Your point of view
is
> certainly reasonable, so we may have to agree to disagree. 
> 
> RNG Options
> -----------
> 
> However, I'll pose the question, "Why do we have a choice of
RNG?"
> 
> It's not essential for backgammon; Snowie, JF and BgBlitz don't need
> more than one, and I'll bet extreme gammon doesn't either.
> The only reason can be to remove doubt that gnubg manipulates the
dice.
> I don't think there is much doubt that gnubg is honest among backgammon
> players, therefore the reason for it's existence is gone.

Hmmm, I constantly see people with serious doubts
about gnubg's dice.
No matter the effrt you put in to try to show them
it's fair, they stick
to their silly position.

We probably don't need 5 RNGs: 1 good one (MT), plus
manual dice and an external
one (random.org) should be enough.

>  Differential reporting of rollout results
> --------------------------------------------
> 
> I suppose this is another matter of personal preference. 
> 
> 
> I agree that the seed should still be displayed so that rollouts can
be
> duplicated, but why do we have to have the same information again
and
> again and again. I don't believe anyone avidly reads this information
> for every time it is displayed. The most they will do is scan the
> settings to see if there are changes.
> 
> I think that my proposal makes the scanning easier and the rollout
> results clearer. Reducing the amount of context data makes the important
> data - the rollout - more prominent. It makes it easier to compare
roll
> 1 with roll 2, and so on, by bringing the text closer together, which
is
> the main point of any rollout.
> 
> By showing only the changes, it also makes the changes themselves
easier
> to spot.

One partial solution would be to put the rollout results
first (all the moves/
decisions)and only after, the settings of each move/decision,
even if they are
all identical.
It swon't help spotting the differences but it's fairly
easy to implement and
will improve readability of results.

MaX.