[Bug-gnubg] Re: Re: Revised tutorial

[Albert Silver]

Actually, I have thought about it. There are two problems:

- From what I've understood, it requires some very special format
rules that I know nothing about.

- Also, from what I've seen, it is a text-only manual. This is a huge
impediment in the usefulness of the manual IMO. I spend quite a lot of
time explaining how to use and setup GNU to a number of players. In
fact, this helps give a clear idea on its accessibility to the general
public, and how it might really compare to a commercial product such
as Snowie. Even with the multitude of pictures in my own 'manual'
(hard to call a 50-page document a tutorial now), many people have a
lot of trouble with it.

I'm going to skip over the good parts of the program, which are many
times many, so don't think this is open bashing. I just want to point
out a couple of things that make it still lag behind, or break even at
best.

- Relational database: this will never be used as it is. As it stands,
it will only be used by a programmer who also has a serious vested
interest in backgammon. As far as I know, this means only GNBG
developers and possibly a small handful of other users. Less than 10 I
am certain. I myself tried and ran into roadblock after roadblock.

- Setting up and analyzing a position. Saving the results. This is the
part I spend the most time explaining as a rule. There are too many
counterintuitive issues still for the user. One can learn the special
steps, but they are mostly illogical, and far too many users freak in
my experience. It's a bit as if I told you that to get into your house
you had to dance on one foot or the key wouldn't enter. The problems
(not including a couple of small bugs) are:

a) You cannot set up a position upon entering the program. The Edit
button is greyed out, and one MUST start a new game or match first.

b) Upon setting up a position correctly, one must not try and analyze
the position immediately. One must exit Edit mode first by repressing
the Edit button, and only then ask GNU's advice.

c) Even if you setup your position as you wanted, and analyzed it, you
can still lose all your info by having the disastrous idea of saving
it as a *position*. It seems completely natural to save a special
position with analysis as a position, but here this means losing all
the analysis.

d) Inability to see the rollout settings or detailed results. Snowie
solves this with a floating tip, so when you pass the mouse cursor
over the move, you will see a floating tip appear with the details of
the rollout. Such as the number of trials, the error margins, etc. In
GNU the only way is to export the data somewhere.

e) Finally, there is the issue of taking a position from a game/match,
modifying it or the score, and then reanalyzing to understand the
effect the changes have on its evaluation. One can't begin to stress
how important a tool this is to learn and improve one's game. Aside
from the rollouts and full match analysis, it is easily Snowie's most
important learning tool.

I love the program, as is clear, but as to lagging behind, the above
on setting up and analyzing positions is where it really lags behind
the competition IMHO.

Albert



On 9/20/06, address@hidden
<address@hidden> wrote:
> >I'm going to make two versions: one downloadable and portable such
> >as PDF (with indexing), and a page-by-page HTML version, with a bar
> >of contents.
>
> Great. In fact I don't know if it makes sense to keep on having a (poor)
> gnubg manual and a good "tutorial", it's kind of wasting efforts.
> Ever thought about working on the manual ? IMO, the manual is the
> main (and probably only) area where gnubg is behind its competitors ...