Re: [Bug-gnubg] GnuBG offline

[Rod Roark]

See 4.5 and 4.7 at:

http://mole.dnsalias.org/%7Eacepoint/GnuBG/gnubg-faq/

And then some of us have noticed that on FIBS the GnuBG-based
bots all seem to have ratings in the 1900's and 2000's....

-- Rod

On Saturday 07 February 2004 10:55 pm, Peter Stewart wrote:
> I find the claim on your website that "It currently plays at about the 
> level of a championship flight tournament player" to be unsustainable, 
> indeed laughable.
> 
> The programme I have downloaded _cannot_ play well at all, and depends 
> entirely for its outcomes on producing optimal dice for itself at 
> critical points in each game.
> 
> I am not an outstanding backgammon player, but enjoy the game - in other 
> words, a typical prospective customer for excellent software if I could 
> only find any.
> 
> GnuBG is, without doubt, the least enjoyable of a pretty poor bunch of 
> computer opponents I have tried.
> 
> It is quite easy to beat, losing gammons and backgammons fairly often 
> and only able to deal effectively with a narrow range of playing styles. 
> But more to the point, there are consistent practices written in that 
> ruin any pleasure in the game and in fact render the software worthless 
> in its present form.
> 
> For instance, when hitting a blot in its home board early in a game, 
> there is an extremely high frequency of double-six dice for the 
> opponent, or of locking out from re-entry by other statistically 
> unlikely means.
> 
> Also, it is very common for the opponent to be trapped in GnuBG's home 
> board behind small arrays because repeatedly unable to get a combination 
> of high & low numbers on the dice.
> 
> But most absurd is that the programme doesn't bear off straight away 
> when it is obviously going to lose, until what it assesses to be the 
> last moment - and this can be _after_ a double throw would have cleared 
> the player's board, so that GnuBG is obviously calculating the next 
> throw in advance or taking a stupid risk of losing gammon for no reason 
> at all..
> 
> Out of interest, I tried building a particular array in my home board, 
> with points 2-6 made and leaving blots on points 1 and 7, then hitting a 
> GnuBG blot.No less than  _four_ times out of five the next throw for 
> GnuBG was 1 and 6, taking out both blots. This is utterly absurd, and a 
> programme that needs to resort to such methods is unworthy of anyone's 
> time & effort to play with, at any level of expertise.