The_Badger, on 2016-August-23, 03:40, said:
but surely the best thing to do is to list and mail all these anomalies to Gibware.com, so that they can look into this. BBO uses GIB software but isn't responsible for the bids it generates.
I thought BBO took over gibware, so the only ones doing any dev-work on Gib today are the BBO-developers?
The_Badger, on 2016-August-23, 03:40, said:
Programming a bridge computer, a friend of mine with a MSc in IT told me, is a 1000 times more difficult than programming a chess computer
Yes, that is quite true, unfortunately (unless you love tough challenges
).
The big difference is that in Chess, ALL input you need is available right there in the chessboard+position itself.
Nothing hidden/unknown at all.
What you need is good position-evaluation algorithm, some smart optimization algorithms, and a lot of computer power --- the more the better.
In Bridge, however, there is this whole lot of unknown/uncertain information you have to deal with.
For example:
- How are the cards placed in the hidden hands?
- What bidding-system are opps using?
- What bidding-system are
we using? --- seems Gib has quite some issues here
- All the different types of bids and bidding-conventions you have to deal with...
- How would the other players act if they held a certain hand? Is it consistent with what they already did?
- What did they
not do? And what does that mean?
- Psychic or irregular bids.
and much more, like percentages, suit-handling, elimination-plays, etc, to mention only a few...