[goffster]

Perhaps utterly deranged.
Bbo forums robot tourney.
https://tinyurl.com/y9eqd8b7

2s!? Lack of double of 1n!?

2s even more incomprehensible since I am known to not even hold 3s
What is the matter with 3D if it feels it simply has to bid?

[johnu]

♠J854 is an excellent suit. Not quite good enough to jump rebid 3♠ so that's why 2♠ was chosen.

[nige1]

goffster 'Perhaps utterly deranged.Bbo forums robot tourney. 2s!? Lack of double of 1n!? 2s even more incomprehensible since I am known to not even hold 3s. What is the matter with 3D if it feels it simply has to bid?'
++++++++++++++++++++++
IMO
- GIB used to bid better but recent ad hoc tinkering (to address specific issues) has introduced more bugs than it fixed.
- What GIB needs is a new bidding program, specifying bids in open-source readable tables (e.g. FD or Kungsgeten's mark-up). This would help to increase coverage and to avoid stupid mistakes. It would also allow GIB to understand opponents' bidding (when presented in that new format); and to play its partner's preferred system.
- GIB would be much better if it underwent rigorous regression-testing after each fix -- to mitigate the introduction of new bugs,

---

[nige1]

a simple regression-check would be pysch-detection: high-light cases where the systemic description of a call differs widely from the actual hand (as in Goffster's example). Sometimes there is a tactical argument for a psych; but rarely in uncontested constructive auctions; and GIB should be especially careful to avoid CPUs

[goffster]

I would like to add, it was robot that bid 2s, not me.