[lycier]

---

[JLilly]

My understanding of help-suit game tries is that with 1.5 or fewer losers in the help-suit, ptr is to bid game; with 2-2.5, ptr exercises judgement, and with 3+ ptr signs off in 3 of the major. Ptr has 2.5 losers in ♥, and so exercises judgement. With 7HCP, ptr's hand is on the weak side of a 1M-2M raise, so ptr passed.

If I were the robot, I'd bump up the value of my hand considering that I have 4 rather than 3 of the trump suit, and T9xx is of much greater value than xxxx when ptr has game-invitational values and five cards in that (bid) suit. OTOH, my short-suit values, inasmuch as they are values, may be for naught. With 5 guaranteed ♠ and a help-suit try in ♥, opener probably has 2 or fewer ♣, so my J♣ may well be worthless.

In the event, you probably had 10 tricks, but the placement of opponents' honors was on the favorable side for your ptrship.

BTW, what was W's lead?

[lycier]

 JLilly, on 2017-March-22, 16:03, said:

My understanding of help-suit game tries is that with 1.5 or fewer losers in the help-suit, ptr is to bid game; with 2-2.5, ptr exercises judgement, and with 3+ ptr signs off in 3 of the major. Ptr has 2.5 losers in ♥, and so exercises judgement. With 7HCP, ptr's hand is on the weak side of a 1M-2M raise, so ptr passed.

I disagree.
Obviously, what you said have difference with robot evaluation. Gibs only apply for Total Point evaluation, never count losers.

[JLilly]

Lycier, can you clarify your comment?

"what you said have difference with robot evaluation."

So humans would predictably evaluate differently?

"Gibs only apply for Total Point evaluation, never count losers."

You have information that the robots don't count losers in responding to game tries?

[smerriman]

It's true that GIB doesn't count losers at all. It's not true that it solely uses total points; it takes 8421 points into effect with game tries, though exactly how, I'm not sure anyone knows.

AQ seems like it should be enough to accept though despite the minimum.

[lycier]

 JLilly, on 2017-March-22, 18:36, said:

Lycier, can you clarify your comment?

"what you said have difference with robot evaluation."

So humans would predictably evaluate differently?

"Gibs only apply for Total Point evaluation, never count losers."

You have information that the robots don't count losers in responding to game tries?

Of course, yes, I do.

Look at this screenshot.





Here is Gib system notes link .

[smerriman]

That is an overview. It says nothing whatsoever about how GIB responds to a game try.

[JLilly]

Lycier, the snapshot you posted doesn't mention anything in particular about counting losers. It also doesn't mention lots of other heuristics that humans use, and indeed is explicit that GIB makes decisions in part based on statistical results from double-dummy simulations. It does however mention the HCP and shortness-point-count heuristics that humans use. Is point-count the only heuristic it uses? Even if so, how does it weigh point-count against its simulations?

[lycier]

[Stefan_O]

 lycier, on 2017-February-13, 17:30, said:

Why does Gib reject long suit game try when with good support?

Doesn't it just use simulation, and simulation indicated 4♠ would go down more often than it makes?

[lycier]

 Stefan_O, on 2017-March-26, 03:35, said:

Doesn't it just use simulation, and simulation indicated 4♠ would go down more often than it makes?

For Gibs, what's the meanings of long suit game try? How to respond long suit game try?