etha

ok got to 50 boards. 63 imps to the preempter. If I work out a way to do this quickly I'll try to see how mad we can go with the preempts and gain. I'll stick to 3♣ favorable first in. Not sure I can go as far as 5 card suits but 6 bad might be possible.

up to bd 42 going a bit mad might have to find an easier way. One thing I have noticed is the preempter never bids again unless forced. Is it normal to have a 4♣ raise either after a bid or not to suggest opener do something either bid on with extras to make, or to bid on with more shape (sac) or to bid on with no defence(sac) or to dbl with unexpected defense or some combination of these? I seem to remember maybe Robson suggesting 3nt was either to play or I want to sac and 4♣ was trying to involve partner in the decision. Can anyone tell me about these sequences. And does anyone play action dbls, which always seemed a recipe for disaster when I heard about them?

I suggested this to jec a long time ago he wasn't interested. If someone wants to play against a team of wbridge or a team of qplus I can do it if I get some volunteers. Jack might be possible but is harder because we would need some people that owned jack. With jack the hands would also probably need to be predealt just because it has no easy to use play one hand feature.

the last time I tested gib against top programs e.g. wbridge and jack, It was losing about an imp a board. I stopped playing bridgez because wbridge is so annoying to play with e.g. 2nt = both majors reply to stayman amongst other things. This test was a few years ago but i can't imagine it has changed much. If the robots bring in money for bbo or they want to make them serious I have no idea why they don't make an arrangement with at least one of the top 5 programs. Gib hasn't competed in the world computer championships for ages mainly because it would have no chance.

Ok so I completed the first playthrough of the 100 hands with mainstream set, but I made a slight mistake and it set ew to moscito and ns were playing sort of sayc. Anyway then I tried to get jack to play its own tournament that had been created by this and it went what I've already played this and refused. So I have to actually have a person not jack sat in a seat so I chose east cos then I just open 3♣ every hand and hit auto play. It is now using the right cc so we may get to see if playing a different cc matters more than the opening of 3♣. I'll redo it with the mainstream setting with the right cc so we can see.

I had a few thoughts about this. What are you doing about position and vulnerability? It seems to me there can be 16 systems depending on these. You surely want to have different agreements based on seat and vul to be optimal. Is the plan to make a system your bot can only play with itself? Is the plan to make the best system you can play with another bot? Or is the plan to make the best system you can with a human with a decent but not infinite memory? If the latter you would need to describe the system bids in some form a human can understand which would be very different from the others.

Moving on to Jack. Jack is the best available robot. It has 8 different preempt settings. On the 3 most aggressive it bids 3♣. So I deal 100 boards setting our hand as East and dealing the others. Next I plan to get jack to play them all with the preempt set to main stream and then again with it set to aggressive, the least aggressive it can be and still open 3♣. I will set jack to use the settings it does in the world computer championship and using that cc by both sides. I'll be back with the results which may take a while.

Ok so the least reliable method I have is to give the hand to q-plus. Q-plus is about the 3rd- 5th best program available in terms of results in the world computer bridge championship. It does have some interesting features though for inputting your own bids and doing simulations. So first I set the opponents to play take out dbls and leaping michaels and then I set the opener to aggressive, normal and conservative 3 level preempts. It chooses to open 3 ♣ at all settings. Then it simulates what will happen if you pass, open 1♣or 3♣. 1000 hands is the max however so that's what I chose. And the results at normal and conservative were almost identical -1.2 imps for 1♣ and -0.5 imps for pass. Aggressive led to 1♣ -1.0 and pass at 0.0 imps. So that was quite interesting I'll see what happens using other methods later.

Here is a simple hand. 1st position favourable imps. [hv=pc=n&e=st54h3dkt6ckqj973]133|100[/hv] So I think the modern trend is to open these 3♣ but I am sure there are people that think this is a clear pass. I am going to try by using computer A.I. to decide whether this is good to open or bad. Feel free to do the same or to make any comments you like. One thing I can't do but someone might be able to is look for similar hands played in real games and see what the result is there.

Why did you bid 3♥ that used up a lot of space for no good reason. Just bid 2♥ unless playing negative free bids. A good general rule is if you have several option that seem to make sense in a constructive situation bid the lowest. I call this garvey's rule but I'm sure someone came up with it before that.

the opposition had to bid 6♠ or higher 74 times to get to par in my 100 hand sim from before.

Hmm well that plan wasn't very successful because spades going off is almost always par. 6♦ made once, 6♥-1 was par once, and 4♥ 5 times. This is over 100 hands dbl dummy.

I'll see what happens if I fix the south hand and make the other hands comply with the bids up to 3♠ and then simply determine the dbl dummy outcomes of these hands. Computer will speed this up and not be insane it only goes mad after the 3♠ bid.

Oh and I gave it to Jack who I expected to demonstrate robots are clueless on these hands. It did not disappoint. First it had the insane agreement that 4♦ was non forcing and second it bid 4nt ace asking over 3♠. It never wanted to play in diamonds which was at least sensible as far as I can see.

Can someone explain when going past 3nt with a new suit is a cue and when it is a new suit?

I gave it Jack for an unbiased answer. He played both in diamonds. Even in sayc he didn't think north was worth anything but 1nt.

Jack the computer program uses these. I think they are terrible personally and would go with Nige1's approach.

if you want to keep the impossible negatives in. Reese thought exactly as you do in precision bidding and precision play he wrote "I feel it is uneconomical to devote so may sequences to them" He proposed over 1♥ to use 2nt for all of the impossible negatives.

There are quite a few relay methods where 1♦ is still negative and doesn't include the 4441 hands. For example the old moscito relayed with 1♥ = 4+ spades, 1♠= 4+ hearts etc. This avoids declarers hand being known mostly. Versace on the rare occasions he plays precision also has 1♥ being spades. meckwell in 2006 at least have 1♦ as negative only with 2♠ through 4♣ used for 3 suiters. So I think it seems most have moved the 4441's out of 1♦. Muller de wijs and Tarzan precision also did away with it. Another thing you have to decide on is whether 1♣ 1♦ 1♥ uncontested shows just a good hand, or hearts and balanced, or just hearts.

I'm pretty sure LTC is just worse than a point range and number of trumps as a method of hand evaluation, which pretty much makes the whole idea not worthwhile.

We played 1nt as non forcing and 2♣ as the relay over 1♠. We were playing the transfer openers were canape though. And it was a strong club system so the openers were limited. I'm never keen to not have a natural ish 1nt bid available to responder, but I haven't got any solid data to say its bad.

If you are in London go to the Young Chelsea Bridge Club if at all possible it is the best club by miles. There will be internationals there every time you go. All systems are allowed and I'm not sure what the split is between acol non acol but I would guess there is more 5 card majors strong NT than it now but I haven't been in a long time. Elsewhere acol still reigned supreme last time I looked but there are bound to be some 5 card major strong Nt people at almost any club these days presumably except Pinner.

So anyway to sum up I dealt 100 hands where the NS hands were the same and west bid 2♦. 6nt is a make on 57 of these and there were 3 where 6♣ makes and 6nt does not. No idea if there were any where 6nt made and 6♣ doesn't, pretty unlikely east has a void. 9 of the 6nt were only makeable by south. surprised we had no strong 1nt openers and no 14-16 nt openers which rather defeated the bidding problem as its much harder to investigate 6 now. I may do some more of these.

how do you want him to invite game ?

with 4+ hearts wouldn't opener always bid 2♥ on this auction? So you know there is no heart fit. partner might have 5 diamonds and be trapping too? Or would dbl be penalties already?