[NickRW]

lackeman, on 2015-March-31, 02:49, said:

A computer cant be "aware" of whats happened in the "past" like a human can...

You seem to be unaware of the capacity of even vaguely modern computers. A computer certainly can be aware of past events and be programmed to accumulate this information on a on-going basis and, indeed, accumulate data about it's opponents and their tendencies too.

[diana_eva]

I have no clue what's going on in this thread at all... I guess bec I'm not an expert.

So is lackeman trying to say computers can't be taught? or the other way around? translate for me pls

[lackeman]

Hi nick!
But how do u built together a lot of information and from that your "thought" of what is right to do her in this particular situation?
Much info from bidding, opponents act, carding, skill, their play ,thinking, making IT hard for them to find etc? All this and more are factors that humans use and that build upp our picture of where the cards are. Also much of this is build up of our experience of players Way to act and for ex "simular situations", there are also factors like our position, vul, Even when a psycic bid was done etc. Also when waluate the Cards in bidding, "point system", what position relative for ex "opner"
All of this kind of factors in my opinion Will be hard for a computer to evaluate and draw conclusions at and act upon. There are so much and diffuse for a computer to act on and if IT does I am sure we often Will se incredable bad act from IT!
If ex if a anknown player does a terrible misstake then u will probably judge him as "incompetent", so if the computer do silly misstake like that it maybe also kan be judged!?

[hrothgar]

lackeman, on 2015-March-31, 06:22, said:

Hi nick!
But how do u built together a lot of information and from that your "thought" of what is right to do her in this particular situation?
Much info from bidding, opponents act, carding, skill, their play ,thinking, making IT hard for them to find etc? All this and more are factors that humans use and that build upp our picture of where the cards are. Also much of this is build up of our experience of players Way to act and for ex "simular situations", there are also factors like our position, vul, Even when a psycic bid was done etc. Also when waluate the Cards in bidding, "point system", what position relative for ex "opner"
All of this kind of factors in my opinion Will be hard for a computer to evaluate and draw conclusions at and act upon. There are so much and diffuse for a computer to act on and if IT does I am sure we often Will se incredable bad act from IT!
If ex if a anknown player does a terrible misstake then u will probably judge him as "incompetent", so if the computer do silly misstake like that it maybe also kan be judged!?

So what?

Computer programs such a Jack already play much better that the overwhelming majority of players.
No, they aren't perfect and they make other types of mistakes than humans, but their level of play is very solid.

It's been a long time since Zia stopped being willing to wager significant quantities of $$$ on a bridge match versus a team of computers.

The main reason that computers aren't better than they are right now is that the market for strong bridge player computers isn't large enough to attract serious sustained interest. If IBM or Microsoft were willing to spend the money that IBM did to on a chess playing program you'd shortly see a computer beating world champions.

[NickRW]

lackeman, on 2015-March-31, 06:22, said:

Hi nick!
But how do u built together a lot of information and from that your "thought" of what is right to do her in this particular situation?...

Well, at the moment the problem is hard because, as others have pointed out, there is no money in it (well, not a lot anyway).

However, if you're interested in computers playing cards and doing it well, look at http://poker.cs.ualberta.ca/ (which is primarily about poker), especially the publications page. (Warning - this is dense technical stuff). Some of those poker bots certainly do "read" their opponents.

[helene_t]

NickRW, on 2015-March-31, 04:38, said:

A computer certainly can be aware of past events and be programmed to accumulate this information on a on-going basis and, indeed, accumulate data about it's opponents and their tendencies too.

I just wrote a goldfish simulator that deletes all files created more than two minutes ago but it makes my computer crash. Any geeks here who can help me out?

[NickRW]

helene_t, on 2015-March-31, 07:56, said:

I just wrote a goldfish simulator that deletes all files created more than two minutes ago but it makes my computer crash. Any geeks here who can help me out?

There was something in the news a few days back (iirc)about a fish that remembered her ex owner months after being given to someone else and the fish could pick out the owner in a room of people.

So... suggest you change your simulator to deleting things a year old. Wait. It must have already deleted itself by now

[lackeman]

Hi hrutgar so does IT means that it is only one computer that beat the worldchampion? Since IT is so exepensive? So the rest is moderate player? Well this doesnt seames for me as a description of reality!
Esther as i remembered ther was many computer for as long as 20 years that beat human!
For me this support my explanation and if tru says that it is atleaste much easier for a machine to beat a human in chess than in bridge.
If I am right that IT much harder to produce a good machine in bridge then in chess. Then we would expect that IT Will costs much more money and IT Will take much longer time before we se a bridgecomputer at the same level copaired w a chess computer!
And that if hypothesis are correct is what we can se ie reality been concistent whith what we could expected!
IF IT only were a matter of economics (ie not harder but same dificulty to produce bridge and chessrobots of same strength IT would not be a history like this).

[lackeman]

So how much money does it taskes to produce this machine ?
And remember that the computer should also be 25 years old also if we shuld be able to compare here!
As said Im not an expert at computers but IT should matter also!

[Dinarius]

I want whatever lackeman's on!

D.

[johnu]

lackeman, on 2015-March-31, 10:19, said:

As said Im not an expert at computers

Maybe slightly out of context, but I 100% agree with this statment.

[Antrax]

PhantomSac, on 2015-March-31, 02:24, said:

Maybe it can't be proven but it seems like just using common sense it is impossible to imagine that computers could not pass humans in bridge. It seems inevitable.

I suspect it's because it's difficult for people who aren't computer experts to understand just how difficult some problems are, to a computer. If forced to guess I'd wager you're right, but I can play devil's advocate here.

For instance, let's talk system. On one hand, if you let the computer play whatever, it can probably play some information-theoretic perfect system that a human would have no hope of understanding or remembering , which would give it an edge. But, such a system would come from a human, probably using a different algorithm to create that system. If the task were instead that the computer learn existing systems and augment them, as human experts do to taylor it to their style and close gaps, that would be one hell of a more complex computer program.

Another issue is something like table feel. if everyone plays online then the computer has the edge since it's all compute. However, put a computer on a human team and seat it at a table, and you might find out the +EV perfect play was wrong due to a tell picked up by the human declarer at the other table, lose 12.

Finally, this is the most tenuous but deep inferences, like "why did he return a club when a diamond seems auto oh he's trying to give me a losing option so he has the trump queen" - good luck teaching a computer those.

these are all small things so in reality probably computers will dominate bridge at some point, but it's not as automatic or obvious as may seem at first.

[rhm]

hrothgar, on 2015-March-31, 06:28, said:

The main reason that computers aren't better than they are right now is that the market for strong bridge player computers isn't large enough to attract serious sustained interest. If IBM or Microsoft were willing to spend the money that IBM did to on a chess playing program you'd shortly see a computer beating world champions.

I am not convinced.
This is unproven. You could be right but you could also be very wrong and the market for strong chess player isn't so big either.
In fact we also do not know how much has been invested in Bridge computer algorithm. It could well be there is not such a great difference, people always claim.
In my youth I remember having read an article from an eminent chess writer, why computer will never beat grand masters in a serious game. He was obviously wrong.
But I still do not subscribe to the notion that computer can beat human mind in almost any field of endeavor if enough is invested.
It is not even clear whether everything of substance can be trimmed down to programmable computer algorithms, but mathematical research done by Alan Mathison Turing almost a century ago suggests not.
For example computers are nowadays sometimes used when proving mathematical theorems.
But I doubt that computer can substitute the ingenuity of researchers in mathematics.
In the end you have to find algorithms you can program which are better than the human mind, which works on experience, associations and inspiration. Creativity is hard to program.

Bridge could well be a game where this can not be accomplished no matter how much you invest. This does not make Bridge a better or worse game than chess.

Rainer Herrmann

[mike777]

"Creativity is a phenomenon whereby something new and somehow valuable is formed, such as an idea, a scientific theory, an invention, a literary work, a painting, a musical composition, a joke, etc"
http://en.wikipedia....wiki/Creativity

I understand computers already are very good at creating new stuff. The new stuff created in 2015 may not be somehow valuable however I expect and see no reason why over the next 35 years or so the new stuff that is formed will not somehow be valuable.

I expect we will gain a much better understanding of how the software and hardware of the human brain works over the next 35 years.

[billw55]

Because bridge is often a game of percentages, there is a sort of hard cap on how "good" any computer (or human) can be. Even a flawless player will lose a double finesse 25% of the time, and so on. And therefore, they will lose matches to inferior opponents some of the time. Compare backgammon: the bots are vastly and irrefutably superior, but humans (even me) still can win matches on occasion.

Against all that, computers never get nervous, or tired, or angry, or go on tilt. They never miscount and never forget a single played card down to the last spot. Yes, I think it is pretty clear that they could be better than human.

Human ego is strong in this respect. People should have learned from the IBM experience with chess, but many are still stubborn. Another candidate for the "computers will never win" crowd was Go. But there too, the machines are advancing relentlessly.

Really the whole premise says much more about human psychology than about technical feasibility.

[mike777]

Emotions are being programmed into computers. I suppose one may argue they only mimic emotions but what will the difference be.

[NickRW]

mike777, on 2015-April-01, 08:20, said:

Emotions are being programmed into computers. I suppose one may argue they only mimic emotions but what will the difference be.

[on Dave's return to the ship, after HAL has killed the rest of the crew]
HAL: Look Dave, I can see you're really upset about this. I honestly think you ought to sit down calmly, take a stress pill, and think things over.

[Free]

I think there are many issues why bridge programs aren't better than they are at the moment. One is obviously money, but I think the most important is that we lack calculating power.

Some have compared it to chess. But that's a whole other game! Suppose we have unlimited calculating power. In that case you could calculate the entire game of chess and you'd never be able to win against a chess computer unless the game is predetermined that one of the players can always win. A bridge computer could only manage to simulate every deal, determine which actions are best for each player on that deal, and make percentage actions. This means they will fail from time to time depending on the hand, so a human may have an opportunity to win. In the long run the computer will obviously win, but in a single match it would be beatable. This is by far the most important to realise.

At the moment we lack calculating power, so we have to do it with limited resources. A chess computer can easily disregard poor looking moves and investigate further on good looking moves. It also has full information, which means it doesn't have do run simulations. It makes a move and calculates the best move for its opponent, does the same for itself,...

In bridge, you have to run simulations inside simulations to get to the optimal line, which makes the complexity exponentially big. Running simulations alone isn't enough, because for example (like someone mentioned already) holding Qx behind dummy's AKJ would mean you'll always lose your Q. No, you have to run a simulation for each player what the action will be in your own simulation.

Also note that I haven't mentioned the bidding...

[mike777]

computing power does increase exponentially. The disagreement seems to be for how long.

I believe one aspect in this discussion that is overlooked is how will human performance be enhanced. We all seem to believe that computers will improve performance at bridge.
What will the future bring in terms of human enhanced performance?

[johnu]

Free, on 2015-April-01, 10:09, said:

I think there are many issues why bridge programs aren't better than they are at the moment. One is obviously money, but I think the most important is that we lack calculating power.

...

At the moment we lack calculating power, so we have to do it with limited resources.

mike777, on 2015-April-01, 10:39, said:

computing power does increase exponentially. The disagreement seems to be for how long.

We've got more than enough computing power right now. Since this is the Bridgebase forum, have you seen how fast the play goes when GIB is declarer against 2 other GIBs? The cards are usually played so fast you can't follow the play because everything is a blur, and Bridgebase computers could be running thousands of GIBs at one time.

Instead of a fraction of a second on a play, give GIB 2,3, 10+ seconds to make a difficult bid or play, so 10x - 10000x as much time. Do you think GIB will bid or play substantially better? There may be some edge conditions where the number of simulations is not sufficient to model unlikely distributions so e.g. GIB may miss a 100% safety play, but GIB will still make almost all the same unbelievably bad bids and plays. Why? Because the programming either has bugs or omissions or just doesn't have the coding to do the correct analysis.

So it's a software problem. Suppose you want to play Precision club with GIB. You can't do it because GIB hasn't been programmed to play Precision. No increase in computing power will ever change that. It's not just the opening bids that need to be programmed, but responses, rebids, responder's rebids, etc. Then for defensive bidding, you have to program a defense to a strong 1♣, a nebulous 1♦, a different 1NT range, natural 2♣, 3 suited 2♦, etc. Not that it can't be done since IIRC some other programs can play Precision, but then you will introduce a whole new class of bugs that need to be fixed and GIB programmers have trouble fixing some of the existing ones. Maybe GIB isn't the best engine to use these days. Some people have said that Jack is very good. I'm sure Jack has its own unique set of issues. Maybe they can be fixed and Jack improved to be world class. Just as likely IMHO is that a new program that hasn't been invented yet will be the first world class ability program that will build on the development of today's programs and be designed from the ground up to avoid some of the problems with today's programs.