[Wayne_LV]

I would think a Robot that plays SAYC per the ACBL Booklet

My link

would be a popular feature for BBO.

The current GIBs bid and play pretty badly, as evidenced by the fact they are relatively easy to beat. I realize that their win/loss performance is a factor of the humans at other tables playing with them, but my partner and I consistently beat them to the tune of over .6 IMP per board.

I would venture a guess, not having any statistics available other than numerous observations, that the vast majority of BBO players are playing SAYC.

I would also venture a guess that most of them are not playing the complete system (i.e. such conventions a J2NT and splinters are seldom evident on players profiles or in play.)

One of the most useful purposes for renting GIBs is to practice bidding and play. Would it not be more desirable to have SAYC robots that beginning and intermediate players could practice with and learn from. Would it not be more of a measure of progress to play against opps playing the same system you play?

Further I would suspect that a number of players not currently using GIBs would rent a SAYC Robot for practice and learning thereby generating more income for BBO.

Wayne

[Stephen Tu]

Wayne_LV, on 2012-September-01, 17:13, said:

The current GIBs bid and play pretty badly, as evidenced by the fact they are relatively easy to beat. I realize that their win/loss performance is a factor of the humans at other tables playing with them, but my partner and I consistently beat them to the tune of over .6 IMP per board.

If you are sitting mostly N/S when you play with them, I think you aren't really beating the GIBs so much as the human/GIB partnerships and a few other human/human pairs sitting N/S. Average human is kind of bad, so it's pretty easy to beat the comparisons your way. If you want a tougher game, rent the advanced bots, and sit e/w! Or even just renting advanced is enough, they really do exhibit substantially better bidding judgment than the E/W mostly basic bot field, and play much better. My partner and I beat the bots by a substantially smaller margin after I started renting the advanced bots.

It might be hard to program the robots to play SAYC well. A lot more temporizing and bidding fragments after 2/1s to create a force. No forcing minor raise is pretty ugly also.

I don't know that SAYC really is all that much simpler than GIB's card. A half dozen extra conventions?

There's an argument that beginners/intermediates might be best served going ahead and learning to bid in the same fashion as experts, Larry Cohen is arguing that we should just start people off with simple 2/1. Just evolution of the game, we switched from teaching beginners 4-cd majors to teaching them 5 cd majors, it's not that huge to teach them to bid 1nt with a wider range. Beginners can muddle along, read GIB's bid explanations, make their mistakes, look at what other tables do. One thing I think that might be a good feature is for the GIB MBC boards to include one board played by 4 GIBs. That would give a baseline for at least how the robots *think* it's supposed to go. Sure buggy, but often better than what beginners/intermediates might dream up.

The thing is I'd prefer the programmers concentrate on the many bidding bugs reported in this forum. Branching out to multiple systems increases the workload considerably IMO, I'd rather work on getting GIB clean on one system than playing two more buggy ones.

[Wayne_LV]

The problem I have with using the advanced robots are they are too expensive. I think a more fair offering would be the advanced robots at a price somewhere in between the current price and the price of the inferior robots and scrap the weaker version. Normal practice when a software product is improved is to stop selling the outdated version.

If players are to play with robots, I would think it would be far better to play with one that is more predictable and plays a system they understand. Defending is normally a lot of guess work against players and systems that are, for the most part unpredictable, so what the robots play as opps is of little importance.

[semeai]

Stephen Tu, on 2012-September-02, 01:18, said:

It might be hard to program the robots to play SAYC well. A lot more temporizing and bidding fragments after 2/1s to create a force. No forcing minor raise is pretty ugly also.

I don't know that SAYC really is all that much simpler than GIB's card. A half dozen extra conventions?

There's an argument that beginners/intermediates might be best served going ahead and learning to bid in the same fashion as experts, Larry Cohen is arguing that we should just start people off with simple 2/1. Just evolution of the game, we switched from teaching beginners 4-cd majors to teaching them 5 cd majors, it's not that huge to teach them to bid 1nt with a wider range. Beginners can muddle along, read GIB's bid explanations, make their mistakes, look at what other tables do.

In fact 2/1 seems potentially easier for humans as well as robots. The extra conventions are a minor problem, but as you point out you can read the explanation, at least if you know to do that.

I wonder what percentage of the population does understand the bit about reading the explanations of even your own bids before you make them. Is it nearly universal or not? This seems hard to test.

Wayne_LV, on 2012-September-02, 16:45, said:

The problem I have with using the advanced robots are they are too expensive. I think a more fair offering would be the advanced robots at a price somewhere in between the current price and the price of the inferior robots and scrap the weaker version. Normal practice when a software product is improved is to stop selling the outdated version.

My understanding is that it's not that the software is outdated for the cheap bots, but rather that they're given less thinking time, and are thus cheaper for BBO. No comment on the price (who wouldn't want a cheaper price?) except to point out that they're not seven times as expensive unless you play every day of the week.

[Quantumcat]

Wayne_LV, on 2012-September-01, 17:13, said:

The current GIBs bid and play pretty badly

No they don't! My GIB partner recently played a perfect elimination, and was beaten by his GIB opponent who performed a Crocodile Coup to prevent his partner being endplayed. In another, I was declaring, and knew I would need an endplay later, so tried to cash an ace early (trick 2) so the GIBs wouldn't know to throw away their doubleton honour, as they might if I do it later. Guess what the GIB did - he stuck his doubleton king in, in second seat. What a champion! Do you call that playing badly? I have seen them perform many other excellent plays that even experts might make a mistake with. When they get a bad result, it can easily be because they made a correct play which happened to fail in that situation, when everyone else did the "natural" or "obvious" thing, which turned out to be right.