Silver Creek Entertainment

I realize they're AI, but they routinely make the worst plays. Some of the ones I notice most often:

1) Your partner leads a queen with nothing to back it up. That leaves their partner (you) to guess. Am I supposed to step on that with the king or ace if I have it? What does that mean? It's certainly easy if you have both the ace and king. You let it go, of course. If I have the king or ace and let it go by, it will likely be taken by west. But you don't know that. Maybe your partner has the other card above the queen. This is so consistent, it's maddening. Burning the queen seems to be the favorite thing for your AI partner to do. It's a stupid lead. If you have any intention of making it good, you do NOT lead it.

2) Your partner is very adept at taking tricks when you don't want them. It virtually never fails. Both you and your opponents have already made your bid. So the only thing left to do is avoid overtricks. Except that your exceptionally retarded AI partner can't seem to figure that simple fact out. You lead, or follow east's spade lead with the ace of spades from your hand to get rid of it. This of course is a perfect time for your partner to get rid of their highest spade in the hopes you can get your opponents to take the rest of the trump tricks. Does your AI partner ever see this obvious play? No! The moron throws their lowest spade and then takes more overtricks with the high spade they had the opportunity to ditch.

3) Your partner is blind to winning cards. In any hand, the way to make non trump cards good is to lead them early and often. The longer you wait to play them, the more likely they will be trumped. For instance, your AI partner will play the ace of hearts to start the hand and pulls it in. You find out later in the hand it also had the king of hearts, but didn't play it immediately after the ace! Why? Your partner doesn't like to make their bid or something?

4) Your AI partner has no clue how to set a nil bid. You can count on your partner almost 100% of time to trump your lead when any human players would know you have the nil bid trapped and can force them to take the trick. Does the dolt throw out an off suit card when the nil hand is caught with the high card? Nope, trumps it.

5) The AI partner has no concept of drawing spades out of your opponents hands. Making a medium high to high team bid is usually very difficult because the AI partner will not lead spades after they have been broken to get them out of the opponents hands. I can't tell you how many hands have been set because the AI player keeps leading off suit cards into a trump by the opponents. They will not lead a spade until they are either out of off suit cards, or nearly so. But by then, your partner has already set your team by allowing the opponents to trump two or three tricks they never should have been able to take.

Unless it is a bagging hand, why would you put the King on a Queen lead by your partner?

If the bot has the King and Queen, it obviously should lead the King and not the Queen.

If the bot or a real pard leads a Queen and you have the Ace, this is the opportunity to either take a finesse or not take a finesse.

Unless it is a bagging hand, why would you put the King on a Queen lead by your partner?

If you have both the King and Ace, as I noted, you wouldn't. But if you only have one or the other, you then can't know if it's safe to let it play as the high card on the trick. An unrealistically high percentage of the time, west has the other high card. So if you put your King on the Queen, you lose both next best cards to the Ace, and then most likely any other tricks in that suit. If you play your Ace, west will simply hold the King if it can. You can't force the King out of your opponent's hands by leading a Queen. That makes leading the Queen with no other higher card of that suit in your hand a dumb play.

If the bot has the King and Queen, it obviously should lead the King and not the Queen.

Correct. But the problem is, the bot relentlessly leads the Queen when it has neither the Ace or King.

If the bot or a real pard leads a Queen and you have the Ace, this is the opportunity to either take a finesse or not take a finesse.

There's nothing to finesse. Your AI partner is simply forcing you to take a high card with another high card instead of saving the Queen for a possible winning card later in the hand.

I guess that I will try again.

If the bot leads the Queen, why would you put your King on it, independent of who or who may not have the ace?

Also, if the bot leads the Queen, and you have the Ace, how do you know that there is nothing to finesse?

If the bot leads the Queen, why would you put your King on it, independent of who or who may not have the ace?

Yes, I was just reading my note again, and that would be a dumb thing to do. There's no point in taking a Queen with a King. Since they're both your team's cards, they have the same value. I meant if only if you had the Ace, what would you do?

Also, if the bot leads the Queen, and you have the Ace, how do you know that there is nothing to finesse?

And that would also be true in a real game. It would be a logical finesse. Sadly, if you don't play your Ace on the Queen to see if east was holding the King (to try and prevent simply losing it to the Ace if you have it), a very abnormally high percentage of the time, west has the King. You can almost count on losing the trick if you don't play your Ace.

But beside strategy, the most annoying part is the bot almost never has anything to back that Queen up with. Now, if you were holding something like the Q, J, 10, 5, 4 of hearts, a Queen lead can be a smart move. You can hope your opponent will play the King, your partner take both with the Ace, and now your J, 10 are both high. The bot however will routinely lead the Queen with nothing under it but low cards. So leading it only does one thing for the bot; it's just throwing the Queen away.

I do not want to imply that I think that the bots play well. They play horribly as a rule.

The worst things that they do are grab tricks as fast as possible on bagging hands and play Nils (on both ends) more than badly.

However, in defense of HW, I think that it is important to consider the following:

I cannot even imagine how one would go about programming Spades play.

Unlike every other game that I can think of, there is an almost constant approach/avoidance situation in Spades regarding taking tricks or letting them go to the opps. Everythng in Spades is situational, and there are endless possible situations.

How often when playing are we not even sure what we want to do on a given trick? Are we reading our pard's play correctly... is he setting or bagging.

My entire book is written around the risk/reward decisions associated with bidding and playing. Especially with bidding, that means very often not bidding one's hand. You will often see players either bragging about or complaining about others not bidding their hands, yet bidding one's hand is absolutely the very worst possible approach to Spading.

I do not know how one would ever program a bot to not bid its hand.

Oh, absolutely! That's mainly why my first comment was I realize they're AI. A programmer would go nuts trying to write if this, do that for every conceivable situation. You'd end up with another IBM Watson computer for just one game.

I really just wanted to noted the worst blunders the bot consistently makes.

Number one is kind of tough. How much of leading the Queen is a ruse? But really, with nothing to back it up, it simply shouldn't be lead. Your goal should be to hope to make it good by waiting for the Ace - King to fall.

Number two should be easy to code. Both teams have made their bid. Stop taking tricks!

Number three. Play your winning off suit cards fer' cryin' out loud! They're much more likely to be useless as the hand progresses. If it gets trumped early, then it is highly unlikely to have ever been good anyway.

Number four. That would be tough. It would have to consider things like, do I need the trick to make our own bid and to heck with the set? Is it worth it to set the nil bid? That is, it doesn't matter in the least because at the current score, you'll go out and the opponents would still be short. A pretty fair number of variables on that one.

Number five. That one should be somewhat easy. You need the tricks, stop letting your opponents trump. The bot will lead a heart and it gets trumped. That would tell anyone, don't lead another heart. Except the bot does just that when it gets the lead back. It will lead another heart into another trump. Draw the spades out first, then see if you can make the rest of your off suit cards good.

I would have to strongly disagree with point 3, especially if your team has Spades strength and a long or solid side suit.

In situations like that, it is far better to try to pull Spades and then run the good side suit cards assuming that the holder of them has an entry.

That would be a great strategy - if you could lead spades before they're broken. But since you of course can't, you have to try and get your high off suit cards in early. Once they're broken, then yes, run out spades then play your off suit winners. But you rarely get the chance to do that.

If Spades are not broken, then one leads something else besides his cards that may well be winners later in the hand.

I agree with one point the OP is making and that's that bots could be programmed better to attack/defend nils.

Doing this one thing would make the bots a much more reasonable stand-in if a player leaves and you can't find a sub!