Bots and Backgames

A correspondent offers as an opinion that the positions I showed in the last post should not even be doubles. The reason that he offers for this is, "That bots do not understand backgames". This is an opinion that is held by my many people, some of whom should be listened to because they are respected thinkers, some of whom are merely parrotting what others have said.
Let's sort out some nomenclature. White in the last post is "playing a backgame", i.e. holding two points in the opponent's home board with the hope of generating shots later, hitting one or more checkers and winning the game in that way. Red is "defending against a backgame", i.e. trying to bear in and bear off without leaving any shots, or surviving them or indeed winning after being hit.
Do the "Bots can't play Backgames" set think that the bots play them poorly, or defend against them poorly, or both? I'm not clear on that. Do they think that they play the checkers badly or do they think that they are very badly wrong in their assessment of the possible outcomes of the game to decide upon cube action? Somebody enlighten me here and I will be very pleased to hear opinions, ideally backed up with some data.
Some people I know produce positions, usually composed, that the bot will handle in a very odd way, I accept that, but I don't think that that is evidence that the majority of games are played badly. I also know positions where the bot can be well off the money in non backgame positions.
I will be very happy to play the first position from the last post as a prop, either double against no double or I will pay White a point to take the cube. Props can be set up and played on Gamesgrid. Any takers?
Alternately I will be very happy to take snowie playing either side against a human, each player to take the same side for, say, 50 games. Any takers? A sporting 5 euros a point shouldn't kill anybody and might settle the discussion in this case at least.

Not What, But Why

Like many other keen players, I run all my matches through a bot for analysis while the match is fresh in my mind. I don't look at every play, or even every error, but the blunders (mistakes that cost more than 0.11) do get looked at. Once I've checked them and satisfied myself that the bot is likely to be correct, it is vital to try and understand the position and how I could have avoided the mistake. Take a look at this next one and you'll see what I mean.



Here I am playing the Red checkers and leading 5-4 in a match to 9. It's a fibsleague match. I think about doubling, but don't. In fact this is already a pass according to Snowie and Gnu and failing to turn the cube is a major blunder. How did this come about? It isn't at all clear that White should take, so by Woolsey's Rule alone I should obviously double. Moreover, there is a very handy rule of thumb for deciding when to double backgames. When you are threatening to clear the third last point in front of a well timed backgame, you are probably in the doubling window. With hindsight I know why I didn't cube here and it's an old failing; I tighten up with the cube when leading in matches that I am very keen to win, particularly when I am playing somebody that I think is not playing well. I drift along, hoping that the match will somehow win itself and tend to be pessimistic in exactly this sort of position. What I should do here is ship the cube in and put the pressure on White, because clearly in this position she would prefer not to have to make a cube decision. It's more or less what I call a 60/40 position, 60 wins including 40 gammons.
Some key points to note: I have lots of spares at the moment, so I can handle all the non-clearing rolls, except 6-1 of course. These spares are very important and when I begin to burn these, my position will get weaker. On White's side she has the nice 6-5-4 block in place, but her other checkers are awkwardly placed, although of course her frontal position should improve. it's important to recognise this "I'll get weaker, while she'll get stronger" scenario, because it happens a lot.
I rolled 6-2 and White rolled 4-1, so we came to this position next.


I have got slightly weaker because I have lost a couple of pips of timing and White has improved a bit because her distribution is better and she has slotted an important point. Now Snowie thinks that this is still a very strong double and White has a tough take. Gnu agrees. Again I didn't double. At least here I was consistent, as I did recognise that I had got a bit weaker, but again the blunder happens because I am reluctant to escalate the cube, a weak-kneed attitude. Backgammon is a game of aggression, be aggressive.
Next I rolled 5-1 and burned my last two spares from the 6pt, followed by White rolling 6-1 and playing 12/11, 9/3. This is a mistake for White. Slotting the next point in the prime is usually right, but here it's one slot too many. White already has the vital bar slotted and it is more important than the 3pt here. In addition, five blots is a lot of weight for any position to carry! The best play looks like the quieter 12/6, 10/9. This is much more balanced and the 9pt is a point in the prime field, six pips away from the 3pt escape hatch. The real weakness of the 3pt slot shows its head on my next roll, 6-2.


>

With all those blots strung out I should play 8-6 and take my chances now, as White will never be so weak again. Sadly this world class play passed me by and I played the feeble 5-3. If White had made the best play last turn then that would have been correct and 8-6 would have been a huge blunder. Clearly the five blots should have been marked with a "More Is Possible" flag (see earlier post!) and the prospect of 7 blots after, say, 5-1 (23/17*) for White should have any predator licking his lips. Here I was more sheep than wolf, but it's a wake up call. If you play me next week I might have sharper teeth! Until the next time, enjoy the game!

More Blitz Stuff

In my last post I concentrated on checker play in the face of a blitz, but I also remarked on the correct cube action and said "Note that White should double any play and Red should take, even if he has made the weaker move". Here's the position again. It's 0-0 to three.




It's fairly easy to see that this is a strong double but a lot of players would pass this, so what makes it a take? Assessing game winning chances and gammon losses in blitz positions is very very hard. There is no magic formula that combines all the features of the position in a way that can be converted into maths, so in the end it comes down to instinct. However, instinct can be informed, so what do we need to look at here? From Red's point of view, he won't be on the bar after the play and although he has yet to make a point, all his checkers are in play and reasonably well placed to make points soon. He is only 5 pips behind in the race after the roll. Looking at White's position, we can see that 6-5-4 block of course and the strength of that can't be overstated. It's immense, but looking at the rest of White's checkers it is actually all that she has got! As noted in the last post, she has the stack on the mid and a slotblot in the outfield and watch out for this, it's a very weak formation. You usually can't help getting into it, but once you have it needs to be repaired and soon. Also, crucially perhaps, White has yet to move her back men. If her back men were split, or if her outfield checkers were less fragile, then the take would be very tight and it would probably be a pass for Red if he made any but the best play from the bar.
So, that would be my instinct and Snowie analysis backs it up. Not a pleasant take, but correct, even at 3-away, 3-away, where we need to be more conservative than usual with our takes.
Doubling blitz positions is a lot of fun for the doubler of course. They are very volatile and in one sequence the defender's game can collapse into a gammonish heap. The play is usually a matter of bashing anything you can see and going for a closeout, with the difficult play only coming after you have had one or two checkers hit. One thing I have noticed though, is that several successful blitzes leads to over confidence and the cube in subsequent games gets earlier and earlier! Here's a good example of one.



Red (me) is on roll and the score is 1-1 in a five point match. I doubled this, what would you have done? White incidentally is a Fibs GammonBot, so Red isn't entitled to add any vig for the chances of an incorrect pass. Mind you, a pass would be fairly horrid here, as the double is a blunder. Let's take a closer look to see why.
Apart from the White checker on the bar, her position is actually a bit better than Red's, with the 11pt made and only two checkers back. The race is even. So, Red is doubling on threats and against a human opponent that might be reasonable. We all get a bit nervous off roll with a checker on the roof and I have seen positions like this passed before now. The trouble is that Red doesn't have checkers in position to be very threatening. What this position lacks is volatility. Red has quite a strong game that should develop nicely over the next few rolls, but he is unlikely to make the sort of great leap forward that will lose his market on the next sequence.
After I doubled and White took, I then rolled 2-2. Having fundamentally failed to understand the position, it's no surprise to see that I then blundered with this excellent roll. Take a moment to choose your play in the position below. The score is 1-1 to five.



I chose 13/11, 8/6, 4/2*(2). This wasn't a careless play, I thought, "Put a second checker on the roof and bring down another builder", which seemed reasonable at the time. However, switching points isn't really viable, because I just don't have the ammo for the blitz. Granted I have brought another checker into the zone, but at the considerable cost of giving up my 4pt. What I really need to do here is import a lot of extra ammo with 13/11(3), 8/6. After that excellent play (and kudos if that is what you selected) then I suddenly have 12 checkers ready to blitz or prime as the dice decide. This is so much more flexible than my play, which comits heavily to the blitz plan. Midpoint fans will hate to see it go, but owning your opponent's bar and your midpoint concentrates too many checkers in one segment to be flexible. One or the other has to go and here it is the midpoint. This is quite typical. Look out for this combination and get ready to dismantle it; the two points don't combine well to cause trouble for the opponent and the checkers on the mid are usually more use somewhere else.
The last thing that needs to be said about this disastrous sequence is that you need to be particularly careful not to give away the cube to a savvy opponent at this score (4-away, 4-away). Your take point on a recube is about 32% cubeless, so any time White gets to be a favourite here a recube is in sight! Didn't know that? Then that alone is enough to make reading this blog worthwhile, so until the next time, enjoy the game!

When Safety Isn't Safe

The score is 0-0 in a three point match, Red to play 3-1.

Once your opponent has made her 6, 5 and 4 points, you can expect a powerful and gammonish cube pretty soon, particularly when, as here, she has put you on the bar in the process. This is particularly true at the start of a three point match, where the initial cube is more powerful than usual and you often have to pass cubes that you could take for money. So, I rolled 3-1. What would you do with this?

I played bar/24, 13/10, a gut instinct play that responds to blitz fear, i.e. I'm in trouble here, let's anchor and try to ride it out. Unhappily, this is a very weak play, a large blunder. Easily the best play here is bar/22, 24/23 and if you played that (and would have played it over the board) well done, great play! Can we see why this is best?
White has achieved this powerful little block using only checkers from the 8 and 6 points. This is of course very efficient, but it leaves her short of ammunition to continue the attack, so stepping up to the best available points is not as dangerous as usual. However, no need to come under the gun, even such a small gun as this, unless there is a point to it, so why should we make this aggressive and risky play, rather than just anchoring? It's because White has left herself with five checkers on the midpoint and one on the 8pt, a notoriously weak formation that is difficult to develop. If Red plays bar/24, 13/10, then White can either hit one of the outfield blots or bring checkers down from the midpoint to strengthen her forward position. Playing bar/22, 24/23 reduces her chances of hitting in Red's outfield and increases the coverage of White's outfield, forcing contact and action now rather than trying to postpone it until later.
Don't like this? Set up the pieces and play it a few times. The difference between the two plays is so large that you should see better results from the aggressive play very quickly. Note that White should double any play and Red should take, even if he has made the weaker move. Before computer programs arrived to take the strain out of rollouts, we used to do them by hand, ideally with a friend to ease the labour and provide idea input. Our rollouts were not very accurate in terms of results, but we did learn how to play the position, something that we lose when we use the bot to do the work. Give it a try and until the next time, enjoy the game!

Blundalongadorbel

I'm on a bit of a losing streak at the moment and at these times, it's always good to keep a close eye on the blunders, the plays where one gives away a lot of equity. If you start to see where the big mistakes are and more importantly, why you made them, you can begin to put right the faulty thinking that led to the mistake in the first place. Here's one where I could have done better with a bit more attention to detail.



I lead 2-0 in a match to five, I trail 148-170 in the race and I am on the bar. The roll is 6-5. of course I enter with the 5 and look round for a 6. It hits on the 7pt and without much more thought, that's where it went. This is a fair sized blunder.
Hitting is usually correct when you trail in the race, so why is it wrong here?
Hitting has several drawbacks. The first is that it doesn't achieve all that much. It only costs White 7 pips in the race and Red doesn't have either a board or a prime to contain the blot when it re-enters from the bar. When the play "works", i.e. when White re-enters and doesn't hit one of the three blots lying about, it still doesn't gain much because the bar point isn't particularly valuable here.
The second big drawback is that it means giving up Red's midpoint. With both players holding a high anchor after this roll, the mostly likely game type is a mutual holding game, where both sides struggle for positional supremacy and try to build points. The midpoint, controlling the outfield and providing a useful landing point for checkers moving from back to front, is vitally important.
It's so important that White on her turn will be very reluctant to give it up in order to hit, so playing 23/17 is relatively safe and that is Red's best play. 8/2 is as bad as hitting.
Of course 23/17 doesn't achieve much either, but it makes a position with only one blot which is nice and keeps the vital midpoint. These quiet plays are hard to find, as we are conditioned to "do" something, to make things happen, to hit and be aggressive, but sometimes quiet is good and a bit more thought here might well have led me to a better play. It's bad to make mistakes of course, but it is worse to make them when you just haven't thought for long enough!
Until next time, enjoy the game!

Even More Is Possible Sometimes

The score in the position above is Red 1 White 6 to 7 and it is of course post-Crawford. What this means in practice is that the cube is dead and the gammon threat is one way. Winning a gammon here is hugely important for Red, effectively getting him to DMP and losing one is irrelevant.
As in the previous position the More Is Possible flag is up and the score is shining a spotlight on it, not to stretch the metaphor too far! When this play came up I made the "obvious" play of 20/9, as I expect many would have done. Snowie analysis preferred 20/14, 20/15, which was startling to say the least! What, give up the golden anchor and leave three blots? It's pretty obvious that White isn't going to mount a blitz if she can help it, so the blots are fairly safe for the moment, but where's the gain? It works like this.
Positions where you hold an anchor in your opponent's inner board and a point in her outfield (more usually her bar point) suffer from two major flaws. The first is that they don't produce anything like as many shots as you might think and the second is that it is very rare to have the timing to keep both points for as long as you need. When you have to run from one of them, the opponent can often hit her way home. Here the race is relatively close, 112-123 after the roll, so Red definitely hasn't got the timing to keep them both. It makes sense to break one now, while White can't take the risk of hitting and the 20pt is the one to break! Why? Because if you keep the 20pt, then it won't pose any great threat to the White checkers on the 8,7 & 6pts and White can run the lesser risk of leaving indirect shots on her 13 & 14 pts. If you keep the 16pt, then White will face the herculean task of safely clearing her four rear checkers against an anchor perfectly placed 6 pips away.
It's rare to see a play that can be described as brilliant, but any player who made this move over the board is probably world class. When I first rolled this out some years ago, I did the rollout with Snowie 3 and it made 20/14, 20/15 clear best, with 20/9 a blunder. However Snowie is not ideal for this as it doesn't roll out keeping the score in mind and a recent Gnu rollout shows that the two plays are in fact fairly close, although 20/14, 20/15 is still best. Anybody like to run this through ExtremeGammon? Incidentally, yes I did think of 16/5 and it isn't a large error, but it can't quite match the other plays, as it gives up what is clearly the best anchor.
I hope that this has demonstrated the usefulness of the MIP flag. These oportunities to find an exciting and effective play come up fairly frequently, so let me know if you come across one IRL. Until the next time, enjoy the game!

More Is Possible.

Before you read on, take a moment to think about this position. It's a seven point match and White leads 2-0, with Red holding the cube. How would you play Red's roll?

If you played 18/11, you are in good company, as that is what I played, but it's a blunder. Because of those two blots in White's home board, Red can play to get a double shot after White's next 6 and try 18/15, 5/1, or 18/15, 7/3. This works particularly well when White throws 6-3 and has to expose four blots! It's quite a rare position, I made this mistake years ago and haven't seen anything quite like it in the interim. It's rare for Red to still have the luxury of a spare on the anchor. However, similar positions do occur fairly regularly and when the side bearing in has butchered his board and has a couple of blots hanging about, I like to raise a flag over it, to remind myself "More Is Possible", to prevent myself from making the hasty, "obvious" and safe play. Note that the two blots don't always make it right to expose a blot, but you do need to give it some serious thought.
I hope to be posting more regularly now, so until the next one, enjoy the game!