Clearly this is a sequel to the huge thread below. I start this new thread for the viewers who have no knowledge of the little go to last post button, or lack the coordination to hit it.
The title is somewhat misleading, as i am indeed not easy to convince that playing two down with 6-4 is a bad choice. It is just meant to draw the crowd.

While my pc is still busy with subsequent rollouts, in the meantime I give my commentary on the finished rollout of worldclass vs casual player. I had set the STD to .01, which is silly, because the casual player makes a lot of random mistakes, and because of reasons mentioned below. Still, to please Taper_Mike, I let it do the rollout for an awful lot of times. On purpose i have not shown the equity difference, because I like to treat the cubeful and cubeless result separately. (Cubeful means the equity with the use of the cube, cubeless means the equity without the use of the cube. The cubeless equity can be calculated from the litte table below by adding the wins and subtracting the losses. The equity difference, which is shown in the window of the hint-button, is from the cubeful and not from the cubeless rollout). You can click on the bar to enlarge.


[img]http://s6.************/h2w4epbap/Screenshot_from_2013_07_31_16_45_04.png[/img]
imageshackus

A very strong reason to play 13/9 13/7 is that you don't want to hear: ¨O yeah, you are reading backgammon books. Well, I can do without.¨ Let's be honest, we don't like to win because we have more knowledge, but because we are more ingenious. Neither do we want to hear that it is all just luck. It is important to have the right strategy against the beginner, both for the sake of your own reputation and that of backgammon.

Openings could be divided into two types:
1. Static openings. Player has established a 'permanent' (nothing in BG is permanent, but you get the idea) asset, such as made new point or ran a checker with 65. Play after such opening tends to be easier for both players because newly made asset often dictates what both players should do.
2. Dynamic openings. No permanent assets established. Player split, or brought down, or both. Play tends to be more difficult because usually more game plans are available to chose from. There could be hitting going on back and forth, sending more checkers back and making game longer.

Which type are 64S (24/18 13/9) and 64D (13/7 13/9)? They are both dynamic, of course. 64S is already complicated enough? Are you sure 64D is even more complicated? I am not. Moreover, assuming replies are played correctly, I feel that if you are missed after this aggressive play, the game will become static more often after playing two down than after splitting.

When people willingly sacrificing equity for complications, they should be asking one more question: if game gets more complicated, then for whom it gets complicated more. I can answer this question for 64S. On average game gets more complicated for 2nd player who ends up with more checkers back than the 1st player. I don't have enough experience with 64D to claim that you will be facing more difficult decisions, but I certainly feel so. Also, note that your gain when missed is not as big as it might appear At 1st sight. Assuming you opponent plays correctly his replies the game plans are very likely to be priming games. Note that he usually misses with small numbers. This means he will be down in the race when this happens. It is more likely than after opening 51$ (13/8 6/5) to end up on the wrong side of timing battle. And what about those back games when he hits you couple times? O.K. He will be making more mistakes than usual, but consider how he might be playing a backgame himself if you split instead, and there is hitting back and forth going on.

I like 24/18 13/9 with an opening 6-4. Sometimes my opponent hits me on the 18-point, and then I hit him back, with a big lead in the race! My runner-up is 24/14, which is also pretty good for the race.

I love playing games where I'm ahead in the race. Just my 2c.

In any way, you get mostly situations which are different from S. You have more ammunition to attack his back checkers, which often leads to a hitting back and forth. Probabably the advantage is mainly against a weak opponent. Generally speaking, I have the feeling that timing issues become more complicated.

Many combinations lead to a homeboard point, with or without a blot. In case he played also two checkers down, you can make your point and leave the 7 point slotted. 31,51,63,33,55 play somewhat less, depending on the situation. You can always try to hit loose.

Suppose 30% of the 40% nonhitting games get more static. 62, 63,64,(65) make up 25%. That's a total of 3%, and you can get a lot of spectacle back.

You are the better player, and you will be more knowledgeable about this point. Again, let's leave the laboratory and get into the real world of backgammon. This is certainly related to skill gap. Often it is the opponent who has more checkers back. Perhaps 64D could also seduce the opponent to play more careless. I got so obsessed with the 64 thing that my mind was playing checkers allnight. I hope there are some open minds out here to do research.


By the way, 51 is the worst move of the whole bunch. I tried the first 12 missing plays up to six moves against an average player (intermediate level), and in 4 of them I have a priming game, and in just 1 the bot has one:
[img]http://s6.************/rtteb8kmp/Screenshot_from_2013_08_01_14_48_36.png[/img]
how to use print screen
[img]http://s6.************/g5zcgovht/Screenshot_from_2013_08_01_14_52_43.png[/img]
studio screenshot
[img]http://s6.************/vsqlu299t/Screenshot_from_2013_08_01_15_11_41.png[/img]
screen capture open source


I have done an extensive rollout of worldclass against intermediate level. Unfortunately my computer crashed after about 7000 untruncated rollouts, but i have been following the process and apart from the cubefull issue, there is not much equity difference. The same is true for wordclass vs advanced level. S has a small advantage over D in the laboratory, which could be fully cancelled out if placed in the field of humanities.

I usually play 8/2 6/2 because in New Zealand there are players who have not read anything since the 80's. If you make the 2 point they instantly think you do not know what you are doing and under-estimate you. This is great.

Against good, knowledgeable, modern payers I have started to play the split with 64. But I am very comfortable with making the 2 point. Of course at DMP 24/14 is correct

removed

8/2 6/2 narrows the priming option, so goes against the principle of noncommitment.

This is great.

Happen there to be any out of the box thinkers, because I start to feel like a don quichote.

The other part of the equation is if your opponent knows how to play the opening replies. After 8/2 6/2 some opponents will make incorrect replies, especially if they are not use to seeing that opening. This equity gain is worth giving up some noncommitment.
Even 21: 13/11 6/5 (my usual opening 21) goes against the principle of noncommitment but I would play that against anyone. I think Bill Robertie recommends it (in Modern BAckgammon) as well but notes it is more commital.

I have tried 13/9 13/7 a lot, and I am not convinced that the bot is able to evaluate complex positions like these. Is the bot justified in its assessment?
On the right is the negative equity move, and 3-1 followed after 3-2


[img]http://s6.************/t6tgw3bld/Screenshot_from_2013_08_02_14_13_33.png[/img]
online photo sharing

To get this thread again on top of that of Taper_Mike, I want to pose the obvious question, if XG is really capable of sizing up whether 13/7 13/9 is somewhat inferior to 24/18 13/9, when we consider 13/7 13/9's potential to provoke a 3pointbackgame in the early stage.

Well, since provoking a 3 point backgame is -EV, we shouldn't be surprised that XG rates 13/7 13/9 as inferior.

I want some authoritative unprejudiced confirmation that a 3 pointbackgame is negative EV. And at the moment I am not decisive about the authority of XG concerning this issue. If mr. Robertie declares that this is true, I will accept it.

Didn't he kind of say that in the thread "Extreme backgammon 3-point backgame", post #10?

Sent from my SGH-I747M using 2+2 Forums

We are talking about the 3pointbackgame now, and I am not sure the moderator referred specifically to this structure.

It is true that by slotting one doesn't necessarily create a threegame, as timing will often not allow this.

As far as limitations goes, I am no giant who plays against giants, and few of us are . I have played this strategy past week against the bot, and sometimes you meet real limitations (which could have a happy ending), and sometimes you meet real boons, and most importantly, very often a lot of excitement. But possibly I have been indoctrinating myself since I was challenged by Taper_Mike.

Here's a few observations on this approach.

A well-timed 3-point backgame is +EV.

When you try to slot a lot of points unnecessarily to get men sent back, you rarely reach a well-timed 3-point backgame. Usually you will get a poorly-timed game, or a lot of checkers stuck behind a prime, or a 2-point backgame which may or may not have timing, etc.

The games you get from playing this way will be very interesting. The results -- not so good.

This is not a new approach. Many players experimented with this approach in the 1970s and gave it up because the results weren't good. And that was against 1970s-era opposition, where few players knew how to defend well against a backgame. In today's tournaments, many players can defend well.

I happen to think that Snowie and XG play well against backgames. But even if they didn't, so what? When you sit down to play a money game or enter a tournament, you won't be playing Snowie or XG, you'll be playing humans. Some of them will defend backgames well, and some will defend poorly.

There's a whole book written about the style you like -- Backgammon for Blood, by Bruce Becker (1975). Check it out. The paperback edition has a great cover photo.

It makes me a little bit sad that the bot really has the last word and supports play that leads to less enervating situations in general. But this is not my last word in the matter.

First, I like to submit to you a hypothetical question, and being aware of the antique historical value of our basic setup, a reply is just optional. Suppose you could travel with a time machine to the 17th century to visit Edmund Hoyle. Would you have liked to persuade him to choose for a different setup?

Second, let me make clear that I am spoonfed and absorbed with your style, and that through your books I was able to give a good performance to the best players in my country within two years. Nonetheless, my zest for the inconventional made me venture the liberal approach. I can remember a couple of players who got really insecure by my free market play, while they were superior in the conventional approach. There was one player who even got fully tilted after I had won several games. Though I don't advocate this style as a standard play, this proved to me that in a certain context it can be really profitable.
It can also cause an aggressive player to double prematurely, and if you are the superior player it is definitely an advantage to have possession of the cube in a game with increased complexity.

Hi Yogiman,

the questions you ask are interesting and it depends (if we translate your first question to the future)
on the probability that a future computer might get so strong and see behind a certain treshhold,
that the current computers don't see.

For instance the sorts of play that Robertie makes against massive backgames (MBG) and which only recently are
supported by computer-rollouts.

Maybe there are such plays possible for the side playing the MBG, which might be invented/supported by future computers.

I'm sure Hoyle had his treshholds, so why don't we.

(p.s. maybe even Robertie has his treshholds)