| |
BGonline.org Forums
51S-21 – A Compendium of Previous Posts
Posted By: Taper_Mike In Response To: Opening Replies: slotting vs. splitting for 21S-21, 41S-21, 51S-21, and 52S-21 at DMP and GG (Igor)
Date: Sunday, 29 June 2014, at 9:30 p.m.
The following is a compendium of previous posts taken from my notes. While most of them do not directly address the DMP issue, there is much that that is relevant.
Mike
51S21
Posted By: David Rockwell <Rockwell.David@gmail.com>
Date: Wednesday, 7 October 2009, at 1:41 p.m.
http://www.bgonline.org/forums/webbbs_config.pl?noframes;read=50244
Many would find 51S-21$ slotting into the double shot surprising as shown on Stick's list. What may be even more surprising is that the most common response, 21S is only the fourth best move, a large .033 error. Both 21U (Up 24-21) and 21D (13-10) play better than the unmotivated split (24-23, 13-11) in this position. Those who won't play the slot here should be happy with U which is very close. 21U plays well when opponent has split the back checkers and there are only two attackers within range of the twenty-one point. It makes a good attempt at an advanced anchor and avoids multiple indirect shots in the outfield.I'm with Stick - it's $ for me.
51S21 - I'll do it. But why?
Posted By: David Rockwell <Rockwell.David@gmail.com>
Date: Wednesday, 7 October 2009, at 4:16 p.m.
http://www.bgonline.org/forums/webbbs_config.pl?noframes;read=50266
I will turn it around and ask, why is splitting the back checkers a big asset in this position?All should understand that after 51S-21$ responder is in bad shape. 21 is a bad roll. Getting hit on the five is NOT OK. It loses plenty of equity. The point of the play is that the split is worth less than it is with some other 2nd move choices. Getting hit by nines and tens in the outfield is very bad after splitting (24/23, 13/11). And, on the upside of course, when the slot is missed, you have a big equity gain.
The topic of splitting with an ace is worth a whole article or more. I'll point out quickly that:
It can be argued that doublets are particularly effective in attacking the split checkers in this position. I'm not yet convinced that this is significant enough to make this list. It may be.
- You are behind seven pips in the race [including the 4-pip, on-roll advantage that your opponent has]. Your primary game plan is containment; not racing, not hitting a shot later.
- Opener's forward position is NOT poised to make points quickly. Splitting is valuable, but not urgent.
- Opener does not have a blot in the outfield. After a move like 41S, that blot on the nine often remains a blot which is more likely to hit by split checkers next turn. There are no targets in the outfield after 51S.
In general, splitting with an ace does not play well on the second move when facing four on the eight point. A slotting response is the norm.
51S21 - I'll do it. But why?
Posted By: Rich Munitz <rmunitz@verizon.net>
Date: Wednesday, 7 October 2009, at 4:47 p.m.
http://www.bgonline.org/forums/webbbs_config.pl?noframes;read=50279
Thanks David. This is just so immediately counterintuitive. I am not the type to ever just blindy accept something without trying to understand the underlying reasons. Sometimes those reasons are conceptual. Other times they are purely technical. In this case, it looks like a mix.Looking a little deeper, while duplication is often overrated, there are quite a few hits after the slot that are otherwise very productive numbers anyway:
11,13,22,23,24,33,34,35,44
46 is a very good roll against the alternative 21 plays S or U.
The hit rolls that would really do nothing otherwise are:
12,14,36,45
Looks like for the most part the slot really does make good rolls out of rolls that are already good rolls.
51S21 - I'll do it. But why?
Posted By: David Rockwell <Rockwell.David@gmail.com>
Date: Wednesday, 7 October 2009, at 8:54 p.m.
http://www.bgonline.org/forums/webbbs_config.pl?noframes;read=50326
Timothy Chow wrote:Neither David nor Matt has answered Rich's question: Why?
A first step for such difficult questions might be to do a semi-manual 1-ply or 2-ply analysis. Play out all the possibilities and see what equities the bots assign the resulting positions. I might do this for 51S21 at some point, although I have other priorities right now.
Here's a summary answer to the why question, taken from the discussion above.
Splitting isn't urgent. Slotting fits the game plan better. Many hitting rolls are duplicated.
This is "why". I performed the analysis you suggested some time ago and these are my findings. It it makes it to the top of your priority list, I encourage you to perform the analysis as well. It's very helpful.
52D-21
Posted By: David Rockwell <Rockwell.David@gmail.com>
Date: Saturday, 17 October 2009, at 3:53 p.m.
http://www.bgonline.org/forums/webbbs_config.pl?noframes;read=51309
If you split after 52D, you are making about as big an error as you can make with a rational looking 2nd-roll move. In general, slot the ace, do not split on the second move when facing four checkers on the eight point. This is often explained by the attacking potential the opener has against the split back checkers. I am not yet convinced this is the primary reason for the sharp slotting preference, although it is certainly a factor. I am in the midst of deep study trying to understand this concept better. (i.e., I have been spinning my wheels.) In any event, the rule is clear even if the logic is not - Slot facing four on the eight point.Note that this concept is relevant to the previous position, the counterintuitive 51S-21.
21U
Posted By: David Rockwell <Rockwell.David@gmail.com>
Date: Saturday, 26 December 2009, at 11:59 p.m.
http://www.bgonline.org/forums/webbbs_config.pl?noframes;read=56521
After 51S (Split 24/23, 13/8), playing 21S (24/23, 13/11) allows the checker on the 11 point to be hit with 63 and 54 in addition to 64. This is a huge difference between splitting here and splitting on the opening roll. Remember this fundamental. The value of developing the midpoint by leaving a blot in the outfield is related, in part, to how often it is hit. Nack and Paul's book discusses this concept frequently.In the opening position, 13/11 generally plays very well because it is only hit with a 64. Even though the blot on the 11pt doesn’t add much to point making potential, it is often a strong move because of this. Look at the difference between an opening 32S (24/21, 13/11) and 32Z (reverse split 24/22, 13/10). The latter play is reasonable, but well known to be inferior. One of the reasons for this is that it is that it is hit in the outfield with four rolls rather than two. Make a habit of looking for these outfield hits. It may influence your choice of plays. For example, after the 5pt is made, a blot on the 10pt is no longer hit by 54 from the ace. This improves the value of the move.
21U
Posted By: Nack Ballard <nack2000@sbcglobal.net>
Date: Sunday, 27 December 2009, at 1:09 a.m.
http://www.bgonline.org/forums/webbbs_config.pl?noframes;read=56529
51S-21 produces a different correct move from that of opening 21 because the opponent can hit from the 23pt with 63 and 54. Instead of being .048 behind, U rolls out ahead of S by .020, .034 or .023, depending on the bot, an average of .026.For 21S-21, U's .026 edge over S is erased and the two plays are tied because opp's hitting 63 is duplicated — it could have been used constructively to make the 5pt.
For 41S-21, S pulls back into the lead: it is about .016 better than U because opp’s hitting rolls from the 23pt, 63 and 54, are both duplicated.
Coming full circle, with opening 21, S increases its lead over U from .016 to .048 because Opp can’t hit with 63 or 54 as she can (and happily would, despite the point-making option) with 41S-21.
Hope that helps,
Nack
51S-21U better as 51S-21$
Posted By: David Rockwell <Rockwell.David@gmail.com>
Date: Sunday, 27 December 2009, at 2:57 p.m.
http://www.bgonline.org/forums/webbbs_config.pl?noframes;read=56555
Andreas wrote:According to Stick’s RO page, Snowie sees U better by a hair, and GnuBG, $ better by a hair. Are there reasons to think U is best?
Maybe. Snowie prefers 21U over $ as you say. GNU does not do a good job of evaluating certain slotting plays. I don't know whether or not this is one of those cases. The handful of times that I have dug into Snowie vs GNU preferences, Snowie has looked better. But, I have never been able to conclusively demonstrate why.
Also, note that U is preferred by GNU in its cubeless measure. This has been another red flag that the GNU live cube rollout may be problematic. A good rollout will find cases where one move is preferred with a cube and another without. GNU varies the relationship between cubless and live cube. Snowie tends to keep the ordering of moves between cubeless and live cube. And, the relationship between cubeless and live cube equities tends to be very steady. I would expect variance between the two. However, my research has left doubts that GNU is doing this well. So, the jury is out.
I play $ even though I have some doubts about the rollout.
41S-21 and 51S-21: merits of $ vs S
Posted By: Nack Ballard <nack2000@sbcglobal.net>
Date: Friday, 1 January 2010, at 3:39 a.m.
http://www.bgonline.org/forums/webbbs_config.pl?noframes;read=57066
Neil Kazaross wrote:I would think slotting would do OK here after 41S since it does OK after 51S noting that slotting is duping some 3's and 4's after 41S. Apparently, it just doesn't gain much to split when the opp has played his early 5 from 13/8 adding more fire power but not creating a blot.
It is natural that you (and others) are wondering why $ loses to S (by about .020) for 41S-21, but $ beats S (by about .020) for 51S-21. That's quite a swing.
I know your comments are only offhand but the underlined your key phrase above demonstrates a somewhat common misconception. I thought many might benefit by the following explanation. The four after-positions are diagrammed at the end of this post, for reference.
The only roll with a 4 that is duped better with 41S-21$ (than with 51S-21$) is 41. In either non-slot position, 64 54 would hit on the 14pt, 43 would make the 20pt (rather than the 5pt), and 42 would make the 4pt.
The 44 roll argues the opposite: for 51$-21$ (more than for 41$-21$). With 51$-21S, you get pointed on while Opp unloads two stacks anyway. Conversely, with 41$-21$, 44 becomes a highly efficient roll for Opp, hitting and using her 9pt builder to make her own 5pt; 41S-21S manages a lot better against this roll.
Granted, 41 is a bigger swing (than 44), but it’s only one roll (i.e., we’re not talking “4s” in the plural) and the effect of 44 offsets about 30% of it (yes, I cut the swing in half to account for doublets being half as likely to arise).
As for 3 duplication, 3s are duped better with 51S-21$ than with 41$-21$ because 3s unstack. Double 3s in particular is better duplicated, but also 53 31 are duped better. Slotting deflects Opp from unstacking her fourth 8pt checker with these rolls.
Let's count 21 as a 3. That is the only roll other than 41 that definitely argues for 41S-21$ more than 51S-21$, because it would otherwise make Opp’s 7pt. Other rolls that use the 9pt builder to make a point (62 52) aren't duplicated/deflected -- they can't hit the slotted 5pt.
Based on my 36-ing, the 41 21 duplication that argues for 41S-21$ is more than offset by the 44 33 53 31 duplication that argues for 51S-21$.
Another factor is that when you leave indirect shots (in this case 64 63 54) and are hit, you have more play against a 9pt blot rather than an 8pt spare; the efficiency of splitting (rather than slotting) is increased because the 23pt checker might roll a 7 (or inch up and then hit in a scramble).
Other rolls, too, speak favorably to 41S-21S compared to 51S-21S: Consider Opp's 61 11 65. You gain more by having split because you have immediate 7s as well as 8s to hit the 16pt blot, and if your 5pt is slotted you're conflicted about hitting on the 16pt anyway (except with 44). That aspect doesn’t exist for 51S-21S-xx (where xx is 61 11 65 and others) because there is no 16pt blot to hit.
You were on the right track when you touched on the value of splitting against a 9pt blot, and some of my analysis addresses the manifestation of that value. It is the issue of duplication that is often radically different than it appears on the surface.
Hope that helps.
Nack
After ~13,000
Posted By: neilkaz <neilkaz@earthlink.net>
Date: Friday, 1 January 2010, at 5:55 p.m.
http://www.bgonline.org/forums/webbbs_config.pl?noframes;read=57123
As I expected, and as was also shown by someone's DMP RO, $ wins a bit more games, but the added G and BG losses seem to be making it slightly incorrect from a technical standpoint. We can also clearly see that slotting leads to a longer game (time shown) and one would expect more complications vs weaker players.Unequal skill RO is very interesting
Posted By: neilkaz <neilkaz@earthlink.net>
Date: Saturday, 2 January 2010, at 4:36 p.m.
http://www.bgonline.org/forums/webbbs_config.pl?noframes;read=57202
What is implied here, is what I thought and that is to complicate vs. weaker opps. We can't be sure what mistakes 0-ply is making here that 2-ply doesn't, but we are certainly sure that 2-ply plays better than 0-ply so we can expect that what is shown here is a decent approximation of real life.With unequal skill, slotting sure seems to win more games, which is some of the reason it does lots better cubefully. I also suspect the 0-ply is making more cube errors from here than 2-ply....hmmm wouldn't we expect that in real life if one player were about 100 elo stronger than his opp?
51S 21?
Posted By: David Rockwell <Rockwell.David@gmail.com>
Date: Monday, 17 June 2013, at 12:24 a.m.
http://www.bgonline.org/forums/webbbs_config.pl?noframes;read=143211
Understanding the slot is not trivial. However, the difference betweem moving 23/21 or 13/11 with the deuce isn't tough. The opener hits the blot on the eleven more frequently than we would like with the back checkers split. And, the blot on the 21 point is only subject to a small number of point on head shots since the opener failed to add a builder to a new point. After 41S for example, moving to the 21 point with 21 is not as desirable because of the extra attacking numbers.So, 51S-21U and 41S-21S
Knowing and understanding this - evaluate how to play 21S-21; U or S?
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.