Forum for Basic endgame theory

Path: Endgame =>

review of Herman's first rewrite [#2240]

Back to forum     Back to page

New reply

reply review of Herman's first rewrite (2010-03-26 09:39) [#7554]

First of all: great start!

Some suggestions:

  • it is confusing to use the sente/gote arguments before explaining them. Better first show that there is something to the order of moves and then conceptualize.
  • at times the explanation is a little verbose to my taste. If you wish I can polish it to my taste, whence you have all liberty to revert.
  • I think I would like to emphasize more that average gain and swing counting (great terminology) are different but almost equivalent ways to look at endgame plays and their evaluation.
  • your final demonstration of whether a move is sente is convincing but hardly practical, because there is easy access to the complete information about the game. The concept of sente is precisely introduced for practical reasons, because there is no easy access to the complete information. So I like your explanation higher up better, where you confess that even strong players have to best guess, bluff or know from experience a move will be sente.


HermanHiddema: Re: review of Herman's first rewrite (2010-03-26 16:10) [#7560]
  • I've reordered it to explain strategic sente/gote first, then introduce the concept of endgame sente/gote
  • Please feel free to polish as you see fit! It is hard to find a good style for this article, needing to balance between beginner style (glossing over complications, taking the reader by the hand) and more formal style (use terms exactly, show calculations).
  • I don't know how you would like to do this, exactly. Please feel free to edit to make the relation between these two concepts more obvious
  • I've removed it, it was not very well thought out. I think the current treatment of sente/gote is better (but still needs work).
Unkx80: ((no subject)) (2010-03-26 10:37) [#7555]

Good work, Herman!

Some meta-comments:

  • Much of go is the struggle of expanding your influence and territory while reducing the opponent's. As a result, there's a lot of emphasis on moves like press, push, block, reduction, etc. that occur at the boundaries between yours and opponent's spheres of influence or territory. More generally, I feel that the beginner study section isn't emphasizing this concept enough. It is easy to forget about this major concept after wading through too much life and death. Perhaps it might be useful to remind the beginner about this again in the endgame pages.
  • I agree with Dieter in that it is better to roughly explain sente and got before moving into evaluating the values of moves. In relation to my previous point, beginners may need to be told why they should block in response to a hane, and what are the consequences of not blocking. Ultimately, such explanations can help the presentation of the concepts of sente and gote, as well as the value of moves.
HermanHiddema: Re: ((no subject)) (2010-03-26 16:13) [#7561]
  • I agree. The subject of boundary plays (yose) needs to be incorporated. I will try to write something on it, but right now my head is swimming with all the editing, and I'm in no state of mind for it. I need some rest, probably a good night's sleep, before I can tackle it :-)
  • I've improved on this concept significantly, I think, please tell me what you think of the current treatment of sente and what should be added/removed
tapir: ((no subject)) (2010-03-26 10:55) [#7556]

I am very grateful that Herman started editing this page. I am not sure, whether it is basic endgame theory or a short review of basic and advanced endgame theory right now. Miai counting and tedomari are probably not very basic - and I fear one can have too much information in an introductory page. Also, I would start according to importance with sente / gote first, mutual damage next and deiri counting last. As far as I remember the problem in my games was not to count how big the swing value of all the moves was, but not to lose sente and starting to ignore "sente" moves for bigger sente moves by myself. Maybe mutual damage should even be the first item (by importance it very well is).

That is, what actually is sente? I admit I am not sure at my level as well. Is it useful to talk of sente in a local context at all - when on the whole board tenuki may be the best answer? Related to this there should be some mentioning of size of follow ups (/ threat size) as this is more important for sente/gote and mutual damage purposes than the size of the "sente" play itself. It may even be misleading to tell beginners this move is sente (easily introducing a reified concept of sente) without context - he may answer blindly in future "because it is sente" without thinking about the size of the follow up moves and possibilities for mutual damage at all.

I am aware that my proposal may look a bit impractical and probably doesn't fit well with Dieter's idea - as it forces to start the presentation with not yet introduced concepts, but if it is possible to demonstrate with diagrams (think of Kageyama's treatment of endgame) this would be worth it.

Cheers Tapir

PS Follow up, sente/gote, mutual damage, size - leaving advanced concepts and techniques for further reading.

PPS I see that you write about the value of the follow up, but isn't it somehow lost. I mean what is sente and gote depends on it :)

X Re: ((no subject)) (2010-03-26 15:01) [#7559]

Hi Tapir

I think Herman's article does a very good job in describing what sente and gote are in an endgame context. On one hand, you need the concept to make the calculations (with swing counting). On the other hand, they are not absolute. The question "what is sente" will occur to any improving player when studying the endgame. The best description possible, I think is, "a move is sente when its local follow-up, when unanswered, represents a clear gain over any other move on the board". (*)

In a situation with easy access to the complete information, say moves A, B, C remain to be played, it is easy to determine the order without conceptualizing it. The concept is useful when the moves remaining are M1, M2, M3, ..., Mn where n depends on the path chosen and Mn is unknown.

The question "is M1 sente" is answered by seeing that, if Black M1, White M2, Black M1' (the follow-up) makes it very unlikely for White to win.

The degree of a move being sente is proportional to the likelihood that the game is won if the move is not answered locally. If the players manage the complete information about the remainder of the game, there is no question about a move being sente. There is actually no question about the game result.

I cannot fathom a meaning of "sente" that is disconnected from the probability of the game result.

(*) The odd thing here is that a move being sente implies that it is best for the opponent to answer it. Playing a clear sente move thus forces the opponent to play something that keeps his winning probability at least equal. So it is probably better not to play sente moves, unless victory is clear and the sente move simplifies the game. I think that is meant by sente gains nothing. But then again, this may be a flawed argument: playing the sente move must be set off not against "doing nothing" but against the opponent playing the reverse sente there.

tapir: Re: ((no subject)) (2010-03-26 16:21) [#7562]

Hi you,

Probably the distinction of sente position and sente move is what I aim for. I found this particularly troubling in my unfinished learning process. Sente positions will often allow to play a sente move there "at the right moment (when the follow up is biggest compared to the rest of the board) - however more often than not with correct play the opponent will play tenuki. And that should be hammered in as early as possible! Play tenuki if you have a move in another sente position with a bigger follow-up. Correct endgame play has in my opinion no relation to winning probabilities, that is why MC engines usually play a bad endgame (losing all but the last point when ahead, being unreasonable when behind).

Beginners often lose games by chance according to who starts endgame first, taking all the "sente" points in row and answering sheepishly. I really really believe that omitting the impression ("the first line hane is sente, i have to answer") is a big step forward in endgame presentation to beginners and a small step towards overall better performance :) And maybe one should even rewrite the beginners heuristics to that end. Not the sente with the biggest swing value first, but that with the biggest follow up. (Ok, it may be difficult to actually count this for most, but even trying makes you stronger.)

Cheers Tapir

HermanHiddema: Re: ((no subject)) (2010-03-26 16:16) [#7563]

I have decided not to go into the concept of tedomari indeed, it is too complex for an article title "Basic Endgame Theory". I've also reordered the contents to focus on the basic concepts of sente, gote and mutual damage first, and going into details of counting only later.

I am more happy with my current treatment of sente, although I think it can still be improved significantly, please tell me what you think and whether you have suggestions for improvement.

HermanHiddema: Thanks all for the feedback (2010-03-26 11:03) [#7557]

Thanks for the feedback everyone, I will try to take it in to effect as I rewrite further today! Once I'm done with that, I'll reply to each of you in more detail, but right now I'm still not entirely sure myself what the final order and structure of the page will be :-)

emeraldemon: Re: Thanks all for the feedback (2010-03-26 17:43) [#7567]

The basic endgame theory page seems deliberately to avoid CGT, which is probably just as well, but it does provide a simple and clear definition of sente: a sente move is one where the position after the play is hotter than the position before the play. This means at the time you make the play, the followup will (usually) be the biggest move on the board, and your opponent will respond. Of course it's not so easy to know the temperature of moves, but that is in principle what sente is all about. It might be possible to explain sente this way without using terms like temperature, I'm not sure.


Unkx80: Re: Thanks all for the feedback (2010-03-26 19:04) [#7570]

It is quite possible to explain sente without using terms like "temperature". Just say something like, you make a move, and there is a followup worth 10 points. If there is nowhere else with a value of 10 points, then the opponent's best move is to answer the move. Thus, in this case, this move is sente. Assessing the value of moves is another story, but the above description should be enough to describe the main idea of sente.

Bill: Re: Thanks all for the feedback (2010-03-27 08:03) [#7584]

Indeed, it is possible to talk about sente without using the term, "temperature". However, during the '90s "temperature" became quite popular on, with a meaning somewhat different from the original. I think that it has entered the English go lexicon with the popular meaning. :)

Unkx80: Re: Thanks all for the feedback (2010-03-27 09:39) [#7585]

I see. Thanks.

tapir: Re: Thanks all for the feedback (2010-03-26 22:15) [#7574]

This is not exactly true. Sente moves have follow-ups bigger than overall temperature, however if your opponent finds a move elsewhere that makes that area burn even hotter you will follow around to answer and only later get your sente move answered. In a way it was a true sente (it got answered), but in a way it was executed badly because the opponent did not need to answer instantly and got another sente in between. This is sometimes unavoidable, but sometimes when involving double sente moves game deciding.

Bill: ((no subject)) (2010-03-26 17:40) [#7566]

I am very glad that this page is being rewritten. :) The original was a mish-mash from the start. I have nitpicked a little, but this is a great improvement. :)

My feeling, though, is that it should be aimed as much as possible to DDKs. Keep it introductory. I would not even get technical about sente and gote. If the opponent answers a move, it's sente. Sente is worth taking, and keeping. (That's two different senses of sente, but neither is technical. :)) Mention that there are technical meanings, and give the links. That's good enough, I think. Say that the biggest move is usually best, but that there are exceptions, then link to tedomari. Talk about miai. Mention that the usual rule is sente, then gote, then reverse sente. Say that in estimating territory you assume that sente plays have been taken.

I would not -- repeat not -- say anything about sente being worth twice gote. That is technical, related to deiri counting only, and not, as is sometimes claimed, an approximation. If you mention it at all, say that it is how you compare sente and gote with deiri counting. Besides, it should be on the deiri page, anyway. :)

I would not get into fractional values. Fractions scare a lot of people. There is a lot of that stuff available on SL already. :) Use links. That is what hypertext is for. ;)

Again, bravo on the rewrite! :)

X Re: ((no subject)) (2010-03-26 20:28) [#7572]

"If the opponent answers a move, it is sente" - ok, but that is the most trivial meaning of sente. Sente is only useful as a concept that helps you make a decision . Hoping that the opponent will answer why you know that would be stupid, is a bad way of making decisions. So, the likelihood of the opponent answering should be nothing else than the likelihood of his losing the game if he does not. Of course, in handicap games, White always has to hope that Black is more pessimistic about his chances when playing elsewhere.

In the calculation of endgame moves at a certain moment of the game, the hindsight of the opponent having answered is irrelevant. Maybe "sente" means both things in Japanese, but shouldn't we make sure we use the strategic concept only in this context? "White kept sente" in a game commentary on the other hand, is a useful conciseness of speech.

Unkx80: Re: ((no subject)) (2010-03-26 20:58) [#7573]

See my other post.

Bill: ((no subject)) (2010-03-26 17:47) [#7568]

Small rant:

I think that to give the impression that the endgame is about points is dangerous. As I was taught, the endgame is about fighting strength. That is especially so for beginners, something that advanced players may forget. For beginners the endgame is about shortage of liberties. Disasters are waiting to happen. That may be better dealt with on other pages than a theory page, but please do not say that the endgame is about points.

Phelan: Well done! (2010-03-31 23:09) [#7598]

Great work! I finally understood what deiri counting and miai counting are! :)

Back to forum     Back to page

New reply

Path: Endgame =>
Forum for Basic endgame theory
RecentChanges · StartingPoints · About
Edit page ·Search · Related · Page info · Latest diff
[Welcome to Sensei's Library!]
Search position
Page history
Latest page diff
Partner sites:
Go Teaching Ladder
Login / Prefs
Sensei's Library