|
How is Vital Points of Go? I don't have that one.
|
#
¿
Dec 15, 2014 02:30
|
|
|
|
| # ¿ May 13, 2024 15:11 |
|
|
Heh, Jasiek and rec.games.go. Those are some memories, though I wouldn't call the flame wars I engaged in back when particularly pleasant. Can't say that I miss it. David Carlton's comment about the density of Sakata's MIddle Game of Go is spot on. It's a good book, but it's *rough*.
|
|
#
¿
Dec 15, 2014 05:10
|
|
|
Any suggestions on working on general fighting/middlegame/sacrifice techniques? When I was playing regularly, I was KGS 3k or so, but have been away from the game for about 18 months. In terms of style, one thing I know I do wrong is to play a bit too passively and just lazily play shape and not fighting back when I should. When I push against this tendency blindly, I find myself in disadvantageous fights against people who fight tactically better than I do, particularly online. To work on this, I've been working back through my selection of (mainly japanese) problem books to brush some of the dust off of reading skills, and I mainly need to play more. That said, studying that would help me work on when to play shape and how and when to fight would be great. I've picked back up Attack and Defense, Breakthrough to Shodan, and Kage's chronicles of handicap go to help there; do these make sense along with getting back to playing, doing problems, and memorizing a few pro games? Also, any particular pros that stronger players recommend studying/memorizing at my level?
|
|
#
¿
Dec 23, 2014 17:04
|
|
|
uranus posted:do you all memorize pro games? do you find it helpful, or enjoyable? For enjoyablility, yes, I find it an enjoyable way to relax at the end of a day. A welcome respite from the workday without the effort of actually playing. In terms of study value, the kinds of moves that the pros find and the places they tenuki or handle indirectly are ones I wouldn't even have considered, which is eye opening. Figuring out the logic behind that is hard, though, and often I cannot; I think there's value, though in just learning that you can tenuki, probe, or handle things indirectly in places you wouldn't have before.
|
|
#
¿
Dec 23, 2014 18:13
|
|
|
uranus posted:so i was going to try to get into chess as well because its very frustrating trying to find human people to play go with in real life, There's not a local Go club where you are? If you're in the states, the AGA website list is usually pretty up-to-date.
|
|
#
¿
Jan 5, 2015 01:21
|
|
|
uranus posted:youre telling me i have to MAKE NEW FRIENDS???/? Who said anything about friends? Victims is the word you're looking for, I think.
|
|
#
¿
Jan 5, 2015 04:02
|
|
|
Wow, I had never seen her videos before. Those are really nicely done.
|
|
#
¿
Feb 2, 2015 19:35
|
|
|
I just like the discussion of how she thinks about where to go, the kinds of sequences she thinks through, and how she evaluates them. No book really goes over that, and it's just nice to have examples of what to aspire to there.
|
|
#
¿
Feb 3, 2015 15:02
|
|
|
Under 15 posted:http://eidogo.com/#url:http://files.gokgs.com/games/2015/2/11/Helsbecter-clipper.sgf There's no direct response to it, but my feeling is that when my opponent plays a strange move like that, the game starts to focus on him trying to make it a good move and you trying to make it a bad move. My feeling is that 30 is a little questionable for that reason, but that 34 in particular helped make 25 a good move because he could link up to it when he jumped out. Tactically, I really don't like 40. Not just because it lost you sente, but because it made it easier for him to threaten connect under on the right since you lost the peep from the other side. I'm tempted to play something around N10 or maybe O9 at that point instead.
|
|
#
¿
Feb 11, 2015 14:38
|
|
|
The AGA posted a link to this online joseki game on facebook: http://www.josekifarm.com/ Overall, it's a pretty decent game-oriented interface to a joseki interface. It certainly pushes me on continuations that I did't know later in joseki, so I expect I'll use it as (yet another) way to procrastinate in work meetings.
|
|
#
¿
Feb 14, 2015 17:13
|
|
|
sensual donkey punching posted:waht the hell is a joseki farm. cant stop clickning Ah. It also why it complains when you pass as opposed to playing a followup that's clearly optional.
|
|
#
¿
Feb 18, 2015 15:12
|
|
|
Nothing specific off the top of my head, though you might have fun looking at the players and games listed on the "amashi" page at Sensei's Library if that's the kind of play you like.
|
|
#
¿
Feb 25, 2015 00:02
|
|
|
Uncle Jam posted:You used to be able to connect through telnet but I think all of that got stripped out a few years ago. You can still use telnet to igs.joyjoy.net 7777 to connect to IGS; all of the clients are built on the basic text-based protocol that IGS has always had. Their client-specific protocol piece support has however, as I understand it, expanded over the years to include additional features in response to other servers. That said, I don't use IGS and am probably still banned from there as a result of the old IGS/NNGS/KGS Usenet rec.games.go flamewars. I'll get signed up for OGS, though. It looks quite nice and something browser-based. On a related front, my KGS contacts (I was an admin there for a long time) tell me there's an HTML5 KGS client in the works. I look forward to not having to go through the Java monstrosity once that finally goes live, particularly given how clunky the original KGS UI (still) is.
|
|
#
¿
Mar 31, 2015 18:16
|
|
|
Must have been an open tournament, as I was able to sign up for it without being a member of the SA group. Someone invite Borachon, please.
|
|
#
¿
Jul 2, 2015 18:23
|
|
|
Xom posted:I try not to spend too much time studying joseki, so I play 4-4 to limit what I need to know. After B 5, Black is probably looking to pincer severely, so instead of a one-space approach, I play a two-space approach. Black pincers anyway. I play what is probably a crude zokusuji. Still, I got strong, so now I go on the offensive with W 18. The way I generally study joseki is to look up and work on learning the common lines after I butcher them in a game. Sensei's library generally has good discussions of the main lines of play on things that I'll try to learn. If it's not on sensei's or seems pretty obscure, I may look it up in a book, but generally don't worry about actually learning it, as it's likely rare and I probably won't remember it anyway. In the case of the 3-4/far approach/pincer, the crosscut is the common response: http://senseis.xmp.net/?34PointDistantLowApproachOneSpaceLowPincer
|
|
#
¿
Jul 3, 2015 14:29
|
|
|
(Double-posting) Since we're analyzing our games here, I'll post some thoughts on my first one, too: https://online-go.com/game/2332298 5: When he approached the corner straight-away, I contemplated extending at the top and making sure I just came away with sente, but decided I didn't want to let him get comfortable on the right side. I played a high approach so I could leave it lightly as needed. 9: jumping out and threatening to connect. After the 10/11 exchange, I felt I could leave the top be and treat it somewhat lightly after 12. I had both Q14 and O10 to make some shape locally, and could always jump out if necessary. 19: Even though I'm "out", I decided to make myself safe here so I can't be chased around more. Wasn't sure if this was too slow. I was planning to make a san-ren-sei on the left if I could get out with sente; my shape is light enough that there was the possibility of it, but I didn't really expect to. Leaving here in gote seemed like the price to pay for getting to three corners first. 21: If he gets L17 first, I'm fenced in while he's connecting up. This seemed like a natural move. 25: Thought about playing E5 instead, but decided I'd just swallow F3 on a large scale and neutralize some of the center influence that white had invested in. He'd get territory for that, but it seemed like a reasonable trade. 37: Thought about playing C7 to reduce the scope of his right side, but decided between helping out my stones on the right side and keeping him from making a double wing at the bottom right. 45: Debated between playing at the bottom and playing on the left. The left keeps him separated, but then my K4 stone starts looking lonely. I was worried about it getting attacked while he built a big bottom right, and getting pressure put on my right side stones at the same time. I guess they're safe enough already, though, so perhaps D8 or thereabouts is better? 48: Honestly expected white to hane here, and was debating between extending to E8 or blocking at C7 when he did. Either seemed okay for me, but wasn't sure what to do. When he pushed to C7, he basically decided for me and I was happy to give up D8 for the opportunity to swallow the F3/4 stones. 60: Thought he might fight a ko here, but I was solid everywhere and he'd have to make the first threat in a huge ko. When I descended at R8, I was planning on giving up P11 stone when he cut at 62 as the cost for demolishing the bottom right and becoming strong in the middle, but then read out I didn't have to. 66: I was going to descend to R12 here before he resigned, since at worst killing the right and connecting at N11 are miai. Even just connecting through having destroyed the right side is a great result. Questions for others:
[EDIT: Grammar is hard.] Borachon fucked around with this message at 15:43 on Jul 3, 2015 |
|
#
¿
Jul 3, 2015 15:33
|
|
|
]I think you want to be careful picking the game to study; a non-trivial set of pro games turn into complicated fights based on deep, complex reading that I'm not sure would the most informative for most people (certainly not me). Maybe one of the stronger players could recommend a few games for us to study?
|
|
#
¿
Jul 5, 2015 16:03
|
|
|
uranus posted:i'm trying to figure out the time setting in these games. is it ten 1 minute periods, or what? From what sensei's library says, it looks like 10 seconds per move plus 10 1-minute byo-yomi periods "for longer thinking".
|
|
#
¿
Jul 5, 2015 19:09
|
|
|
uranus posted:this is correct but play people, not ai If you don't have time to play people, then basic problems (e.g. Graded Go Problems for Beginners vol. 1 then 2) are not terrible once you have some of the real basics down. Coupled with regular play, GGPv2 is great for pushing towards 10-12k once you have the basic tactics handled.
|
|
#
¿
Jul 8, 2015 16:43
|
|
|
YagerX posted:Thanks! The feedback is really helpful! There are a few things I find I really struggle in games that if like to read up on but not sure what it's called. The middle game is hard, particularly because tactics and strategy start to interplay in complicated ways. The classic reference for this is Attack and Defense, but just watching lots of games of somewhat stronger players and playing more will help, too.
|
|
#
¿
Jul 9, 2015 21:01
|
|
|
Mederlock posted:I will get to 20K *shakes fist at self* The Proverb is "lose your first 100 games as quickly as possible" for a reason. It just takes a number of beatings to get a better sense of what kinds of moves to consider when, and that, barring special circumstances, some moves just aren't worth considering. uranus posted:i just ordered Tesuji and Lessons in the fundementals of Go They're both fabulous.
|
|
#
¿
Jul 10, 2015 00:13
|
|
|
AttackTheMoon posted:Where do you people go to play Go? Noone around me plays it so its a bit sad that I keep having to play on my own. There are local clubs all over in the states ( http://www.usgo.org/where-play-go ), though many of them are small and may meet intermittently. If there's a club listed for you on that link, contact the person on it for more information. I do vastly prefer playing in real life compared to online.
|
|
#
¿
Jul 10, 2015 16:15
|
|
|
http://senseis.xmp.net/?BentFourInTheCornerIsDead
|
|
#
¿
Aug 1, 2015 20:15
|
|
|
That is just so wrong.
|
|
#
¿
Oct 5, 2015 23:46
|
|
|
Symbolic Butt posted:https://code.facebook.com/posts/1478523512478471/teaching-machines-to-see-and-understand-advances-in-ai-research/ You know, there's a lot of interesting things there, but a lot that really bugs the crap about me about how the cool things that paper does are being hyped:
EDIT: The actual paper is by a Toronto/Google team, not facebook. Unlike the Google paper, the facebook post gives not actual information about how it does, just "hey, it's strong!". My guess is that this is a fruitful research direction (MCST programs have been looking at neural nets for move evaluation for a long time), but some of this looks like a lot of unsubstantiated marketing hype, too. EDIT2: The facebook video also looks to be against GnuGo 3.8, which was released in 2009. Borachon fucked around with this message at 17:07 on Nov 3, 2015 |
|
#
¿
Nov 3, 2015 16:25
|
|
|
According to Nick Wedd on the Computer Go mailing list, the facebook program appears to be playing on KGS as darkforest and darkfores1. It doesn't play ranked, but general feeling seems to be that it's about KGS 1d or so.
|
|
#
¿
Nov 6, 2015 13:47
|
|
|
pointsofdata posted:Cool that they used machine learning rather than mc They used machine learning to guide MC. MCST is still at the core of the engine. Details of the algorithm published today in Nature, and the article can be found here: http://deepmind.com/alpha-go.html. It's really nice work, and the games are really good. Importantly: "AlphaGo is scheduled to play Lee Sedol 9-dan, the Korean pro widely recognized as the strongest player of the last decade, in a 5-game series in March."
|
|
#
¿
Jan 27, 2016 22:22
|
|
|
pointsofdata posted:Without using mc they still beat all the current programs Only if they give current programs a 5s/move time limit. MC programs are highly sensitive to how much computational power they get. I would be surprised if they did nearly so well against current programs at reasonable time limites.
|
|
#
¿
Jan 28, 2016 03:25
|
|
|
Uncle Jam posted:I think the point of that part of the paper (?) is to show that what they've added initializes the program better for each turn than MC does. Not that what they've done is superior to MCST in all aspects. The paper does *lots* of interesting things. From a quick glance, the main things are that it uses a DNN to weight/initialized MCST searches, it truncates rollouts and uses a different DNN to evaluate those abbreviated rollouts, and it started training these DNNs against a large database of 6d-9d KGS games, and then against its own games later. I expect they have been and will be training against Lee Sedol games. THey also use a lot of horsepower - 1200 cores and 170 GPS - for this.
|
|
#
¿
Jan 28, 2016 15:45
|
|
|
Uncle Jam posted:On the AGA YouTube channel there is a video of Kim reviewing the alphago games, it's good but a lot of the video is Kim demanding to know how the program counts ko threats or other things. When he was commenting on the games and what AlphaGo did well and poorly or well, it was really interesting. When he started to speculate in ways that made assumptions about how the underlying program worked, though, things were much less useful. The whole discussion on the value of sente in a particular position was really interesting and why AlphaGo's play in the first game was problematic was really interesting.
|
|
#
¿
Feb 1, 2016 19:11
|
|
|
sunaurus posted:Could I ask for a game review? https://online-go.com/game/5241383 Generally you did fine, and black's opening was really bad. He put all of his territory in one basically uninvadable spot and banked on being able to make 2-point territories inside yours. When someone does that, the key is to find opportunities to reduce his territory while growing your own. That generally means being more ambitious about reducing the edges of his territories than you were. There were a number of times you could have A few key points: 20: Why block in this direction? If you block the other way, you make a bigger territory, and white is already strong int he lower right, so any thickness you build here isn't that effective. 24: Another chance to switch direction by playing at around S5 since black didn't play S5. Black's already low on the bottom, so switching to S5 would both keep black low on the bottom and get you more of the right side. 26: P1 is small endgame. Don't play moves like this, as it asks him to fix a weakness while keeping sente. 34ff: All of these do expand your middle and reduce his some, but they also strengthen his middle so you can't go in a little deeper later. If you're going to go here, what would he do if you played L8 or L9? By being more ambitious, he either has to risk a fight that would destroy his entire middle territory or cede more reduction to you. 58: I would have played C18 - C13 limits how far black can go on the left side, and if he crawls out that way, he reduces his own lower left. 62: Again, don't do this. Locally, if you just play F17, black has to either give you sente or leave a weakness behind. Your move here asks black to completely fix his weaknesses and keep sente. 68ff: You need to decide on an approach here. Are you trying to make his area smaller, your larger, or what? You're basically securing a small middle area of your own while solidifying a large area of his. I'd at least go farther into E11. 76: I'd play H11 - you're threatening more damage to his big area than he's threatening to yours. 104: This is really small, especially compared to A11 or M1 or even S4. What do you do if B ignores it? 110: Gote, compared to wither A11 or M1.
|
|
#
¿
May 14, 2016 15:26
|
|
|
Pander posted:Why? White R2, black hane, white cut, black atari, then what? If B hanes at the head of the two stones there, W R2/B Q2/W Q5 is really ugly for B and pretty simple. W Q2/B P2/W R2/B P3/W R6 is also bad for B. Sensei's has a section on it: 4-4 Low Approach Invasion Interception
|
|
#
¿
May 19, 2016 16:06
|
|
|
PerniciousKnid posted:How do people teach Go to would-be casual players? I read somewhere that 9x9 games are the standard entry point, but I feel like 13x13 would be more like "real" Go and therefore more interesting without being too complex/long. I'm trying to lure people in without either boring or overwhelming them. I like teaching on 9x9 and 13x13, because so much of the early learning process are tactics, scoring, and basics of how stones interact. 9x9 and 13x13 let them do more of that. That said, some people are really turned off by not playing the "real game". For people who insist on "playing the real game", I generally just play the full board and loan them a problem book or point them to a website so they get basic tactics down a little faster.
|
|
#
¿
May 31, 2016 17:26
|
|
|
Overwined posted:Words to live by. 11: This has to be at D17, making the "hane at the head of two stones" shape. 17: Neither attachment is correct here. The attachment on top is okay when you can connect your two sides and make thickness. The side attachment is okay when it makes your opponent heavy and you don't have to back up. Generally, though, you don't want to attach to weak stones if you're strong and you want to attack because attachments usually make both sides stronger. Here you want white weak and running. E3 and D3 are both much more natural. 29: Read before you play, not after.  41: This move is vague. What's the worst that can happen here? If w comes into this area, b can attack him strongly, reducing right's right side in the process. 47: Does this threaten two cuts, or just ask white to reinforce his large area? 51: See how big N14 is? It threatens to swallow the white stone at the top on a large scale, and reduces w's right side. What you played asks w to connect to his weak stone while attacking you, instead. Which he did. 59: W did indeed do B a favor there. 90: This isn't a aji, it's just a weakness - w takes pure profit in sente. 105: Is this really a serious weakness as it stands? W has a the edge and lots of supporting stones nearby. Simply pushing like you did is just a reduction. If you want more from it, you're probably going to have to start something above it near R14/R16 and see what comes before forcing w to strengthen in the area. 169: Take and make the ko. You can't live in that much space directly. More generally, you and your opponent both try to make modest-spze space in the middle too directly with gote moves, and these moves end up being a little vague. In particular, I'm thinking of moves like N5, H11, and H7. Center territories tend to have lots of openings into them, so it's hard to use gote moves to secure center territories. For building in the middle, you generally want to either: 1. Attack or lean on something to close off part of a center in sente, or 2. Sketch out something so large (a moyo) that the opponent has to make a hard choice between either (a) diving in and giving you something to attack to solidify a smaller but still significant portion of it, (b) reducing it, giving you something still sizable because they're scared of jumping in. Borachon fucked around with this message at 18:37 on Jun 3, 2016 |
|
#
¿
Jun 3, 2016 18:34
|
|
|
I found 38 basic joseki + a decent joseki reference (I use Rui Naiwei's Essential Joseki) to be a nice starting place. Basically, 38-basic gives you a good idea of what you're trying to get out of a corner, what an even exchange looks like, and when you might steer one direction or another. Then you play games, and after the game, look at a somewhat longer reference like Essential Joeski to find out "what should I have done in this corner I butchered?". The online references are of uneven quality - I like them as an additional "what do some people do here", but, for me, they're hard to use for more than that.
|
|
#
¿
Jun 17, 2016 18:26
|
|
|
Athanos posted:I'd say volumes 3-6 are the only ones you need to read. Tesuji and attack and defense are the two you should read imo. Endgame is actually quite good, too, though not as "sexy" in terms of learning how to make flashy plays as Tesuji or Attack and Defense.
|
|
#
¿
Sep 30, 2016 14:19
|
|
|
HummedExplosions posted:In the first case I tend to take a point like A and try to invade later at B and build outside thickness. In the second case I've tried things like E5 and C3 but I never seem to get a great result. How should I be thinking about these two situations? Is there a better way to handle them? I'm a 9k on KGS in case that's at all relevant. In the first case, A is somewhat small - you're being greedy for a corner territory when the game is going to happen in the middle, and that corner is still invadeable. At the top, I'd favor K16 instead, so that if w comes into the top, he's already pincered. At the bottom, the one space extension on the left instead of the knight's move would put you in a better place to play B later. Again, you're being greedy for territory when the middle is where the game is going to happen. In the second case, E5 is the right move. What's your opponent doing that leads to a bad result?
|
|
#
¿
Oct 4, 2016 14:15
|
|
|
derp posted:no wtf is that? Cho is playing DeepZen tonight starting at 11 pm EST. Commentary by Myungwan Kim on the AGA streams on twitch and youtube starts 2 hours later. Borachon fucked around with this message at 01:01 on Nov 19, 2016 |
|
#
¿
Nov 19, 2016 00:44
|
|
|
empty whippet box posted:Anyone got a solution for this? I've been staring at it for a couple days and I can't figure it out. G2/F2/G3/G4/H1? EDIT: F1 or G1 instead of H1, since H1 forces B to erases the aji of the cutting stone that you still need to capture some B stones, and W loses the capturing race if she forces B takes the cutting stone by playing H1? Borachon fucked around with this message at 17:28 on Oct 14, 2017 |
|
#
¿
Oct 14, 2017 16:55
|
|
|
|
| # ¿ May 13, 2024 15:11 |
|
|
Shadonra posted:a2 b1 (without this exchange, Black can give up 2 stones at d2 and monkey jump you to death) then e3 e2 g2 f2 g3 g4 f1 I obviously left off the e3/e2 exchange from my sequence, which I had assumed. A2/B1 does seem necessary first.
|
|
#
¿
Oct 14, 2017 18:06
|
|