Life In 19x19

I just started playing go 3 days ago, and was wondering what the best way for a beginner to improve is? I got 'Many Faces of Go' and have been playing the 18kyu and 15kyu opponents on that, both of whom I can beat comfortably. I presume they are not actually anywhere close to 18k/15k level as far as humans are concerned since I'm most likely 30k myself since I just started. Today I did ~650 exercises on 321Go, but I'm not quite sure how useful this is. Will doing all of these exercises help me improve a ton? Or is there a better way? Also, I was wondering how applicable are doing exercises to improving in a 'real game'? Will I just learn to spot patterns in real games where I can exploit my opponent based on having seen a very similar situation in an exercise? When I was working hard on chess, I could do the tactics puzzles very easily in books, but it was many, many times more difficult to spot them in actual games (since I never really knew when to look). I've submitted one game I played to the 'Go Teaching Ladder' (the first game I ever played!) to hopefully get some feedback but I've not heard back from them - do they refuse to review some games?

Also, is it feasible for a complete beginner to reach shodan level in 1 year or less if he works hard? My country's go association has a 'shodan challenge' on their website where they encourage weak players to try to get to shodan in a year. Is this feasible? 1d corresponds to around 2000 ELO from what I've seen on histograms, and I know someone going from being a complete chess novice to ~2000 ELO in a year would be pretty difficult!

Thanks a lot for any advice - much appreciated .

Simba wrote:I just started playing go 3 days ago, and was wondering what the best way for a beginner to improve is?

For now, play lots and lots of games is best, asking questions about things you don't understand (e.g. in this forum's review section, or the KGS teaching ladder).

In general, life and death problems in great numbers are generally accepted to be best for pure improvement. I generally recommend doing whatever you enjoy studying, on the basis that this doesn't seem like work and so is the most fun way to improve, but if you're absolutely fixated on maximum improvement then tsumego is the way forward.

I got 'Many Faces of Go' and have been playing the 18kyu and 15kyu opponents on that, both of whom I can beat comfortably. I presume they are not actually anywhere close to 18k/15k level as far as humans are concerned since I'm most likely 30k myself since I just started.

Ranks like this are hard to quantify. Particularly in computer cases, where a weak player can learn fairly simple tactics to beat a nominally much stronger computer player. Unfortunately, these tactics are not generally good against humans, so you learn bad habits. For improvement, it is a much better idea to play on KGS, which has a good number of beginners and lots of friendly helpful people.

Today I did ~650 exercises on 321Go, but I'm not quite sure how useful this is. Will doing all of these exercises help me improve a ton? Or is there a better way?

Again, for now, play lots of games. In general these sorts of problems are good though, certainly if you enjoy doing them.

Also, I was wondering how applicable are doing exercises to improving in a 'real game'? Will I just learn to spot patterns in real games where I can exploit my opponent based on having seen a very similar situation in an exercise? When I was working hard on chess, I could do the tactics puzzles very easily in books, but it was many, many times more difficult to spot them in actual games (since I never really knew when to look). I've submitted one game I played to the 'Go Teaching Ladder' (the first game I ever played!) to hopefully get some feedback but I've not heard back from them - do they refuse to review some games?

Spotting patterns is a good part of things, because even non-identical patterns may have similar shapes and vital poitns that you can instinctively recognise. I think shape is a much bigger concept in Go than in chess (and can be more instinctively internalised), but it will always be true that spotting things on the board is harder . Doing tsumego also trains your general ability to read out situations (pruning the search tree, keeping moves in your head, reading deeper sequences etc.), which is something involved in every single move of every single game without being tied to the specifics of the problem.

Doing the problems is undoubtedly of great benefit if improvement is your goal, though is not necessary to improve to a fairly high level, albeit possibly slower.

The GTL don't refuse to review games as far as I know, but it can be a very slow system. I recommend the 'KGS teaching ladder' or 'beginners' rooms on KGS, where peopel will look at games and answer questions for you. The review and study sections on this forum are also very good.

Also, is it feasible for a complete beginner to reach shodan level in 1 year or less if he works hard? My country's go association has a 'shodan challenge' on their website where they encourage weak players to try to get to shodan in a year. Is this feasible? 1d corresponds to around 2000 ELO from what I've seen on histograms, and I know someone going from being a complete chess novice to ~2000 ELO in a year would be pretty difficult!

Thanks a lot for any advice - much appreciated .

Shodan in one year is very feasible, with study, or even without much study in some cases. Assuming you're in the UK, the shodan challenge isn't strictly about reaching shodan, but at improving to a level chosen by the player at the beginning. Shodan is a fine goal, but don't get hung up about it - if nothing else, frustration will make improvement even harder! Shodan is probably a bit easier to reach than 2000 chess ELO, from what I can tell.

Preview Edit: amnal ninja'd me, but here it is anyway:

Simba wrote:I just started playing go 3 days ago, and was wondering what the best way for a beginner to improve is?

Fastest way is to do as many exercises as you can, and keep playing. Best way depends a lot on the person.

Simba wrote:Today I did ~650 exercises on 321Go, but I'm not quite sure how useful this is. Will doing all of these exercises help me improve a ton? Or is there a better way?

Yes, they will help you improve a lot. Better way, again, depends on the person.

Simba wrote:Also, I was wondering how applicable are doing exercises to improving in a 'real game'? Will I just learn to spot patterns in real games where I can exploit my opponent based on having seen a very similar situation in an exercise? When I was working hard on chess, I could do the tactics puzzles very easily in books, but it was many, many times more difficult to spot them in actual games (since I never really knew when to look).

It'll come to you, although mostly it'll be problems you can solve easily, or have memorized (I have [sl=CranesNest]Crane's Nest[/sl], for instance).

Simba wrote:I've submitted one game I played to the 'Go Teaching Ladder' (the first game I ever played!) to hopefully get some feedback but I've not heard back from them - do they refuse to review some games?

Nope, they just take some time to reply. They'll get to you. A faster way to get feedback is to post games here in the Game Analysis subforum, or ask in one of the beginner rooms on KGS.

Try to finish 100 games as quickly as you can.
9x9, 13x13, and 19x19 are all OK -- and the smaller boards can save you time.

Hi and welcome to the forum

Doing problem is hugely useful in improving your strength and is widely regarded as the most effective (it is also widely regarded as the most boring, but that's life I guess) About its applications in a real game: Whenever I hunker down and study for a few weeks I notice a sharp increase in reading skill so it helps a lot. Of course there are different circumstances in a game than when doing tsumego (time pressure, not knowing that there is a actual solution to the problem, not knowing the framing of the question, ect) so you can expect a certain lag. The second thing that helps you improve is playing, especially at your level. There is a saying that beginners should play their first 100 games as quickly as possible.

Reaching shodan in a year is probably possible but not very likely if you want to have a life Since I personally haven't reached shodan and am nowhere near it I am not really qualified to answer this question, but I believe that there are several people on this forum that have reached or are about to reach shodan in roughly 2 years. Not to discourage you or anything but there are several people who have noted that there is a wall somewhere in the low sdk's where you can't improve no matter how hard you try.


If you want to improve really fast right now I suggest you get a teacher and play lots of games against him with reviews. The teacher doesn't have to be particularly strong, anything above 15k should probably do. Using this method I was able to improve to 20k in a week

Simba wrote:I just started playing go 3 days ago, and was wondering what the best way for a beginner to improve is?

The best advice I ever got about improving was this:
In the time you have for studying Go, 1/2 should be used to play games, 1/4 should be used to work on life & death problems, and the remaining should be used to study and replay professional games. The rate you improve will be determined by how much time you can spend on these three things.

For the sake of improvement replaying professional games is only worth doing later on--maybe when you're 5kyu, or 1kyu, or maybe 3dan.

If you really enjoy, go ahead, but if you want fast improvement, play games and do problems. I think computers aren't a bad opponent for now, especially if you play the 9x9 boards, which are also good for improvement. Small boards help you learn fundamental tactics that are what hold you back the most at the beginning.

The game analysis forum on here will give you rapid feedback. If you're playing as much as you say you are, the GTL might be too slow.

Redbeard wrote:

Simba wrote:I just started playing go 3 days ago, and was wondering what the best way for a beginner to improve is?

The best advice I ever got about improving was this:
In the time you have for studying Go, 1/2 should be used to play games, 1/4 should be used to work on life & death problems, and the remaining should be used to study and replay professional games. The rate you improve will be determined by how much time you can spend on these three things.

I wouldn't take these proportions literally. I'm sure it varies from person to person. The most important thing, if you want to improve, is to do things that you enjoy. The more fun you'll having the harder you'll work at it.

I will say that if you found that you really loved life & death problems and spent 90% of your time solving them, and 10% of your time playing games, you would probably improve faster. Redbeard didn't mention Go books; there are a lot of great books on go out there aimed at beginners (like the Elementary Go series). A lot of strong players will tell you that beginners can't learn much from replaying pro games -- but it can still be very fun!

Play lots of games!

I think until you're 12-10 kyu, just playing games will teach you important shapes more quickly than problems, especially things like when a group is in atari/cannot escape and such. Get onto KGS or another server and play 2-3 games a day with real people and you'll rocket up the ranks. Problems are still good (goproblems.com is fantastic, do the easy ones there) because they'll teach you other skills, but mostly play for now. When your rate of improvement slows then start to worry about other things.

Simba, a couple questions.

How strong a chess player were you? And while you say you hadn't played any games till three days ago do you mean that you knew nothing about go till three days ago or that you had studied about go from some books but never had anybody to play against? (and so had perhaps a good grasp of concepts in spite of no game experience).

The reason I am asking is that I myself was a weak SDK before I played a game against an actual opponent, human or computer. Just "book knowledge". So it is not impossible (and sounds like you are solving "problems")

Also, precisely what did you mean when you say that you got ManyFaces? Does that mean you bought the program or have just downloaded the "trial version". The weakest level that program has is labeled 18 kyu but I think we would need to ask Fotland whether the trial version alone (no database) is playing at the indicated strength. Perhaps the "trial version" is just to allow potential customers to get a "look and feel" of the product and doesn't play at anything like 18 kyu (the program is the same, trial version and complete version that you pay for; the difference is the presence or absence of the database and programs can behave very differently depending on the result of a check "do I have this database or not" -- aka "data driven programming" which I used to do a lot of during my days in the "cypher mines").

A couple people here have mentioned goproblems.com. As far as internet l&d problems go, I like gochild2009.appspot.com a lot better. (A matter of taste, I'm sure. I find GoChild is more consistent, has better designed problems, and I learn both from the problems that are a bit too hard and from the ones that are a bit too easy.) But I like hard copies of l&d problems even better, because it prevents the "what if I click here" syndrome and forces me to read out the problem.

Best advice I ever got from a pro about how to improve:

it doesn't matter that much WHAT you do, as long as you spend at least 10 minutes every day with nothing but go on your brain. (paraphrasing - this was a few years ago)

Problems are great, playing is good, reading is fine, just make sure your brain is aware that this is something it'll have to think about every day and it'll work on making the connections.

Gosh, thanks a lot for all the replies! I wasn't expecting quite such a response !

Amnal: Yeah, I really enjoy tsumego problems, and go puzzles in general (tesuji etc.). Is there a name for go puzzles as a whole (since 'tsumego' is life/death specific)? I'm not quite sure why people say they hate tsumego problems. I think they're very interesting. I love to solve puzzles; I do a lot of nurikabe/kakuro/sudoku/slitherlink puzzles in my spare time. Tsumego are just a new type of puzzle to me . I'm doing a maths degree too so I love solving puzzles and logic.

I played a 9x9 on KGS earlier which I won, but my opponent didn't have a rank next to his name. He resigned after 30 moves or so, didn't seem too strong (yay for me not getting smashed ). Might try a 19x19 on there later if I can find a 28-30k player.

Phelan: Cool, I've posted a position in the game analysis forum just now for feedback.

Mnemonic: Hmm, I'm quite poor - I can't afford to have a go teacher. And if I got a player who was say, 15k to coach me for free, wouldn't he teach me a lot of mistakes (since a 15k player is not playing extremely well objectively)?

A few people: Will playing games on its own really improve me? If so, why? Is this just so I can build up a kind of subconscious/'real life go' experience of situations in real time and build up my response speeds? And should I review every game afterwards? Or most? Or not many?

Mike: I'm not entirely sure - I was invited to join the England chess youth squad (or something like that, can't remember its official title) when I was 13 or 14 but the time controls were very long - at that age I used to like to play 5 minute games a ton. The events were a long way away too. My parents and I thought it would be best if I concentrated on school work rather than chess (I'm now doing a maths degree at Cambridge Uni. so it's gone quite well .)

I started studying go then. I knew the rules, but I'd never really played or studied it before (might have had an odd game when I was a kid online, can't remember). I have the full version of MFoG. I've played and beaten it on the 15k, 12k and 9k levels now, and for fun I played against the supposedly 3d level (Monte Carlo) on a 9x9 and lost by 1.5 points. I posted a late game position against the 9k opponent on the game review forum here (and as you can see it was quite one-sided... I'm not quite sure I give this program much credit...) I'm extremely dubious that the program is anywhere near as good as it claims to be; a real 3d player would absolutely destroy me, surely, not barely scrape a win.

Jts: I've done some of the problems on GoProblems but not on GoChild. Site bookmarked, will check it out, thanks!

Ethanb: Hmm, will do!

Thanks again everyone for your replies!

Simba wrote:A few people: Will playing games on its own really improve me? If so, why? Is this just so I can build up a kind of subconscious/'real life go' experience of situations in real time and build up my response speeds? And should I review every game afterwards? Or most? Or not many?

There are a number of reasons playing games will help you improve. One is that you develop pattern recognition skills and can learn to apply those you learn (whether from Go games or from Go problems) to the broader context of a Go game.

Secondly, it teaches you to think on a whole board level. There are problems you can do for this too, but the difference is that playing a game of Go forces you to create a narrative where you need to think on the whole board not only about the current situation but about future situations and how the game has developed. No Go problem can teach you to give your Go games narrative/consistency.

Thirdly, it's just as good practice reading as Go problems, albeit less efficient in terms of time.

Reviewing games is great, but don't worry about reviewing every game especially if it means you play fewer games.

About a year ago, I was improving about a stone a week in the ddks and high sdks from just playing three medium paced games a day. This eventually slowed down to about a stone a month from 6k to 3k. Then I've run out of time to play a lot, and my improvement has slowed hugely.

In my case, at least, nothing works as well for improvement as playing a lot of serious games.