Project Shodan, KGS nick - Vladimir

[vpopovic]

I really would advise not getting hung up on what your KGS rank is until it is very, very stable (as in over 100 games played at that rank).

I don't plan to stick to the same rank in 100 games or more

[Boidhre]

I really would advise not getting hung up on what your KGS rank is until it is very, very stable (as in over 100 games played at that rank).

I don't plan to stick to the same rank in 100 games or more

Ah, the optimism of those fresh out of the ddk ranks.

[vpopovic]

8k again
I hope this time it will last at least till tomorrow morning

[foeZ]

8k again
I hope this time it will last at least till tomorrow morning

If you're serious about getting 1d within a year you should consider completely relinquishing your rank. Play games on a guest account or play unranked games on an unranked account against players that are no weaker than 1 dan.

Ask each and every one of them to review the game afterwards and try to understand their combined advice to come up with your own moves which are supported by your knowledge and experience.

Playing for rank is hindering true progress because you're too eager to win rather than to play the right move.

I hope that made sense.

[vpopovic]

Thank you for this advice but I don't understand.

If I'm playing for the rank then I'm trying to win. If I'm trying to win then I'm trying to find the best possible move.
I don't see how's that different from playing w/o ranks and trying to find a proper move.

I'm sorry, I just don't understand. I would be most grateful if you could explain that to me a little bit more. Thank you in advance.

[skydyr]

If you're playing to win, you will fall back on what you know instead of experimenting, because you know it works (to a degree).

[jts]

I think the short version of the idea is that playing games will make you better, regardless of what your particular rank is at each game. Let's say, for example, that we predict it will take you 1,000 games to get to shodan. It doesn't actually particularly matter whether at game 500 you are at 4k or 3k or 5k. What matters is how many games you play (and, to a lesser extent, how long they are, how challenging the opponent is, whether you review, how you pair the game with problems and study). A change that makes a 100 game difference in when you hit 4k for the first time might only make a 10 or 20 game difference in when you finally hit shodan.

Now, my own view is that one of the things that makes getting better at Go more fun than, say, exercise is that you get a shiny number to prove how hard you've been working... sort of like a Boy Scout badge. As superficial as it seems, for some people that is exactly the sort of positive feedback that helps them put in a lot of effort.

At the same time, the longer you play, the more spaced out that particular form of feedback becomes. If you let yourself become acclimated to the steady drip, drip of new changes on your rank graph, it will suddenly be very disappointing when these become imperceptibly slow. (Even though you expected, or should have expected, all along that the day would come when you would play first 100, then 200, serious games without gaining a rank.)

Furthermore, it becomes possible to divert effort from the basic mission of playing go into the side mission of boosting your rank, or thinking about your rank. You can find pages and pages of whining on this forum from people who KGS should change its rank system to a less accurate one which will give them the rank they think they deserve. You've already had a little taste of this - it actually is interesting to learn why your rank on KGS can fall without you playing a game, and I doubt you lost much sleep over it. But if this is beginning to happen to you already at 8k, can you see how you might spend more and more time worried about it? Can you see how you might start to avoid playing because you're "too tired" or "too stressed out", and you don't want to play when you're "off", because it will effect your rank?

[foeZ]

Thank you for this advice but I don't understand.

If I'm playing for the rank then I'm trying to win. If I'm trying to win then I'm trying to find the best possible move.
I don't see how's that different from playing w/o ranks and trying to find a proper move.

I'm sorry, I just don't understand. I would be most grateful if you could explain that to me a little bit more. Thank you in advance.

If you're playing with the sole purpose of winning, you're not playing to learn something.
You can only learn from games that you lose because in the games that you win, your opponent wasn't strong enough to punish your mistakes.

It takes a few weeks to develop a bad habit, and it takes a few months to unlearn this bad habit. If you play only to win, you will make overplays that "work" against people of your own rank. But people of higher ranks will punish it and you will lose the game. By the time you will face these stronger players, you have already developped this bad habit and you will hit the infamous "wall" around 5kyu. It will take a lot of time to unlearn everything you have learned by playing against "weak" players and to learn how to play a better move that also works against stronger players.

This is why I suggest you just relinquish your rank and only play against players that are much stronger than you. This way you will skip the whole part where you learn bad habits because they will immediately be punished and you won't make the same mistake over and over and over.

Instead, what will happen is, you will learn some good moves and strategies from seeing how your opponent crushes your moves that you consider to be good. Once you go back to playing people that used to be as strong as you, they will make the mistakes that you have made against this stronger player, and you will remember how to punish those mistakes because you have been punished for it too.

I hope that makes sense now.

[Boidhre]

Thank you for this advice but I don't understand.

If I'm playing for the rank then I'm trying to win. If I'm trying to win then I'm trying to find the best possible move.
I don't see how's that different from playing w/o ranks and trying to find a proper move.

I'm sorry, I just don't understand. I would be most grateful if you could explain that to me a little bit more. Thank you in advance.

There are a few aspects. One is psychological, when you play to learn not to win there is less pressure. The second is as others have said you'll play more conservatively when focused on your current rank. You'll be less likely to try a larger extension than normal because you're not sure how to handle an invasion into it whereas if you're playing to learn you'll go out of your way to create novel situations on the board, not just play the same approach to a corner stone all the time and so on. Thirdly uneven handicap games will be rare, it can be very good to take white and give two stones to a player a bit stronger than you, you pretty much cannot win barring a very large blunder but it can be eye-opening to the game because you'll see exactly what you'll face in several stones time. Ditto taking two stones from someone who can beat you on four. Such games make excellent review material for you because you will get soundly beaten and figuring out how this was done is valuable.


Someone said to me quite a while ago, I think it was Bill but I might be wrong, that you'd be better off not knowing your rank at all until you were a strong single digit kyu strength. Before then it'll just distract you from what you need to be doing to improve.

[vpopovic]

Thank you all, guys. Now I understand what you mean.
I'll give to all this written a deep thoughts.
As jts wrote, there are people who are very motivated with their rank. I'm one of them. I just need to measure benefits of this motivation versus all bad things you mentioned, applied to me and my mental construction.

On the other side, I'm perfectly aware of the fact that any ranking system can not absolutely measure one's ability.
I gave it a thought or two and the conditions to determine your absolute rank are:
1. To know absolute strenght of all go players in a world at a given moment
2. At that given moment (so players' strength could be constant), to play infinite number of games with every single world go player, so you could establish absolute difference between you two
3. To computate all data at the same moment into one absolute ranking system

Since all this conditions are impossible, there's no way to know your real strength. Not to mention differences between KGS, IGS, EGF, AGA, Japan,China, Korea etc. ranking systems.

Knowing all this, I'm not obsessed with my KGS rank. As Bill Spight told me, I just take them as a stepping stones to my final goal. And final goal is to become Shodan. Not because KGS Shodan title means something as a title in a real world (hey, I can't even brag outside of this forum because IRL almost nobody even know what go is ), but because when I reach Shodan, I'll know I'm good go player. Not excelent, not fantastic, but good.
For me, knowing that will be enough. Next best thing to being worlds best go player

I hope I explained well my motivation for this journey and this study journal.

Cheers

[jts]

I hope you don't mind if I think out loud about how to make this point best in your study journal.

I actually disagree with something that Ed said a few days ago, and has said before: that there is a difference between your official strength (what he calls "rank") and your true strength ("level"). I would say, instead, you can talk your go ability or strength in two ways. You can talk about with predictions about who you can beat, and how often; or you can talk about it concretely, in terms of how you would analyze a specific position, what move you would play (and, more broadly, how you think about broad classes of positions). If you want to talk about strength in terms of predictions about winning and losing games, you are implicitly comparing two things (wins and losses) to see which is bigger. And, to paraphrase xkcd, it's hard to predict which of two things will be bigger without using numbers.

Now, the reason why I think this is different from what Ed says is because Ed's way makes it sound like there is an observed number that we can calculate by adding up wins and losses and then, in addition, a secret, unobservable number that generates the observed number. But I think that's wrong - or at least, it's an unhelpful way of looking at it. The rank-number is just a prediction about who you can beat 50% of the time under certain conditions (your EGF rank and KGS rank might be slightly different because we're predicting slightly different game-settings, for example), and the only way we can gather evidence to justify this prediction is by observing who you have beaten in the past.

This means that a game you win in May 30, 2013 can be very important to your rank - that is, it can be a very important data point when we're trying to predict who you can beat 50% of the time on June 1. It will be less important, but still quite important, when we're trying to predict who you can beat on June 15. But by the time you reach shodan - call it May 30, 2014 - a game you won today will be dodgy evidence indeed.

Now do you see what we're getting at? If you want to drive up your rank, winning games is the only way to do that, because it's the only thing that will give us any reason to change our predictions about who you can beat. But (assuming you continue to play regularly), changing our opinions about who you can beat today will have almost no effect on what our predictions will be in a year; we will use the games you play next May to make that prediction.

It's easy to become confused about this because when you're starting your wins drive up your rank at almost the same rate that you're improving, so it feels like "when I win, you change your predictions about who I can beat" is the same as "when I win, I get better at Go." In fact, your play improves just as much whether you win or lose - perhaps you improve more when you get thrashed - but nonetheless, we correct our prediction about your win-rate up only when you win, and actually push it down when you lose. This is because our margin of uncertainty about this prediction is normally very wide compared to the amount that a player improves after each game.

[foeZ]

Another way to improve that's pretty interesting. One that I've been using for a while.

First you have to think to yourself: "Why do I review games?"

The answer to that is simple. When you review games, you have a more objective view of the board. In general, your review strength is 2 stones higher than your playing strength. Ofcourse this isn't the case for everyone, but let's assume that this is the average.

If you can put yourself into a mindset where you're watching someone playing a game, and you're in a different game, commentating that live game for weaker players. After each move, you should figure out the motive, explain it (in your head) to those weaker players and come up with a move that you would play to counter it, and explain it (in your head) to the weaker players.

This kind of playing stimulates active thought processes during games. It'll make you play more thoughtful moves and if you get punished, you can analyse what you thought during that move and how your opponent countered it. Combining your thought process and your opponents counter, you can find errors in your thought process and improve it.

[vpopovic]

I hope you don't mind if I think out loud about how to make this point best in your study journal.

I actually disagree with something that Ed said a few days ago, and has said before: that there is a difference between your official strength (what he calls "rank") and your true strength ("level"). I would say, instead, you can talk your go ability or strength in two ways. You can talk about with predictions about who you can beat, and how often; or you can talk about it concretely, in terms of how you would analyze a specific position, what move you would play (and, more broadly, how you think about broad classes of positions). If you want to talk about strength in terms of predictions about winning and losing games, you are implicitly comparing two things (wins and losses) to see which is bigger. And, to paraphrase xkcd, it's hard to predict which of two things will be bigger without using numbers.

Now, the reason why I think this is different from what Ed says is because Ed's way makes it sound like there is an observed number that we can calculate by adding up wins and losses and then, in addition, a secret, unobservable number that generates the observed number. But I think that's wrong - or at least, it's an unhelpful way of looking at it. The rank-number is just a prediction about who you can beat 50% of the time under certain conditions (your EGF rank and KGS rank might be slightly different because we're predicting slightly different game-settings, for example), and the only way we can gather evidence to justify this prediction is by observing who you have beaten in the past.

This means that a game you win in May 30, 2013 can be very important to your rank - that is, it can be a very important data point when we're trying to predict who you can beat 50% of the time on June 1. It will be less important, but still quite important, when we're trying to predict who you can beat on June 15. But by the time you reach shodan - call it May 30, 2014 - a game you won today will be dodgy evidence indeed.

Now do you see what we're getting at? If you want to drive up your rank, winning games is the only way to do that, because it's the only thing that will give us any reason to change our predictions about who you can beat. But (assuming you continue to play regularly), changing our opinions about who you can beat today will have almost no effect on what our predictions will be in a year; we will use the games you play next May to make that prediction.

It's easy to become confused about this because when you're starting your wins drive up your rank at almost the same rate that you're improving, so it feels like "when I win, you change your predictions about who I can beat" is the same as "when I win, I get better at Go." In fact, your play improves just as much whether you win or lose - perhaps you improve more when you get thrashed - but nonetheless, we correct our prediction about your win-rate up only when you win, and actually push it down when you lose. This is because our margin of uncertainty about this prediction is normally very wide compared to the amount that a player improves after each game.

Games you won increase your rank and your strenght.
Games you lost decrease your rank, but also increase your strenght. Those games will capitalize in rank later.

So, winning or losing, playing a lot of serious games will help you get stronger.

Of course, increasing your strenght is possible when you play serious games with enough time to think about moves. Otherwise, you can end up as a player who plays a lot of fast games every day but never improves in strenght because he doesn't review or give serious thoughts to his games.

At least, this is how I see that.

[vpopovic]

Another way to improve that's pretty interesting. One that I've been using for a while.

First you have to think to yourself: "Why do I review games?"

The answer to that is simple. When you review games, you have a more objective view of the board. In general, your review strength is 2 stones higher than your playing strength. Ofcourse this isn't the case for everyone, but let's assume that this is the average.

If you can put yourself into a mindset where you're watching someone playing a game, and you're in a different game, commentating that live game for weaker players. After each move, you should figure out the motive, explain it (in your head) to those weaker players and come up with a move that you would play to counter it, and explain it (in your head) to the weaker players.

This kind of playing stimulates active thought processes during games. It'll make you play more thoughtful moves and if you get punished, you can analyse what you thought during that move and how your opponent countered it. Combining your thought process and your opponents counter, you can find errors in your thought process and improve it.

I'll try this in my next game. Seems very reasonable.

[Bill Spight]

It will take a lot of time to unlearn everything you have learned by playing against "weak" players and to learn how to play a better move that also works against stronger players.

This is why I suggest you just relinquish your rank and only play against players that are much stronger than you. This way you will skip the whole part where you learn bad habits because they will immediately be punished and you won't make the same mistake over and over and over.

Instead, what will happen is, you will learn some good moves and strategies from seeing how your opponent crushes your moves that you consider to be good.