Life In 19x19
http://www.lifein19x19.com/

guessing right too often ! ?
http://www.lifein19x19.com/viewtopic.php?f=10&t=5767
Page 1 of 1

Author:  MLC [ Fri Mar 30, 2012 8:00 am ]
Post subject:  guessing right too often ! ?

When I give up on a Go problem I take a guess before looking at the answer. A surprising number of these guesses are right but I don't consciously know why I'm making these choices! Does this happen to anyone else? It's kind of neat but kind of strange ...

Author:  xed_over [ Fri Mar 30, 2012 8:26 am ]
Post subject:  Re: guessing right too often ! ?

yes, I often just play on what looks like it should be the vital point without reading it all out to the end, and am often lucky (with the easy problems :) )

but if you want to get better and understand why the right answer is right and why the wrong answer is wrong, then you've got to take the time to explore all the options -- trying to play the best response for each color at each step.

then when finding a bad result for one side, back up and try a different variation for that side, until you can find the best result. then at each step do the same for the other color.

it can be tedious and slow, but it will help your understanding, then when you can "see" all these variations in your head without playing a stone, it will improve your reading.

Author:  Bill Spight [ Fri Mar 30, 2012 8:47 am ]
Post subject:  Re: guessing right too often ! ?

MLC wrote:
When I give up on a Go problem I take a guess before looking at the answer. A surprising number of these guesses are right but I don't consciously know why I'm making these choices! Does this happen to anyone else? It's kind of neat but kind of strange ...


Kind of human. :)

Author:  MLC [ Fri Mar 30, 2012 9:01 am ]
Post subject:  Re: guessing right too often ! ?

xed_over wrote:
but if you want to get better and understand why the right answer is right and why the wrong answer is wrong, then you've got to take the time to explore all the options -- trying to play the best response for each color at each step.
.
I've started to dabble in that direction but only against EasyGo problems where you can hit the "Try" button and actually play it out. I really appreciate that feature.

Author:  Joaz Banbeck [ Fri Mar 30, 2012 12:06 pm ]
Post subject:  Re: guessing right too often ! ?

MLC wrote:
When I give up on a Go problem I take a guess before looking at the answer. A surprising number of these guesses are right but I don't consciously know why I'm making these choices! Does this happen to anyone else? It's kind of neat but kind of strange ...


That type of thinking is at the forefront of current semi-pop psychology right now. I recommend a book called 'Thinking Fast, Thinking Slow' ( http://www.amazon.com/dp/0374275637 ).

some reviewer wrote:
The mind is a hilariously muddled compromise between incompatible modes of thought in this fascinating treatise by a giant in the field of decision research. Nobel-winning psychologist Kahneman (Attention and Effort) posits a brain governed by two clashing decision-making processes. The largely unconscious System 1, he contends, makes intuitive snap judgments based on emotion, memory, and hard-wired rules of thumb; the painfully conscious System 2 laboriously checks the facts and does the math, but is so "lazy" and distractible that it usually defers to System 1. Kahneman uses this scheme to frame a scintillating discussion of his findings in cognitive psychology and behavioral economics, and of the ingenious experiments that tease out the irrational, self-contradictory logics that underlie our choices.

Author:  SoDesuNe [ Fri Mar 30, 2012 1:02 pm ]
Post subject:  Re: guessing right too often ! ?

Most Tsumegos I solve by verifying my initial guess. The problem really is the thing Joaz Banbeck quoted: If you end up with a high success rate, you might get a handful of problems wrong because you become lazy to always check the variations.
I have quite a few blind spots in Tsumegos (not even with hard Tsumegos). In a lot of cases it has something to do with making an empty triangle (popular example is the one corner shape where you live uncoditionally by making an empty triangle) or when this shape is the strongest counter-move because I don't even bother thinking about playing a move like that.

Author:  CnP [ Fri Mar 30, 2012 3:19 pm ]
Post subject:  Re: guessing right too often ! ?

Correct guesses like that is you learning to spot vital points I imagine. Especially for easy problems it's likely that a) there is a solution and b) it starts with the obvious vital point. In real play and tougher problems though you've got read out whether it works or not - remember "almost a tesuji is a duffers move"

Author:  gogameguru [ Fri Mar 30, 2012 10:50 pm ]
Post subject:  Re: guessing right too often ! ?

Joaz's quote sums it up quite nicely. Here are a couple of other random thoughts on this...

In a game, you can play mostly by intuition like this too and it gets easier to do so the more fluent you become in Go. If you're strong enough, you may not even play any 'bad' moves in a particular game, but you'll also miss quite a few opportunities to play 'strong' moves. Instead, you'll mostly be playing 'normal' moves. (let's just accept my very loose definitions of bad, normal and strong moves for argument's sake)

There's nothing wrong with normal moves and probably 80-90% of the time, a normal move is also one of the best moves. A strong player will usually come up with that sort of move within seconds and being able to do that, and play normal moves consistently, accounts for a large part of a person's Go strength. When more than one normal move is available, the choice is dictated by a player's style.

After that, time is spent; first making sure it works and next looking for something better. So these are the two things that you give up on when you don't or can't spend much extra time.

But the point is that playing strong moves (or giving your partner an opening to do so) is usually what decides games. And since you never know when you'll strike gold, you have to search on every move. It's not efficient, but it's important nonetheless.

In addition, based on research on learning from other fields, my belief is that playing a game where you exert yourself by trying to do better on every move helps you learn much more than playing on cruise control does. This (in my opinion) is also the difference between blitz and slower games. Or solving problems quickly vs slowly, which is basically the same thing.

I'm not saying there's no value in blitz games, or fast drilling of problems. My experience is that they seem to help improve fluency, but overdoing it also seems to be damaging, because it teaches you to think in overly shallow ways. Bill Spight has also mentioned this sort of thing when talking about what he called 'overlearning', and I believe he's read papers about that, but I haven't, so I'm just relaying my personal experience here.

It differs from player to player, because some people mainly play fast games and some people almost never do, but an individual's ability to play well under different time conditions can vary wildly.

For example, in my case, I tend not to play many fast games anymore because I find they don't help me improve much. So I'm relatively weak in blitz games. On Tygem I play as a 7d in 30 min games, but on another account (which I sometimes use when I can't find a slow game) my level is more like 5d, because I make mistakes or miss opportunities to properly counter unreasonable play. This is also why it's important to review your games, because you will spot stronger moves (and learn from it) during a review. Two stones at this level is quite a huge difference and I may be an extreme case. I think it illustrates the point I'm trying to make though.

Author:  Mef [ Sat Apr 14, 2012 6:24 am ]
Post subject:  Re: guessing right too often ! ?

There are many problems where the first move is forced or near-forced (ex: It is black to kill and a white move at 'A' lives instantly regardless of where else black might have played...so black's first move must be 'A' if there is to be a killing sequence). These are the types of problems that get solved much more frequently when they are presented as tsumego than when they appear in games (because if you know there is a solution, you know it must start at 'A'...but if you don't know there is a solution you may not see the way to kill). This is why it's so important read out the sequences in advance (though admittedly in a game if you were under time pressure, you could just start with the "forced" part of the sequences and read as you go).

Author:  tchan001 [ Sat Apr 14, 2012 6:38 am ]
Post subject:  Re: guessing right too often ! ?

If you start trying some of the harder problems, you'll often find that the intuitive move is the wrong move when you finally look at the answer. When you do easy problems, the problems are usually purpose built to teach a certain concept which if you recognize will often lead you to easily find the right move. For harder problems, you often have to choose between different concepts or even apply multiple concepts. If you only depend on what you recognize then you will miss the right answer if you don't read out the problem. If you are not strong enough to read out the problem, I do suggest looking at the answer to learn the new concepts and why the move you thought was right is a failure.

Author:  entropi [ Wed Apr 18, 2012 5:15 am ]
Post subject:  Re: guessing right too often ! ?

I remember some time ago, solving problems of inappropriate level had a bad effect on me in the sense of getting paranoid.
Seeing so many "surprising" tesujis above my level, had the following effect:

I : This group is obviously alive, no need to spend another move to defend it.
Me : Ok, but what if he plays this and that and these and those?
Myself : Confusion alarm Confusion alarm!!! I cannot read any more -> better safe than sorry -> unnecessary move

This kind of attitude not only loses the game, but also reduces the teaching value of a game.

Page 1 of 1 All times are UTC - 8 hours [ DST ]
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group
http://www.phpbb.com/