Well it turns out translation is slower harder work than I thought, so the promised videos will take longer than I expected. Sorry to anyone who is waiting.
Also, the series is guidelines for those that want the advice of a pro. It's not a thorough argument designed to sway those who disagree. (Just to manage expectations
)
@Kirby: Then the beginner problems I'm doing are too hard for me
I might try looking for some really simple stuff I can do quickly as supplementary study though.
Bill Spight wrote:
Splatted wrote:
Would you consider a problem easy if you found a successful line quickly but struggled to read it out thoroughly?
To quote myself from the Fudge Factor page on SL:
Much, much later (2019): Now I have a strict criterion for solving a problem. You have to be able to answer all of the opponent's replies.So you wouldn't consider it easy. That's how I initially interpreted that quote but I wanted to be sure. We mostly agree then. One side says "Go back to the basics and hold yourself to such a high standard that they seem challenging once again", and the other says "Do not dismiss problems as not challenging enough if you can not solve them to a high standard of completeness." They are essentially the same thing.
Our differences are in the smaller details, and imo are more about efficiency and personal preference than right and wrong.
Bill Spight wrote:
Splatted wrote:
I didn't mean to imply there were only two variables, just that I suspect people often match the amount of reading they do to the difficulty of the problem and so attribute their improvement to studying difficult problems.
I guess we are talking to different people.
What I have heard, over and over, is the advice to do a lot of easy problems, and do them quickly. Until this discussion, I have not heard anybody attribute their advancement to doing hard problems. (Edit: And I have heard more than once people claim that they are working on problem sets that they get right at least 90% of the time instead of moving on.) The 50% rule is a Goldilocks heuristic: not too easy, not too hard, but just right.
Edit: The handicap system in go is, in terms of advancement, an example of the 50% rule. You keep on adjusting the handicap until you get a challenge that is just right for learning. That's one reason why reviewing your own games is one of the best ways to advance.
Well I've heard all sorts. As you say, "do a lot of easy problems and do them quickly" seems to be the most common advice, but that naturally leads to a lot of people trying it and finding it doesn't work. I actually think easy and quick is not wrong though, just missing important clarifications. It's almost like Chinese whispers. Pros say solve lots of easy tsumego, and that gets repeated without the proper context. It's intended to be a subordinate goal to the approach described above. I have heard similar advice time and again from expert musicians, artists and writers etc, as well as pro go players of course.
There is a simply massive amount to cover in order to reach a high level, and each step along the way requires more than the last. In order to cover it all, you
must increase the speed at which you work. If you prioritise speed over accuracy that's putting the cart before the horse, but if you can improve your speed without sacrificing accuracy, you are suddenly capable of learning more in the same amount of time. Even small differences can have a cumulative effect that is significant over time. There will be more depth
and breadth to your study.
Imagine that player A has solved all the problems player B has solved, but also done them more times and had time for some others. Is it not obvious that if player A hasn't sacrificed accuracy they will be ahead of Player B? That may seem like a big if, but when you treat speed as another aspect to be worked on, it is entirely possible to improve.
Bill Spight wrote:
Splatted wrote:
*I haven't actually heard of the 50% rule. It sounds worth looking in to but for now I suspect this may be the one point of genuine disagreement. I think if I'm failing 50% of my problems because I overlooked variations, that means I'm not being thorough enough in my reading, and if I'm failing to find any solution that means I'm attempting problems I can't possibly expect to read thoroughly. I like 50% even less when applied to other things. A musical passage I get wrong 50% of the time? I'm spending as much time drilling the mistakes as I am the correct version!
Not so. As for drilling mistakes, I have made the same point here on L19 about attempting to read too deeply.
The 50% rule is combined with overlearning, so that most of your practice or drill is in doing it right. Overlearning means that you don't just work on something until you get it right, you keep on doing it right, again and again.
I realise my mistake. I was assuming you were aiming for 50% percent across the whole study session, but I now realise that reviewing is different and it sounds much more reasonable. I still think a higher success rate is better though. In fact the 90% you cited as a mistake seems like a good number, though towards the upper limit. The reason is basically what I wrote above. Getting the problem right is only part of the equation. Speed and fluency also matter, and a problem cannot be considered fully learned if you struggle to read it out, even if you do get everything right. A 90% success rate means the problems still require careful reading or the failure rate will shoot way up, but they are accessible enough that you can also challenge yourself to read quickly as well.
In my experience, this difficulty level, where you mostly know what's right but still have room to improve, is the true goldilocks zone of learning. Not only can you work through the material quickly, and thus cover more, but everything is easily internalised because it's all at that "next step" level.
I presume you have seen enough evidence for the 50% rule that individual opinion doesn't count for much against it, so let me propose another way of looking at this. I think you're actually being too lax in what you consider successfully solved. You are considering
only that which can be easily measured and aiming for 50% in that area. Perhaps consider a time goal like Kirby suggests. Even with a modest goal, getting 50% of the problems correct in time would naturally mean that the percentage of answers that meet your current standard would be higher than 50%.
My point isn't that you should have a time goal. My point is that even were the 50% rule the literal word of god, your application of it would still be just one possibility. Those 90% success rates may in fact be in line with it when working on more than just accuracy.