What I Get Out of Deeply Analyzing My Games Without an Engine
There is, I think, a big difference between deeply analyzing your games (without an engine),
lightly annotating your games (with or without engine assistance), and simply looking at engine analysis of your games.
The hardest and most effort-intensive of these is, of course, deeply analyzing your games without an engine. I never did this until I started taking lessons from @JesseKraai and, ever since, I have not looked back (you may want to check out Jesse's video here).
I'm not a GM (nor an IM, FM, NM, or even an expert), just a lowly USCF Class-A player. So I'm not going to presume to tell anyone why they should analyze their games this way. What I will simply do is tell what I get out of it, that I don't think I can get out of, say, light annotation or simply reviewing an engine analysis, and let anyone who reads this decide for themselves whether they think it's the right thing for them.
There are three things I get out of deep, engine-free analysis that I don't think I can get out of light annotation or engine analysis.
1. A deeper understanding of positions. The computer is great evaluating positions, but it isn't so great at explaining why a position is good or bad. Yes, I can look at a position the computer says is better for White or Black and try to figure out why one side is better. But doing that I'm already cutting out half the equation—deciding which side is better to begin with—because the engine is already telling me. So instead of looking at the features of a position and trying to decide for myself, the computer would tell me who's better and I would then try to justify the computer's evaluation. This is probably not without merit—kind of like solving puzzles when you know there's a tactic to be found. But I'll argue that starting from scratch and first identifying the key elements of a position and then deciding on my own, based on those elements, who is better is the more fruitful exercise.
2. An evaluation of my own thought processes. The computer is of no use here. One of the things I want to do to improve is to root out errors in the process of my thinking. That is, rather than just figuring out that a move I made was wrong, or even why the move is wrong, I want to figure out why I thought that bad move was okay to begin with. What was the error in my process? What was I afraid of in the position? What made me overconfident? What did I overestimate? What did I underestimate? Most importantly, why did I over- or underestimate it? This is something I can work out only on my own.
3. Heavy mental "lifting" to bulk up my analytical and evaluative "muscles." This is probably the biggest thing I'm getting out of my engine-free analysis. It's like a lineman in American football who trains by bench-pressing 400+ pounds, squatting 550+ pounds and deadlifting 600+ pounds. They're certainly not going to move anything like that kind of weight during a real game. But lifting that kind of weight in prep for their games makes them stronger and better able move the kind of weight they have to move (i.e. pushing another 320-pound dude out of the way) during a game. I spend no more than four hours playing any given game, (assuming a time control of G90+30) and most games I play are shorter than that. But it's not uncommon for me to spend around 10-14 hours analyzing a games. I'm convinced that this lengthy process helps me to build up evaluation skills and catalog positions and concepts to bulk up my "inner database" of patterns and ideas in a way that no other training activity could, especially where positional thinking is concerned.
So there it is. All I can say is that I spent years at a USCF rating plateau in the 1600s. Then I started analyzing. Analysis has now been the core of my training for two and a half years, and my peak rating has climbed 171 points. As a result, I am now a true believer in the value of deep, engine-free analysis of my own games, and I don't plan on walking away from it anytime soon.
To wrap up, here is the latest analysis I completed with some lines added by Jesse who went through this game with me. My analysis is far from error-free (as Jesse points out to me in lessons). But in a way, making those errors is part of the process.
