What I've Learned About Tactics

I have a friend who is much stronger than me, maybe 400 points stronger than me. The first time he played blindfold he could get through the whole game without making an illegal move, while I get lost once the middle game hits. Recently he is taking up tactics training on a major tactics website (since I believe chess.com doesn't allow mention of other sites, you'll have to guess).

I'm not good at chess. I'm not good at tactics. But, I still think I have learned some worthwhile lessons about tactics I've learned the hard way over the years of being a beginner and plowing the tactics fields:

1. In the beginning, I underestimated what you can learn from tactics. I think this was because a lot of people undersell what tactics can offer. They say vague things about vision or pattern recognition or they say you will become able to calculate deeper lines or whatever. I'm not convinced they are super clear on what they're talking about.
2. I thought tactics were inauthentic--in a real game you can't assume there is a tactic and look and look until you find one. So why rehearse a situation that you won't face in a real game?
3. I also thought, most positions there is no tactic to find--so focusing on tactics too much seems imbalanced when you only see a tactic in a very low percentage of positions you face throughout the game.

Lesson 1: Solving tactics involves a lot of the a-b-c's of what you need to be seeing in every position. You spot all the types of weaknesses. You learn to check out whether they are exploitable or not. tactically, but you also learn to spot them in general, which can be part of becoming able to create them positionally and also avoid creating them in your own positions. As you start facing more difficult tactics, there will be false starts you have to rule out. This is completely applicable to real chess. You also learn how to get more efficient at ruling out different lines by spotting the devastating replies. You learn to spot the blocking move that kills your idea if you do it like that--the in-between move. Of course you also practice visualizing lines but they aren't all totally forced. You have to use skills to justify ignoring other lines without going too deep into them.

The other lessons I have learned have to do with the benefits and drawbacks of different types of tactics modes.

1. For instance on chess.com the very good tactics trainer has comments and a blitz-style system where you can get heavily punished if you don't find the solution very quickly. The counter-intuitive result of a system like this is that the problems end up getting sorted by how quickly people can spot them and get them right. Lower problems in a mode like this are simpler and more guessable--they conform to what guessers expect. In the past I have avoided modes like this because I wanted to learn to really be sure and not guess--guessing can be really toxic over the board in a long game and I don't want to feed that impulse.
2. A different mode that some other sites have, I'll call this mode "Standard," gives unlimited time to solve tactics. I have seen various players advise this approach to getting stronger. Spend as long as you need, 20 minutes, 60 minutes, anything goes, to really get the problem right. I think struggle like this can get you gains. It's not the only kind of gains in town, and it might not always be the best use of your time, but there's something to be said for this approach. Anyways, again the counter-intuitive result of an approach like this is that problems end up getting sorted based on the number of players who truly go really slow and nail problems, vs. the people who give in to guessing after 2 minutes, 5 minutes, etc. Obvious problems and guessy problems still are the lowest, but the way difficulty progresses won't be the same as the "Blitz" pool because you'll get upward into the mountains where the air is thinner and now you're basically playing problems that only the "long think" guys get right. These problems at first are often pretty simple but have "bait" that catches the faster players, e.g. a tempting check that doesn't work, or a wrong way to capture because it misses a mate threat.
3. Another approach I've seen is called "Mixed." At first I really liked this idea--I can't be sure if there is a tactic or not! Ambiguity, just like a real game! That's what I thought. But in truth there still have to be clear forced answers, which is unusual in the grand scheme of things. Just, sometimes the forced answers are still a lost position, or a draw. Not clearly winning, as with typical other tactics modes. So I thought "mixed" would mean this--sometimes you have a regular tactic, but sometimes you have this mushy, just stop his tactic, or find the only move where you aren't dead, kind of stuff. This was only part true. It turns out mixed also has to do with time. It's a hybrid between the above Standard and Blitz approaches as well. It gives you a lot of time--to a point--but if you go beyond that, then it's possible to lose points for being too slow again. So there isn't pressure to be "as fast as possible"," but there's pressure to be faster than a specific, longer amount of time.

Lesson 2: There are two kinds of gains you want to get from tactics--"seeding" gains for the future and "solidification" of your existing agins. People often say it like this: pattern recognition from fast tactics, and calculation and visualization from slow ones. I think this is mostly right. It is right to take the "unlimited" time approach and struggle to "really be sure"--the one that "Standard" mode seems to imply. Doing this gains you will power, visualization, struggle, and you constantly are learning all those little features and skills I mentioned tactics is good for. However, eventually you need to apply that knowledge and those skills in situations where the time you use matters. Maybe playing a game is enough for that. I think of these two modes as "long term gains, research and development" vs. "solidification." When you see a pattern for the first time, struggle hard and work out all the details, and get the answer, that's "seeding" your future growth. Later on you'll solidify it when you use it under more pressure in a game or perhaps in a tactics trainer with a timer running.

I haven't worked out the ideal balance between these two extremes. Sometimes I feel like I can only do one at a time, and later it seems like time to shift to the other one. I might try tackling a new time control and feel everything start to come together. Other times I'm gaining new things and I feel like my play is going backwards a bit but it's somehow "different", like the horizon of some better play.

All the different modes, like different time controls, have different things to offer. For example Blitz tactics doesn't mean you have to play fast--it simply means you will go down if the problem takes longer for you than for the average. The result will be you won't skyrocket upwards to the higher rated problems. Maybe you'll be getting 97% of your problems right, but just not always quickly. I like this idea but in practice when I play "blitz" tactics I feel the pressure to just take the first idea I see and play it as quickly as possible. The different modes rate problems differently and progress you differently.  In mixed as you go upward you start to get defensive problems and shocking positions where you need to keep equality because those are what people miss all the time--but then you start to expect it.

On some sites you can choose between Easy, Normal, Hard--should they serve you problems you have a 70%, 60%, or 50% chance of getting right? There is no science on this as far as I know but in other areas the ideal is taken to be 70% or 85-95%. On the other hand struggling for an hour and nailing it is a totally different--and also valuable--experience.

Tactics traininers sometimes have some downsides or suboptimal features:

1. No scaffolding. Maybe you simply haven't seen the motif before. You'll struggle and struggle, you won't find it (probably) and you'll have to fail the problem. Will you guess? Will you quit? Will you struggle until you finally find a mirage?
2. Even after seeing the solution, maybe it makes no sense. It seems like the real tactic is invisible. The reasons behind the moves are tactics that aren't being played--tactics that we don't see are restricting the choices of both sides.
3. When you "find the answer" and it isn't the same as the problem, you might not see why your answer was wrong.

These tie into my last lesson. Lesson 3: To make our own practice useful we can't just naively use the site (or book) as-is.

There are tactics books (and websites) that offer problems curated by chess teachers that they feel are perfect for learning the various motifs and skills--rather than "natural" problems just grabbed by a computer from master games. Maybe these are more productive gains because they represent human knowledge rather than the "wild" that computer-mined tactics offer us. Another approach may be to make a custom set (some sites let you do this) where you narrow the range of problems you're facing to work on specific motifs, problem difficulty ranges, etc. you want to focus on. Then you can gradually work your way up to facing the normal tactics trainer and be stronger for it. Scaffolding.

To really learn from those annoying problems where you "don't get it" or "don't see why I can't also do this" it can be a chore. It can be harder than solving more problems, but it's important to do it. On a web browser this might be easy--open analysis in a new window or tab, play around with the computer and find out what happens if they don't do such and such. Recently I saw a guy give some really good advice on this. He said to spend 2 minutes trying to find the refutation yourself. Good advice, but what if it will take me 20 minutes? Following up on the problems where there's something you don't understand has a lot of value. Otherwise you're just guessing through problems, maybe reinforcing bad habits like giving up too early or playing hunches--just doing a numbers game mining problems that work for this approach, but the most valuable thing--knowing why you miss things--is being skipped over.

I use a "training journal" to try to motivate myself. In it I write down what I did. It forces me to put insights into words which is a good way of checking myself. I find naming my mistakes particularly useful. For example I tend to go down the "rabbit hole" of calculating a variation right at the start of a problem when I haven't looked enough at the position yet. I waste time looking at some capture, for example, some potentially forcing line--but something far more important, like a pin, or some threat elsewhere, or three important pieces all on the same diagonal, went unnoticed. I waste 5 minutes looking at nonsense when I should have seen those features before going down any rabbit holes. Writing down why I missed a problem helps me gain these kinds of insights into where I'm going wrong and giving them a name. Another typical problem I have is making a capture without looking at a better way to also make that capture. Sometimes there's 3 ways to do something, I look at 2, pick one, and I could have looked harder and realized there was a 3rd, but I didn't. Once aware of these things I can push myself to practice a new approach when I begin a problem, to make sure I don't do these anymore.

It would be nice if tactics books and websites integrated the perfect motivational features and learning features, but until then, we have to use whatever techniques we can to bridge the gap.