Has anyone ever combined the tactics by rating and the motif list you mentioned? That way you could have 5 of the same motif and then just randomly test 1 from each group. Then, cycle through again?
That's not enough training. The way you speak about tactics implies that major improvement can happen from solving a few hundred puzzles. You need to do a lot more than that.
wow, the three videos on that channel that equal up to 10 minutes of content made me grow by almost 500 rating points in less than a day!
Why not h3? In hindsight, looking back if I thought of Qh4, I would have played h3.
At a glance, 1.h3 Nxf2! 2.Rxf2 Qh4 3.g3 Qxh3 where Black can play ...Bxf2+ whenever he wants with a material advantage and the initiative. I've solved countless puzzles on this theme.
Why not Ng3 liberating white's queen in the process?
1.Ng3? Qh4 2.h3 Nxf2 3.Rxf2 Qxg3. These sorts of tactics should be seen very quickly or even automatically when thinking about 1.Ng3.
If you have to use logic to decide between these three candidates rather than a rapid calculation that takes a fraction of a minute, you are going to blunder a lot of tactics in your games or use all of your time checking the basics.
The example you provided was essentially 100% tactics even though this was not your intended purpose and it is clear that you need to improve your pattern recognition.
You can intake just so much at one time.
I see no reason to believe that this is true. I remember doing 7.5 hrs of puzzle rush in 1 day a couple of months ago, and my metrics suggest that I improved faster per hour doing more tactics than less.
The only possible issues are attention span and energy. Sufficient sleep and food can help with energy, and attention span can be trained.
You can make custom sets on chesstempo if you are premium. Chesstempo allow you to cycle through them and has a variety of settings to suit your preferences.
How are you supposed to see Bf4 and then Bg3 when you have a long diagonal from a1 to h8 open? Why put the bishop on an opposing diagonal not pointing towards your enemy's king? If blocking the e1 to h4 diagonal on g3 is needed, then why not Ng3 liberating white's queen in the process?
These positional considerations are only relevant when there are no tactics.
'A knowledge of tactics is the foundation of positional play. This is a rule which stood its test in chess history and one which we can not impress forcibly enough upon the young chess player. A beginner should avoid the Queen's Gambit and French Defense and play open games instead! While he may not win as many games at first, he will in the long run be amply compensated by acquiring a thorough knowledge of the game.' - GM Richard Reti
It would have been 1. h3 kicking the knight out or forcing a Nxf2. Apparently, this is what Stockfish likes but I don't understand why giving up 3 for 1 is better here. If the bishop takes, then it's at worst 6 for 6.
...Qh4 will give Black the initiative, Black will release the tension with ...Bxf2+ on his own terms (later after improving the circumstances), and the White king will be exposed.
You keep bringing up general considerations with shallow variations. These generalities are only relevant when you aren't missing concrete details. If you are consistently missing basic tactics within the first few moves, it is clear what you need to work on.
What are the patterns?
Primary Motif: Bc5's x-ray against Kg1
Possible reason for tactic: favorable assault ratio on the kingside.
Weaknesses: dark-squares (primarily h2 and f2).
Threat: ...Qh4
All of the above should be registered in seconds at most.
1.Bf4 is essentially automatic for me because I detect the weaknesses on sight and 1.Bf4 is the most active way to solve the problem.
1.Ng3 Qh4 2.h3 Nxf2 3.Rxf2 Qxh3 is a pattern. I've seen many similar examples. There are similar patterns the other forcing variations, but essentially there is a geometrically standard response in forcing variations that should be automatically detected and quickly visualized.
Trying to verbally and logically understand every pattern is not efficient enough. Almost all of pattern recognition improvement is non-verbal, though naming patterns and motifs is helpful.
You should solve a huge amount of puzzles, hopefully organized by theme (either when solving an organized set or subconsciously after having automated a lot of archetypes), and get faster and faster until the basic patterns are automated.
Chessable as a way of training tactics is especially helpful for automation due to spaced repetition, but adults need absurd volume either way. Judging by the fact that you are using logic to try to understand pattern recognition, I assume that you are an adult. I highly recommend that you read this article.
Elijah's lessons were very insightful; my play improved quickly when I was taking lessons with Elijah (his heavy emphasis on tactics was more than enough for me to improve significantly). I would definitely recommend the lessons if you would like to improve, especially in a tactical aspect.
I also question that list.
It's not meant to be all inclusive. You can't learn tactics from only one source.