Productive tactical puzzles: random or themed?

Question

Simply, which of the following two approaches will typically improve tactical ability the fastest:

1. Doing puzzles around a known theme or tactical motif - i.e. do 100 puzzles that involve a discovered attack, or a skewer, or mate in 3, mate in 2 etc. etc.
2. Doing random puzzles where you don't know the theme or motif behind the puzzle. You may even throw in a number of puzzles without a tactical solution to simply force you think about the next best move.

I always assumed number 2 would be better, but is there any research/authoritative anecdotes/good reasoning that show it is better to do 1?

Answer 1

In my experience if you are starting out it is better to solve same thematic tactical problems in a series and slowly moving onto other themes. After you get accustomed with the themes random problems or problems which doesn't tell you what to do helps.

Answer 2

I had the same question when I thought about tactical improvement. What seems to be working for me is an "inclusive" approach. Don't stick to one method, as each method seems to be helping in different ways. This applies to degree of difficulty of chess problems as well. At one point in time, I could solve tactics problems rated 1800+ on lichess very easily only to stumble on problems rated 1500 and below. So a mix of all seems to be working well for me.