In 1970, grandmaster and world-renowned trainer Alexander Kotov published his most famous book, "Think Like a Grandmaster". In this book, Kotov introduced the concept of candidate moves.
One of the first things that you learn in chess is the importance of considering multipled options. The main cause of mistakes is not considering something, so naturally, it's helpful to consider more options. Some trainers say you should always have at least 3 candidate moves. This is the most basic level of candidate moves: you come up with multiple ideas. Instead of a single variation, you now calculate 3 variations!
Now, there's an obvious problem here. You can compare and contrast your three different ideas, but what if you miss your opponent's reply? You need to consider your opponent's responses as well. That's 9 variations... not that bad, right?
But wait, you need to consider different options on every single move. In non-critical positions, the recommended depth of calculation is 2.5 moves, or 5 ply (why 2.5 and not 3? because you want to end on your move). If we have 3 candidate moves on every single move, we have a total of 3^5 = 243 variations. That's ridiculous, and we aren't even considering sharp positions where you have to calculate until the position is clear. Unless you are a computer, you cannot handle that. In fact, many books in more recent years have criticized Kotov's system for being ridiculous and inhuman. The first to make a new system was Jon Nunn. How should I explain his system?
If you are familiar with computer algorithms, then this will make sense to you. Kotov utilized a sort of depth-first searching algorithm, going all the way to the end of each line and then on to the next. John Nunn's idea ("Secrets of Chess Calculation", or "Secrets of Attacking Chess", I forgot the name) was to use breadth-first searching instead; essentially, he advised players to look a little bit into one line, then switch to another, all the while comparing the two. This idea is widely accepted by the chess community to be the best way to approach calculation. Funnily enough, though, it will take just as long as Kotov's method, unless you end your lines early, which is a great way to make blunders. Even more recently, books have criticized Jon Nunn as well, perhaps seeing that his ideas are not so effective, although these books (For example, "Move First, Think Later", by Willy Hendriks) are not very popular. So then, what do I suggest?
First of all, a disclaimer. You have to figure out what works for you. I'm sure Kotov had great success with his ideas, and Jon Nunn with his, since they became grandmasters. However, it doesn't work for everyone, and people who write books often fall into the trap of justification, imposing their ideals on situations. Simply put, most book writers are pretty biased (or they just want to make more money, and suspicious generalizations sell well).
Personally, I think that your calculation needs to have a goal, a direction. The ideas should always come before the moves. At first, you may be skeptical, since you probably think in moves, not in plans (your first thought is probably a move instead of an idea). However, your subconcious is guiding your consicous with ideas. While the subconcious mind is a powerful thing, rational thought is needed for chess, so you should make an effort to actively think about ideas. Pick out a few ideas, and only look at moves that help your ideas along (note that I am not telling you to ignore your opponent; your ideas can be prophylactic), as well as forcing moves. Never ignore forcing moves, no matter how stupid; you will either overrule them in seconds or discover an amazing possibility, and they are easy to calculate becuase they limit the opponent's options. With your search for candidate moves narrowed down to a few specific ideas, the process should be much easier and faster. However, we still have to consider all of our opponent's possibilities. The issue with this system is that if you miss an idea, you won't be finding it. Therefore, instead of coming up with a specific number of ideas, come up with as many logical ideas as possible.
This is by no means an optimal system, but it's what works for me. How do you guys structure your calculation? I'm very interested to know since I might eventually write a book about the thought process...