When I begin to instruct a student, one of the pertinent questions I ask is “What is your main goal each move?” Sometimes I get reasonable, but erroneous answers like:

“I am trying to win” or
“I am trying to make my position better.”

…but the correct answer is “I am trying the play the best move” or “I am trying to play the best move I can find, given the time constraints.”

But the student is not off the hook yet! The obvious follow-up question is “OK, but then what makes one move better than another?” After all, if you are trying to find the best move, a criteria is that you have to know what makes one move better than another. We touched upon this question in The Thinking Cap #2, “The King of the Hill”, but now we are going to focus on exactly what this question pertains. And make no mistake – this is important, for the question “What makes one move better than another?” is at the very heart of what it means to play chess!

First, I would like to eliminate the plausible but out-of-bounds answer “The move that best follows the correct plan.” While this may be an excellent way to help choose your candidate moves, at this point I would rather deal more with the theory of best moves rather than the practice.

Again, let’s start with some other common (but not correct) answers:

1. “It leads to a win of something”
2. “It does more.”
3. “The position after the one move is better than the position after the other.”

This last answer is almost correct! In fact, in theory it is correct, but for humans it is usually impossible to apply (as will be shown below), so it won’t do. However, this third answer does address the fact that we evaluate positions rather than moves, so if one move is better than another, that means it must lead to a position that is better than another, but which positions should we use?

The problem with using the position immediately after the candidate move is that the position may be clearly non-quiescent. For example, suppose you compare a “nothing” move like Kh1 with a move like Qxh7+, capturing the h-pawn. It may seem better to win a pawn with check than just move the King, but what if your opponent’s next move is Kxh7, winning the Queen for the aforementioned pawn? So we must look further, to see what happens after Qxh7+, or else the evaluation that we are ahead a pawn is meaningless.

Generally, we stop our analysis at a point where we can evaluate as follows:

1.  The position is quiescent – there are no more meaningful checks, captures, or threats which would change the evaluation,
2. We use our judgment – for example we sacrifice a piece to expose the enemy King and even though we cannot practically analyze to quiescence, we judge whether the exposed King is worth the sacrifice, or
3. A sacrifice fails since the further possibilities cannot possibly give us a return equal to or greater than our sacrifice. For example, if you sacrifice your Queen and then later you might possibly win back a Rook, no sense analyzing further to see whether that is true, since either way you would reject that sacrifice.

The first of these is the most common.

However, even knowing where to stop analyzing a line still leaves us with the question: With a large tree branching from each move, which positions are the ones we want to evaluate? In other words, which branches of the tree are meaningful? The answer is that we must assume – to the best of our ability to judge – the best moves for each side:

A move is only as good as the positions that will be reached from it, assuming best moves for both sides!

Of course, if you are not good at judging what the best moves are during analysis, you will arrive at the wrong positions and reach the wrong conclusions when you evaluate. This weakness, along with an inability to accurately evaluate positions with approximately equal material and the inability to recognize quickly and accurately the common tactical motifs, are the three biggest “thinking” problems of weaker players!

So now we have our answer. Suppose we have moves A and B. Then we try to determine the best sequence after A and B, only going as deep as necessary to evaluate, for example:

My move A followed by his best move A’ followed by my best move A’’ => A – A’ – A’’ vs.
My move B followed by his best move B’ followed by my best move B’’, his best move B’’’ and my B’’’’ => B – B’- B’’ – B’’’ – B’’’’

This leads to some position after A’’ we feel we can evaluate. Let’s call this position A* and the evaluation of this position E (A*). The same holds for B: the position after B’’’’ we call B* and we evaluate it as E (B*). Then we compare the evaluations (good players spend a great deal of time doing this when they are thinking!) and decide which position we like better. If E (A*) is superior to E (B*) then we like move A better!

Then we apply the King of the Hill method discussed in The Thinking Cap #2, and apply this to all our candidate moves in order to find the best one.

There are shortcuts to this method (depending upon the position), and good players don’t consciously go through all this, but this is basically what should be happening when you think, or else you will end up with a less than optimum move. For example, many beginners don’t even follow the guideline When you see a good move, look for a better one – you are trying to find the best one!

Finally, let’s see how this comes out when a computer does it. Did you ever see the analysis window in Fritz, where it defaults to show three lines? The first line, using the example from The Thinking Cap #2, shows something like:

11 ply (1/42) +0.83 19.Rhe1 d5 20.cxd5 exd5 21.Bh4 …

This means that after 11 ply (assuming quiescence and search extension, etc.) Fritz has evaluated the final position to be 0.83 pawns better for White, and the best move sequence follows. Since playing the best move keeps the current evaluation from the previous ply, (by game theory, your position cannot get better – or worse – if you make the best move in a game like chess), Fritz assigns the value from the final position back to the original move! So it judges 19.Rhe1 to be “+0.83” and, since it has already ordered all the moves and placed this line at the top of the list of lines, then 19.Rhe1 at 11 ply is thought to be the best move, and the second best move, with a lower evaluation, will be on the next line. The top line is called the Principal Variation, or PV.

So next time you are trying to find the best move, try to keep in mind that what you are really trying to do: Find the likely optimum position(s) that arise from this move and compare it to similar optimum positions that arise from each other candidate moves. The move that, in your judgment, leads by best play to the position you like the most is the one you should generally play. Of course, doing this in practice is difficult (and many books are filled with practical advice on doing so), but at least it helps to start by understanding what should be theoretically happening. I hope laying this groundwork helps!