Chess players use analysis primarily to determine what might happen if a certain move is played. This creates a “tree” of analysis with moves branching at each ply. However, the number of legal possibilities is very large, so humans “prune” their tree of moves to only include ones they think might be (at least at first glance) reasonable. If this is not done, then the tree can grow astronomically! For example, if each side has 30 legal moves, then to look at even 2 moves deep (4 ply) would require 30 x 30 x 30 x 30 sequences, or 810,000! Hardly possible for a human in a year, much less in a game requiring 40 moves in two hours. Similarly, pruning the tree is necessary even for computers, which use full-width search to examine all moves, but alpha-beta cutoffs to avoid looking at the entire tree. Of course a discussion of computer tree-searching is outside the scope of this article!

One way to help categorize what happens when a human prunes his analysis tree is to break the type of moves into different categories:

Checks

Captures

Threats

Moves that increase your piece's activity or decrease the opponent's pieces activity

Forced moves, like only legal moves, or necessary recaptures (note that forced moves are almost the opposite of forcing moves like #1-3, although it is possible for a forced move to be a forcing move!).

Other moves

One way to control the tree – and, happily, to find the best move – is to try and maximize the number of opponent's moves that fall under category #5 – in other words, the tree is not only smaller, but more under your control if your opponent has to play a forced move and does not have the flexibility to perform a more constructive purpose. If you can consistently force your opponent to reply to your move we call this pleasant situation having the initiative.

For this reason, when you are searching for your best move, or considering your opponent's best reply, you want to start with (or assume) the most forcing moves, which are almost always checks, captures, and threats! Note that while most checks are more forcing than most captures, and most captures are more forcing than most threats, this is not always the correct order of how forcing your potential moves may be. For example, a threat to mate-in-one on the next move is often more forcing than any capture, and even some checks! So when searching for forcing moves, always consider mate-in-one threats near the top!

I would like to point out that most students, while understanding intuitively what a threat is, cannot define it – which usually means they cannot easily find all threats!

A threat is a move by player A which, if nothing is done by player B in response, can do something positive on the next move for player A.

By “positive” we mean win material, checkmate, or even gain some positional feature like control of a file or wrecking the opponent's pawn structure. But in order for a move to really be a threat, the execution of the threat next move, if allowed, should be meaningful. For example, if you are behind a Rook and “threaten” to win a pawn but in doing so you have to trade Queens, winning the pawn may not really be a threat since the net result may be positive for your opponent, not you!

In the previous Thinking Cap we mentioned that a good thinking process requires you to look to see if your candidate move is refutable before considering it further, much less playing it. At a minimum, this usually means you must say to yourself,

“Suppose I make this move, then in response what are all my opponent's most forcing moves, and can I safely meet all of them?”

At this point you consider each of your opponent's checks, captures, and threats, making sure you have an adequate answer to each. If you do, then the move is plausible and it may remain a candidate move.

Remember, not all forcing moves are good moves, but some bad-looking moves are often actually good moves because of their forcing nature. Let us consider two examples. The first is problem #225 in the first edition of John Bain's excellent book on beginning tactical motifs, Chess Tactics for Students:

This is White to play and win. The answer, if you don't know, is 1.Rxe7 Kxe7 2.Bb4+ winning a piece. But for the moment let's ask a different question: Excluding checks and captures (which sometimes can also produce threats), how many moves does White have which are threats in this position? In this case we want immediate threats, not long-term threats such as to create an outside passed pawn. Hint: not all threats are good moves; they just have to be legitimate threats. To find a threat, make a move, skip the opponent's reply, and then move again and see if you can gain something.

The answer is 5. They are 1.Ba6 and 1.Bc6 (they both threaten 2.Bxb7), 1.Bb4 (threatens to capture on e7), while 1.Ba5 and 1.Be5 threaten the removal of the guard tactic 2.Bxc7+ Kxc7 3.Rxe7+. On the other hand, 1.g4 does not strongly threaten 2.g5 since after 2.g5 Nh5 holds g7 for a while. Notice 1.Ba6, 1.Bc6, and 1.Bb4 are terrible moves that just allow a capture, but they are legitimate threats. But not all terrible-looking threats are actually terrible and you need to differentiate!

Consider the second example, the famous position Bernstein-Capablanca 1914, with Black to play:

Black's move 1…Qb2! at first looks like a typographical error. However, it not only contains two strong threats, but the combination of both threats is unstoppable. White of course cannot play 2.Qxb2 due to 2...Rd1 mate, but White must stop both of the threats: 2…Qxe2 and 2…Qxc3. White resigned, but to show what might have happened, suppose White had played the tricky 2.Rc8. Then Black would play 2…Qa1+ 3.Qf1 Qxf1+ and then win the Rook. Of course not 2…Qxe2?? 3.Rxc8 mate, nor 2…Rxc8?? when 3.Qxb2 stops the mate. Or White can try 2.Rc2, but then 2…Qb1+ 3.Qf1 Qxc2 wins. Or 2.Qe1 Qxc3!.

From these examples we can see that if you have the time, it pays to consider all forcing moves, including all threats; chronologically, these are: anything your opponent may be forcing from his previous move, your current forcing moves, plus (importantly) his possible forcing moves in response to your candidate moves. If you have the time, even consider the ones that may look silly at first.

If, in looking for forcing replies to your candidate move, you do find a threat by your opponent on the next move that you cannot meet on your following move, it is possible that such an opponent's move may refute many of your possible candidates. For example, suppose in some imaginary position you say to yourself, “If I play 1.Qd2, then suppose he plays 1…Qh3 and threatens 2…Qg2 mate. How can I stop this? I cannot, so I cannot play 1.Qd2.” But if this is true, then possibly 1…Qh3 might force checkmate after many of your possible first moves, for example 1.Rb1. So you say to yourself, “Suppose I play 1.Rb1, then does 1…Qh3 still checkmate me?” A move like 1…Qh3 is called a killer move by programmers, and they will use this as the first reply to all candidates in order to efficiently cut off their analysis as quickly as possible. You can do the same thing, so that your thought process then includes something similar to “What moves can I play which would stop 1…Qh3 and 2…Qg2 mate?” Thus your search is made more efficient by only considering moves which prevent the killer move.

In summary, when you are analyzing a game and looking for continuations, the ones that should most concern you are the forcing moves: checks, captures, and threats. On offense, if you can successfully continue to play moves of this type, you have the initiative. Defensively, if you can successfully identify and plan to meet forcing moves by your opponent (or avoid them altogether!), then you will not be surprised and can usually stay in the game and have good chances to win – or at least draw – against even formidable opponents. Therefore, learning to identify and deal with forcing moves is an important part of becoming proficient at chess analysis!