Studying many positions in which an error has been made helps develop this understanding. Consider the following position with White to move.
Black has just played 6...b6? This position occurred in Rubinstein -- Heilmann, Barmen 1905. I found it in Akiba Rubinstein: Uncrowned King (1994). The authors, John Donaldson and Nikolay Minev, identify the move as an error. I have not played against it, but had Black with the position before the error in one internet blitz game--I played 6...cxd4, which is the second most popular move. I lost that game, but not due to the opening. Rather, I blundered away a knight on move 25. The main line is 6...a6. Both my choice and the main line are presented in the Encyclopedia of Chess Openings as offering Black good prospects to achieve equality.
Knowing the main line is no assurance that one will recognize errors. A deviation from a main line is not always an error. It could be an improvement. In many cases, deviations are simply another way to play a given position. There are more than 85 billion possible legal chess positions after the fifth move. The majority must be quite bad for one player or the other, but certainly there are many good positions that have yet to be examined.
Minev has faced this error over the board twice.
From the position after White's 6.a3, twice Vassily Ivanchuk has played 6...dxc4, drawing both. That move and 6...Be7 are given in footnotes in ECO. Bobby Fischer played 6...Ne4 during his Candidates match with Tigran Petrosian in 1971. 6...Ne4 is the first line in ECO, but is presented as favoring White slightly. 6...b6 is the most popular move that is not in ECO. The Chess Base online database turns up 108 games when 6...b6 was played. White scores over 75% in these games.
What is wrong with 6...b6 in the symmetrical variation of the Queen's Gambit Declined, Semi-Tarrasch?
It weakens the a4-e8 diagonal, exposing the c6 knight to a pin. Does this pin allow White to gain material or positional advantage? In Akiba Rubinstein: Uncrowned King, the authors give the game Minev -- Morcken, Moscow 1956 with the line 7.cxd5 Nxd5 8.Bb5 Bb7 9.Ne5 Rc8 10.Qa4. "White wins at least a pawn," they observe (21).
Other moves have been played, but other moves fail to exploit the weakness on c6. 7.dxc5 might have merits, as after 7...bxc5 White plays 8.cxd5.
Black has two replies: 7...Nxd5 and 7...exd5.
Studying the games that have been played by strong players following both replies offers a method for developing the skills to exploit this error. While trying to show the Rubinstein -- Heilmann game to a friend on Saturday, I forgot about Rubinstein's 9.e4!
Rubinstein -- Heilmann continued:
7...exd5 8.Bb5 Qd6 9.e4
Black to move
Heilmann quickly found his pieces lacking mobility and the game ended ten moves later. Most other efforts by Black fared just as poorly.