2025/06/15: "No, That Wasn't A Blunder"
The following type of post shows up frequently in the Comments section of the Daily Puzzle: "The opponent blundered by playing X and therefore the puzzle is flawed."
This is incorrect.
The statement should read "The opponent played a sub-optimal move X but that's OK because puzzles allow for sub-optimals."
There are 2 issues: was move X a blunder and what does that imply as to the soundness of the puzzle?
.
What is a blunder? There are many definitions. Here's one: "A blunder alters the projected outcome." A win becomes a draw or a loss or a draw becomes a loss. In other words, the projected outcome is worse than if the blunder hadn't been made.
Think of it in terms of the standard 3 phases of matter [I'm ignoring plasma]: gas, liquid, and solid. Consider water: it can be gaseous, liquid, or ice.
A blunder lowers the temperature and alters the phase from gas to liquid, gas to ice, or liquid to ice.
With me so far?
Now some examples:
.
What is the projected outcome? White is winning. The engine eval is > +7, a huge advantage. But if White plays 1. Rxd5??, that's a blunder because it allows Black to win with 1. ... Rc1+ 2. Rd1 Rxd1#.
A win becomes a loss ---> blunder
Air becomes solid ice ---> phase change
.
This is a draw. But if White plays 1. Kd5??, Black wins with 1. ... Rd8+, skewering the King.
A draw becomes a loss ---> blunder
Liquid becomes solid ---> phase change
.
What is the projected outcome? White is losing [the water is solid]: the engine eval is -4.19 at depth 25 [whether this is "enough" to be considered losing is a subsequent question].
This eval is dependent on White finding the optimal 1. Qg3, blocking check and allowing Black to win a Queen for a Rook.
In the puzzle, the creator chose 1. fxg5, which allows a forced -M4 [1. ... Bxh2+ 2. Kh1 Bg3^+ [discovered check] 3. Kg1 Qh2+ 4. Kf1 Qxf2#].
Some conclude that 1. fxg5 must have been a blunder because White got checkmated but after 1. Qg3, there was no checkmate.
But did the projected outcome change? No: White was losing before he moved and he was losing after he moved. The water did not change from gas to liquid or ice. It was ice before and it was ice after.
.
1. fxg5 was a mistake:
The interface has 4 classes of errors. In order of increasing severity: inaccuracy, mistake, miss, and blunder.
What's the difference? A change in projected outcome [or a change in the phase of the water]. The interface considers 1. fxg5 to be a mistake, not a blunder, because the projected outcome before the move was a loss for White and the projected outcome after the move was STILL a loss for White.
The fact that the loss changed from a finite eval to a forced checkmate does not make it a blunder, just like lowering the temperature from 20 degrees F to 0 degrees F does not change the phase of the water: it's STILL ice.
In order for 1. fxg5 to be considered a blunder, White would have had to have been winning or drawing before the move and that's not the case here.
.
A subsequent point was brought up: would any given Solver be able to convert the position into a win if White played the optimal 1. Qg3?
The answer is that it doesn't matter: the puzzle is judged on the eval when the puzzle is over, not whether any given Solver can actually win the subsequent game.
This is how it should be; otherwise, the puzzle will be judged sound if Magnus Carlsen solved it [because he can easily win the subsequent game] and unsound if a beginner solved it [because he cannot]. That's completely dependent on the Solver so puzzles aren't judged that way.
.
Finally, puzzles allow for sub-optimal opponent moves. The primary purpose of the puzzle is to showcase some idea and this is sometimes best done by using opponent sub-optimals. It gives the puzzle creator more latitude.
For example, had the optimal 1. Qg3 been played, the beautiful -M4 would have not been shown. Since the creator wanted to show it, 1. fxg5 was chosen instead.
Here's some reference material on that subject:
https://www.chess.com/blog/EnPassantFork/no-the-puzzle-is-not-wrong