Last month we learned about Game Analysis, which is based on an advanced analysis algorithm. This month we’ll examine Find Blunders, which is a different approach to automatic game analysis in Rybka Aquarium. About Aquarium’s Find Blunders Find Blunders is intended as a quick search for critical blunders in a game.

Compared to Game Analysis, the main difference is that Find Blunders only does what the name indicates and skips all other types of annotations. Additionally, because of its narrower focus, the analysis algorithm itself is not as refined as the one used by Game Analysis.

Find Blunders is a good choice for:

• A quick overview of major turning points in a game.
• Analyzing fast games looking for tactical oversights.
• Analyzing games of beginners and lower rated players.

Running Find Blunders If you are viewing a game when you start Find Blunders, only that game will be analyzed. If you want to analyze multiple games, open a list of games and then start the analysis. It works exactly the same way as Game Analysis, as explained last month. As with all other types of Aquarium analysis, Find Blunders is started from the Analysis tab.
	Find Blunders is started from the Analysis tab.

The Find Blunders button is a split button. Clicking the downward pointing triangle will drop down a menu allowing you to set the analysis options as shown in the next image. Clicking somewhere above the triangle will start Find Blunders using the current options.

Clicking Options on the drop-down menu opens a dialog box that allows you to set all the options for Find Blunders. At the top of the window there is a drop-down list (“Engine”) with all installed chess engines. Select the one you want to use for the analysis. In this example Rybka 3 1-cpu has been selected. Below the engine there are two panes: “First pass” and “Second pass (verification)” where you specify the analysis parameters. The first pass analyzes every position in the game after the opening moves. The second pass does additional analysis on moves that seem to be blunders based on the results of the first pass. The definition of a “blunder” depends on the settings in the “Commenting options” pane, which is described below.

Find Blunders settings.

	Both “First pass” and “Second pass” provide you with a flexible way of specifying how deeply each position is analyzed. This method of specifying analysis time and depth is common in Aquarium. The following examples explain how it works.
Analysis time and depth.

Using AND. You can connect the time and depth/plies conditions either with OR or AND. If you select AND, the analysis runs until both conditions are fulfilled. Using the example shown in the image above, the analysis would run for 2 seconds. If depth 7 was not reached, then analysis would continue.

Using OR. If you select OR the analysis runs until either condition is met. Let’s say that we had specified “2 seconds OR 7 plies.” The analysis will run for at most 2 seconds. If depth 7 is reached earlier the analysis will stop, even it the 2 seconds haven’t passed yet.

Time only. If you want the analysis to run for a certain length of time, for instance 10 seconds, use “10 seconds AND 1 plies.” This ensures that the analysis runs for at least 10 seconds, but since the engine will have reached depth 1 long before that, both conditions will always be met after exactly 10 seconds.

Depth only. If you want the analysis to run until it has reached a certain depth, set the time to a low value; for instance, “10 seconds AND 15 plies”.

In addition to setting the time and depth, you can select “Thorough blunder check” for the second pass, in which case the following analysis is performed for each potential blunder:

1. The position before the blunder is analyzed.
2. The position after the blunder is analyzed.
3. The position after the best move suggested by the first pass is analyzed.

If you don’t select “Thorough blunder check,” only the third step is performed. This of course takes less time, but may give you less accurate evaluations. You should choose longer/deeper analysis for the second pass than the first pass. Otherwise the second pass would be of little value in verifying and deepening the analysis. The “Commenting options” pane let’s you define what constitutes a blunder and rules for adding the analysis results as annotations to the game.

Blunder definition and rules for annotations.

Minimum delta (centipawns) defines when a move is considered a mistake based on the absolute size of the blunder. If it is set to 100 centipawns, then a move is not considered a mistake unless it loses a pawn (or equivalent). The higher you set this value, the more serious the mistake must be to be considered by Find Blunders. In the screenshot above it is set to 30 centipawns, which may be a good choice if you are an intermediate player.

Minimum delta (percents) defines when a move is considered a mistake based on the relative size of the blunder. It measures to what extent the blunder is likely to affect the winning chances. Losing a pawn in an equal position can seriously affect the outcome of a game, but the loss of a pawn in a completely lost position makes little difference. This parameter causes Find Blunders to skip mistakes that have little or no effect on the outcome, allowing you to focus on the moves that really mattered.

A move is considered a mistake when it meets the conditions of both Minimum delta parameters.

Maximum variation length specifies the maximum length of variations copied from the analysis into the notation.

The variations produced by chess engines can be quite long and the first moves of a variation are usually more accurate than later moves. The Untrusted engine halfmoves specifies how many moves should be cut off the end of the engine variations.

Like other Aquarium analysis methods, Find Blunders saves all its analysis to disk while it is running. This means that your analysis isn’t lost even if your computer shuts down while Find Blunders is running. When you resume the analysis, it continues from where it left off. In some cases, however, you may want to repeat the analysis from the start and discard previous results. In that case select “Discard previous analysis”.

You should always select “Discard previous analysis” if you have analyzed a game and want to analyze it again with different parameter settings. This applies even if you interrupted the first analysis before it finished.

After setting all the parameters for Find Blunders you can “Start” the analysis or “Save” the new settings. Clicking “Cancel” discards any changes you have made. While Find Blunders is running you will see this status window: Here you can see the progress of the analysis. We see that Current stage is “1” meaning that the analysis doing the first pass through the game. The analysis of 27 positions has been completed (Positions analysed) and 22 positions are waiting to be analyzed in this pass (Positions left). The Estimated time left tells us how much time is remaining to complete this pass.

The Find Blunders status window.

The Results of Game Analysis When Find Blunders finishes its analysis, the commentary is automatically added to the game as shown in the following example: Let’s use White’s 21st move to explain what Find Blunders is telling us. The move is marked with a question mark, which means that it is a blunder (remember that you can define what constitutes a blunder). The evaluation of the position after 21.Re6? is +0.00, which means that the position is equal. The variation that follows the move tells us that 21.Nde4 would have been much better. It would have given White a nice advantage of more than 1 pawn (+1.07).

Rybka Aquarium is a very powerful analysis tool and with this column we conclude the basic description of its most commonly used analysis methods. Previous columns have covered:

• Interactive Deep Analysis (IDeA)
• Infinite Analysis
• Game Analysis

All these analysis methods deserve a deeper look. Even the “simple” Infinite Analysis has many interesting options that could not be covered in a single article. And IDeA, in spite of simple options, is a truly deep method that can be used in so many different ways that there is endless room for creativity and improvement.

Next month I will demonstrate how a high-level game can be analyzed using Aquarium’s analysis tools.

Dadi Jonsson.