The scope and limits of tactical classifications

Before discussing the topic of the post, Knights Errant, quixotic blundering fools feeling our way in the night trying to improve at tactics, time to make room in the fire circle for allegroknight. May Caissa smile upon you. He's already got about five posts up. Please give him a warm welcome to the blogosphere.

Also, I've been thinking a little about how to classify tactics. We all know that rather than simply crunching through variations in one's head, it is helpful to construct a higher-level narrative about the position based on more general principles (e.g., see Temposchluckers's recent post for such an example with bishops of opposite color). Many psychologists would describe this as constructing higher-level cognitive chunks to describe the position at a higher level of generality than the flat-footed 'this piece is here, that piece is there' variety. Indeed, this is what most positional evaluation treatises are supplying, general principles that can be used to evaluate a position to see what weaknesses need to be addressed or exploited.

This also holds in tactics: it is very useful to have the concept of a 'fork', for instance, to classify the multifarious forks that occur in real games. This is not just for aesthetic pleasure, but can be used to provide top-down cognitive feedback in the move selection process. This is a process I call guided pattern recognition, introduced here, and see a non-chess example well-known in the psychology literature here. For example, when I think 'Is there a fork here?', a fork will often pop out at me that was previously invisible, sort of like when someone pops out in a crowd when someone says 'Look for someone in a lime-colored shirt.' (Cognitively, this is likely a top-down attentional affect, as it is well-known that we can selectively attend to particular features of a scene, over and above the previous anemic view of attention in psychology as a little beam or spotlight that is used to only pick out all the features in a spatially focused region of the visual field: rather it seems we also have a spatially extended attentional mechanism that can select individual features anywhere in the field). Helping us with pattern-searching is just one possible benefit of having a good way to classify tactics. It is also useful in discussing games with others ('Then I forked his queen and rook with a pawn': everyone knows what you mean and you don't need to give any specifics of the position).

In the previous post on guided pattern recognition, I listed the procedure I use to search for tactics. In particular, I scan the scene partly guided by a short list of tactical patterns. The problem is, many tactics I am starting to see in CTB don't neatly fall into any of the categories. This is partly a result of deficits in my classification scheme. The challenge is to find a classification system that is exhaustive but practical. This topic has been discussed fairly extensively by King of the Spill and Temposchlucker in two important posts from about a year and a half ago.

One problem with trying to classify everything is that I become somewhat blind to tactics that don't fit into my schema.For example, here is a problem that I ran into today at CTB, with white to move:

I'll put the solution in the comments in the next 24. I had a really hard time with this problem, partly because it didn't fit neatly into my tactical classification scheme. Part of my problem is that my classification scheme is really only for very simple tactics (fork, skewer, pin), whereas this problem is truly a combination that relies on seeing not only the ultimate goal, but also being imaginative enough to see how to reach it.^{See Footnote} I guess that's the thing with brilliancies: they aren't obvious. If I could quickly find combinations like this in real games, man, I'd quit the Circles!

Is finding an exhaustive and practical classification scheme possible? Is it worth the effort? There seems to be a consensus that thinking-via-principles is very helpful for strategic thinking, which generally relies less on variation-crunching. But my hunch is that for sharp, tactical positions, it may be that aside from the basic tactical alphabet (forks, pins, skweres, etc), the Caissa smiles mostly upon concrete analysis of variations. With more than 1-3 move tactics, things get too messy and the simple categories can actually cloud the precise analytical processes needed to find answers. I guess I should revisit King of the Spill and Tempo's posts to help me think more about this.

[^Footnote While such combinations may not arise a whole lot in real games, when they do it would be really sweet to see them. But, then again, I lose most games because I miss 1-3 move tactics. One nice thing about many of these multi-move combinations in Stage 5 of CTB is that little kernels of the solution have already appeared in previous Stages of the problem (e.g., the last three moves of a a five-move tactic in Stage 5 appeared as a three-move tactic in Stage 2).]