I have exactly two games installed across all of my computing devices. (Obviously I’m not a gamer and not likely ever to be one.) I’ve been musing on long-term trends in my performance in these two games, and I have come to two conclusions (unsupported by any sort of rigorous analysis): that my performance in the two games has a negative correlation, and that the overall magnitude of my success on one and failure at the other depends on my mental state — specifically, whether or not I’m depressed.

I expect that this comes as no surprise to psychologists and cog. sci. people.

The two games are classic 2-D X11 games. Seahaven Towers is a solitaire variant, known to many through the Microsoft version called “FreeCell”, and Jewel Box is a Tetris-family spatial recognition game. Seahaven differs from the Microsoft game in one important respect: it does not reject (randomly-generated) starting positions that have no solution. This allows a very natural scoring system, in which a “win” constitutes either solving the puzzle (getting all the cards in order with four or fewer units of temporary storage) or correctly identifying that the puzzle has no solution, and a “loss” is giving up on a puzzle that has a solution. (If there is a solution, it is guaranteed to be found by a depth-first search, but this algorithm may require greater stack depth for backtracking than most humans — including me — are capable of. Computers have no problem with this, since there can never be more than 50 conceptual stack frames in a single-deck solitaire game.) My long-term “win” rate is around .699, but when I’m down in the dumps my patience and short-term memory both take a big hit, and as a result my performance takes a nose-dive. Seahaven stores its state in my home directory, so it doesn’t get reset when the package gets reinstalled; thus, I normally only see changes in instantaneous performance through a decline in my long-term average (in more than 32,000 games over the past two decades).

Jewel Box, by contrast, does not stress short-term memory at all; like other Tetris-family games, each randomly-generated game piece is revealed as the game progresses, and the speed of the game increases over time until the player runs out of “lives”. This exercises visuo-spatial skills — recognizing the colors of the jewels on the playing board and how the next set of jewels will fit into the gaps — and, in later stages, fast reflexes, but not a whole lot of memory. (The scoring system of the game does reward cascades exponentially, so there is some benefit to planning ahead, but the PRNG has a way of upsetting whatever plan the player may have, so there is a limit to planning, particularly as the game speeds up.) This means that a fairly mindless playing style — see next piece, move it into position, watch it drop, repeat — is sufficient. When depressed, my scores certainly don’t decline; if anything, they improve. However, xjewel uses a shared global high-scores file that does get erased when the package is reinstalled, so I can’t easily compare my performance today to last year or last decade.

One huge caveat: when I’m feeling good, I’m not playing games all that much, I’m doing something less repetitive and more rewarding, like writing and researching for my Web site, or annotating my photo galleries. Or making progress on my enormous to-read pile. Just one anecdote, by no means data, never mind evidence.