I have finally put my hands on a physical copy (paperback) of The Elements of Typographic Style. It’s a rather hard book to get a hold of! The adventure of acquiring it, however, is nothing compared to the adventure of reading it. The book is a gem, and I know I am just the millionth person to say it, but I can’t help it.

Topping my stand-out list are:

  • Conciseness. Every sentence, and each paragraph, is filled with content. I should know this, yet I am stunned by what careful selection of words can do. Even more so in our noisy digital world, where so-called content creators make it their priority to spoon-feed the meaning of everything (their chosen meaning, that is) with an endless stream of useless words. How reassuring it is to be asked to actively think. How honorable it is to let the interpretation happen in the mind of the reader.
  • Beautiful prose. Who said that conciseness and beauty cannot coexist? It’s usually called poetry, and, although this book is in prose, some rather large parts feel like music.
  • The sheer amount of technical points. Like I said: a gem.
  • The typesetting. Well, I knew it already having looked at it on the screen several times; what did you expect?!

The bonus feature is that, like only great books do, it makes me want to write.

In the past six weeks I have attempted, quite successfully, to resurrect an old hobby of mine: chess. I used to dedicate a good part of my day to it during the last two years of high-school, settling in the 2150-2199 Elo Fide range; Gave up on it during the five years of college; Got back into playing it (not working on it) during my three PhD years; Gave up on it again for ten full years.

As I said, I am quite satisfied with the first six weeks of my self-assigned resurrection project, and before I say more about it, here are a few things I noticed that I believe are a consequence of the ten-year hiatus.

  • Visualization and memory. I’ve never thought of myself as an exceptional player. After all, I started playing at the late age of 16. But my visualization skills were good. I once played two blindfold games in parallel (like a simul). And my memory was also quite good. I am disappointed to say that when I started the program six weeks ago I had difficulties remembering a position on the board after going just a few moves down in a variation!
  • Accuracy. I was pleased to discover that my imagination and candidates selection is still fine. But my accuracy in the calculation is … very disappointing.
  • Assumptions. I noticed I make stupid assumptions about positions occurring a few moves down in a variation; I mean mistaken evaluation of these positions, based on “intuition” when intuition is not the tool to use. I suspect this is a consequence of mindless, or mind-numbing blitz games online.

A few sidenotes. In his very well known book, Yermolinsky writes that poor visualization may be a consequence of looking at games and variation on a computer screen. He wrote that twenty-six years ago! I agree with him on that, and that’s why I have a few ideas on how a chess software should look like (more on this below).

To get back into something after ten years cannot be easy, and I was expecting it to be even harder. That’s why I am satisfied with how the first six weeks went. I can feel my brain tingling in frustration, which is exactly how it’s expected to react to a new type of learning stress.

I do think that I have an advantage over my previous self: I am ten years older. That is, I have had ten years to learn more things about myself. That’s perhaps the first main point, a re-formulation of the old know thyself.

Most of the learning accumulated during this time comes from my other hobby: running (I sit at the moment at 2:33:xx marathon, and have every intention to improve some more). If I try to pinpoint the items, they look like this:

  • Active learning is the efficient way, and therefore the only way under time constraints (as I have no intention to be a full-time runner, nor a full-time chess player).
  • Passive learning is comfy, but nearly useless past a certain threshold.
  • A “training program” is of utmost importance. Tracking some type of measure is equally crucial.
  • A training program is only half-effective without a goal.

Now, most of these things are rather obvious in the context of athletics, but perhaps not so much in a more brainy context like chess is. Here is where the knowledge of my own strengths and weaknesses turns out to be useful. On running, we would say the individualization of the program is what makes the difference.

Take the first two items above. Here are some examples of passive learning techniques, alongside possible active alternatives.

  • Play blitz games / Solve puzzles
  • Watch a game live / Review a commented game
  • Review a commented game / Review a uncommented game, then check your analysis

I put “review a commented game” in both categories: it can be active, but it can also be passive. In fact, that’s nearly all I did in those two years of high school, and it did take me from zero (well, 1440 was the starting Elo, I think) to 2150-2199. But then, it stopped doing me any good. This is the individualization principle in action: what used to be an active “workout” became a passive one, leading to stagnation of my “fitness”.

I have learned this concept, the hard way, in running training. If the program doesn’t evolve according to the evolution of one’s fitness, with time the athlete’s fitness will first stagnate, then even decrease. The reason is the lack of new stimuli in the body. I like to think that this applies equally well to the muscles as to the brain. Reading a chess book, however well commented, doesn’t trigger any new stimulus to me. Nobody says that it never again will, it may very well do it, in some future, but at the moment it does not.

Another key idea that I believe I can transport from athletics into chess is that of variability: a training program must introduce a healthy dose of variability. The idea beyond it is again that of stimulating various facets of one’s “fitness”. Typically, in running training programs, time is divided into phases, and each phase emphasizes more one type of stimulus over the others, without neglecting any. For example, as an athlete gets closer to a race, she would prioritize workouts specific to that kind of race. I think this last concept has its place in a chess training program too, but I haven’t tinkered much with it yet, as I feel that, after ten years of idleness, I need to brush up on the basics.

And so, speaking of basics, I came up with four kinds of training stimuli that I think my brain needs to get back to where it was. These are not very granular containers, and for each I then defined a few more specific “workouts”.

  • Calculation (e.g., puzzles, problems, endgames)
  • Strategy (e.g., review games)
  • Openings (e.g., key structures, or rapid games)
  • Play. Probably the most effective for me, but also the hardest one. I am leaning towards games against various engines, with different configurations.

I mentioned briefly that I have a few ideas about how a chess software should look like. The story is that I have had these ideas since … 2010, perhaps? Anyway, it was some time during college, as a software engineering student, without money, I did not want to pay for ChessBase. And I am grateful for the existence of some free, open-source alternatives, but none of them quite hit the mark (my mark, that is). I do use SCID, but as little as I can, and only when I really have to; I just can’t stand the overcrowded GUI1. Now that I have more time on my hands, and that the silicon has become a trusted partner, I have revisited those old ideas and have developed what I think a good chess training software should be.

  • A chessboard. Yes, that’s number one! But an efficient one: just the pieces and the notation and a small panel for engine analysis. Also, “Copy FEN” and “Copy PGN” are must-have options, just a click away.
  • A puzzle view. This is very simple: PGNs open with the notation hidden and a position to solve on the board. A single button that says “reveal solution”. No need to search for this very basic form of training across a million context menus (I know it exists in SCID, but I cannot find it!). I added a button “mark as solved” so that I can later add a spaced-repetition feature, just like in flashcards.
  • A Play vs Engine view. Again, it’s all about active learning, and playing is the most active session of all. I made it very easy to set the starting position and the clocks.
  • Databases. I haven’t implemented this yet, and I am not sure I want to. I think SCID and the others do a good job at reading, indexing and showing databases. Their interfaces are cluttered though, so maybe I can rework the concepts

Circling back to The Elements of Typographic Style, it gave me the push I needed to get my hands dirty again with ConTeXt. I designed a Lua module to typeset chess books, and used it successfully with ConTeXt, LaTeX’s smarter brother. It has come up real nice, although I am still debating what the best font faces are. For the figurines, I am quite settled on Noto Sans Symbols 2, as the alternatives are quite horrible. But for the body … I am unsure. What goes well with a chess mind?

  1. I should point out, at this point, that I am Vim user, so the “overcrowded” adjective may be a bit subjective. ↩︎