April 2022: Learning Shogi

Chess has experienced a fascinating rise in popularity in recent years, even in terms of a spectator sport by way of twitch streams and hit TV shows. It's one of those easy to learn, hard to master games, with books upon books written about the openings, tactics and mathematics hidden the 64 checkered squares. I've always enjoyed a casual game of Chess, mainly because almost everybody knows how to play, and it's a nice thing to do while having a conversation. I'm also not great at Chess, so it's also a great exercise in humility. One of these days I also want to see how high I can get my ELO, if I try to touch one of those aforementioned Chess books, but this month I wanted to take a look at one of those games that's not mentioned as often, but I think might be fun as well, if a little niche.

Shogi (lit. "general's game"), or "Japanese Chess", looks a little like Chess, sounds a little like Chess, but for me is a much more frantic experience. For one, when I first tried to learn this game a few years ago, I noticed that standard game pieces looked a little like this:

Eagle eyed readers might notice that the game pieces can only really be told apart by the kanji on the piece, and as somebody who couldn't at the time read kanji, that was a recipe for disaster. These pieces also promote differently than they do in Chess, so what initially looked like ten or so kanji turned more into twenty, and at that point I gave up. Now however, I can read some kanji, including some of the ones on the board, I decided to take another crack at it.

Basics for these games are of course what piece is allowed to do what. All pieces can only capture by moving, so that's old news, really. Except for the pawns. They also may only capture in the direction they move. None of these Chess-exceptions that makes the pawns go diagonally all of a sudden. Speaking of pawns:

The Pieces

The Pawn

As in Chess, the pawn is the most numerous and the weakest piece in the game. It'll be cannon-fodder for the most part, and it'll delight in standing in the way of stronger pieces a lot. It can only move a single tile forward. I could at the time or writing recognize the kanji as the operative component of "aruki", meaning "to walk". The piece itself shows only the first part of 歩兵 (fuhyō). If it enters the furthest three rows of tiles, or the "promotion zone", this piece may promote to:

This is the (と金) tokin piece. It can move into all adjacent tiles, except for the ones diagonally behind it. It shares that movement with the gold general.

Gold Generals

This one is the guy everybody on the board wants to be, exemplified by the fact that almost all promoted pieces move like him. The kanji on the piece is also just the radical for "gold", so maybe that plays a role in that. It's fully readable as 金將 (kinshō), with the same "shō" that appeares in "Shōgi". He's also so damn sure of himself, that he doesn't promote.

Silver Generals

Silver generals, or 銀將 (ginshō) move to any tile in front of them, or diagonally backwards. Similar to the kinshō, the kanji on the tile is the radical for "silver".

Once promoted, they'll turn into 成銀 (narigin). The kanji is supposedly a cursive version of "kin", but in all honestly, this confused me a lot while differenciating between the tiles at the beginning.

Knights

Knights in this game are a bit of a headscratcher for me, honestly. In Chess they're great, because they cover situations other pieces just aren't equipped to handle. They're very versatile in densely populated areas, and are one of my favourite pieces to check a king with. The 桂馬 (keima), or laureled horse (?) however only gets the two frontmost moves of the Chess horse, meaning it gets three or four moves before it gets promoted.

Then it becomes a 成桂 (narikei), or promoted knight and moves once again like the kinshō. This one also supposedly shows a cursive version of the "kin" character. You get how this started to get confusing for me, right?

Lances

Lances, or 香車 (kyōsha, lit. incense chariot) moves forward and never looks back. It feels a little like the lawnmowers from "Plants vs. Zombies", meaning a failsafe, in case somebody stumbles into the edges of your board.

成香 (narikyō) move of course like the kinshō, and features yet another cursive version of the "kin" character.

Bishops

角行 (kakugyō) can move diagonally and far as they like, kinda like bishops, you know? The piece itself reads "kaku", or "angle". 角行 then is something akin to "goes in angles".

Errr... I'm still learning and very much on an expert on the language, but this doesn't look like kanji. I looked it up, and apparently it's not. It's a derivation of an equivalent piece in Janggi, or Korean Chess. I wonder why that is... Anyway, the promoted bishop is called ryūma, or "dragon horse", which by the way, is a great name for a game piece. Dragon Horse. Who needs royalty, when you've got a Dragon Horse? I'd pull a quick Richard III, if that means the resident Dragon Horse will notice me. It also doesn't really care about the kinshō, because the ryūma can move like the kakugyō, and to all adjacent tiles.

Rooks

飛車 (hisha), or flying chariots move like the rooks in Chess. The "hi" in hisha refers to "flying". I originally learned the second character that's not on the piece as "kuruma", meaning car. I guess a flying chariot is kind of a flying car.

龍王 (ryūō) moves like the hisha and also into all adjacent tiles. The name translates into "Dragon King", which is also fun, but nothing beats the Dragon Horse in my humble opinion. The character on the itself is supposed to be the ryū, which I'm sad to say, I don't really see. Maybe this is another cursive, but until I know to read this properly, I'm gonna believe them.

Kings

The kings are actually the only piece to have a black/white variation, though it's not literally black and white. Instead there's 玉將 (gyokushō), the jeweled general, and 王將 (ōshō), the king general. It works very literally the same way the king in Chess does: Moves to every adjacent space, and if it's set checkmate, that's a loss to the respective side. The "sente" moves first, the "gote" second, and which is which has nothing to do with which player has which piece. Rather, the higher ranking player will play the gyokushō, the lower ranking the ōshō.

With all these rules in mind, I thought I'd go with learning by doing. It's how most games make their way into my memory anyways. I found a website where I could play against bots and people (eventually), set up a match against the weakest bot I could find, and went into it. The match went more or less alright. I still had to check which pieces could move which way a lot, but I was getting the hang of it. I also noticed that I had to be a lot more careful in how I use the pawns, since they can't support each other like they can in Chess. In Chess, a lot of my early game is very much focused on pawns, so the stronger pieces get a little space. That's not as good of an idea in Shogi. Here the bot pulled out a rule that's technically standard, but somehow the guide I used managed not to mention to me: Dropping. Whenever a player captures a piece, they gain the ability to drop the same piece on the board, within a few restrictions.

Basically, you're allowed to place any piece you've previously captured from your opponent on the board, as long as it can still make a legal move on the board. That really mainly applies to the pawns, knights and the lances, since they're the only pieces that can only move forwards. The pawns additionally may only be dropped into columns that don't already contain an allied pawn, and can't be dropped to checkmate the opposing king. Promoted pawns don't count towards the dropping rule anymore, so yes, technically one can have two entire sets of pawns on the board.

This also makes trading pieces not as sensible. Trading pieces only becomes a viable approach, if you can actively trade down the opponent, otherwise the same pieces might turn out somewhere where they're much more dangerous. Of course having to read doesn't make this feature any less stressful.

Aside from the reading problem (I have no ideas what the promoted tiles say, I'm just going by the rule of thumb that everything that isn't the bishop or the rook move like the gold general), I have little intuition of how to quickly and efficiently to cover pieces. Oddly enough, not having something like a Knight in Chess is what's difficult to adjust to. Sure, they have a knight piece that can jump pieces and moves in an L-shape, but it can't go backwards, so it's infinitely less versatile. It makes for quick promotions later in the game, but by then the game is more or less decided at the stage I am at. It's a fun game though. I finally get to use my pawns like the kamikaze-weapons I usually do when playing Chess, since trades with pawns are always favourable. Oddly enough, I've had games where I've had almost two entire sets of pieces in my control. I'm still not good enough to reliably win against the weakest lishogi bot, and when I do, I rarely do so elegantly. I've sneaked some peaks at other players' boards, who usually played against much stronger bots, and they tend to be more deliberate in positioning and board control, so I know it's possible.

Pushing Around the Generals

I think I might have figured out a decent opening. I can't really move the bishop or the rook early, since they are almost ineffective, if the opponent has a good number of pawns, and I should avoid blocking the second row, since that row helps secure the bishop, once attacks start coming in. I also wanna bring pawns forward to give myself some space to position pieces like the lance and the knight, but pawns can't support one another. Instead I'll move the generals forward and cover pawns in groups of three or less. That can get me decent pawn trades, and I can always drop pawns back, if it's a pawn trade. The gold Generals are a little more convenient here, since they can basically remain stationed in the position they've traded into, since they can move a space horizontally. The silver general is going to expose all other pawns, if they are moving from their position.

Bishops and Rooks

I noticed something in some of the early games I played that the computer heavily favoured attacking the left side of my board, and that I was usually not well prepared for even very simple maneuvers. I think this is one of the key features that differentiates Shogi from Chess. There is a very heavy asymmetry on the board.

The asymmetry only really expresses itself in the placement of the rook and the bishop. There's 9 squares both ways, so the if one were to remove the pawns and trade bishops, that's an option. The only pieces covering the bishop is the rook, all the way across the board, and the left silver general. Trading equally is an alright way to diffuse a situation, but it opens the option for the enemy to drop one of the strongest pieces in the game somewhere where they already have their advantage. The other attack on the bishop however, is a little slower, but much more effective.

For context, I usually open with the King's Gambit in Chess, and then I do trade down, so my enemy can't establish the bullhorn formation. I usually try to get something into the field that can move diagonally, to navigate the enemy pawns. This instinct doesn't work in Shogi. It's also kinda pointless, because pawns can't form blockades, and considering you can drop pawns, trading pawns is a better option. On the the other hand, getting the rook out there is not exactly trivial, because you'll probably trade pawns along the way, and like in Chess, the rook has difficulty navigating a full board.

So technically, moving the rook's pawn forward, and trading it, makes things can open an attack opportunity for your rook to capture the bishop. That just capturing the bishop in its default position with the rook will trade you a bishop for a rook. That's not great. What makes this a much more dangerous situation is once again the dropping rule. If you've traded pawns during the attack, then you can drop the captured pawn directly in front of the bishop. In the default position, not only is there no piece that can preemptively capture said pawn, the bishop also can't evade. That's a great trade. Should the enemy drop a pawn in front of the rook, that does nothing. A defense against this attack involves at least one gold general, or completely dismantling the left side of the default position. I favour what I did with the gold general above, since that defends against both attacks involving the strong pieces, and takes two to three turns to set up. The next move is to capture the pawn with your own and then you can carry on with whatever you were doing. In this game, that would have been moving the rightmost pawn up far enough to force the enemy lance to capture it, so I can take it in turn.

Queens and Dragons

I think the point where learning the game differs from learning Chess is the absence of the queen, or an analog. Since the board doesn't necessarily have to clear up at any point, the queen wouldn't be as effective either. Instead of the queen, we have the bishop and rook as the objectively strongest pieces. They similarly aren't as universally effective, meaning beginners like myself never really fall into the trap of over-relying on the stronger pieces. I think that's a pretty good thing as far as game design goes. Generally, the dropping rule makes things so very much more interesting, and maybe also chaotic than Chess. For example, there's a reason why I likened the lances to the lawnmowers in Plants Vs Zombies. Since they can't go backwards until promoted, and then are pretty slow on their way back, there's a good reason to drop pawns to promote behind the lance and clean house.

The Takeaway

Judging by my own experience with Chess, it's going to take me a while to get decent at this game, and resources for this game are a little scarce for people who speak the languages I speak. It's quite fun though, and I could be convinced to get a a physical board. It's such a simple set as well. I would just have to put somebody else through the first five or so games where the kanji aren't memorized yet. Semi-seriously, one of the biggest takeaways for me personally was, that cursive kanji are weird. Who would've thought that there are so many cursive variations of the same kanji out there. I'm not looking forward to looking at cursive kanji that aren't just "gold". I was quite pleasantly surprised, that my kanji reading skills have progressed enough to identify the important radicals well enough to more or less play without needing help checking on the moveset. All in all, I'd recommend this game, if only to have somebody to play with. Maybe I'll take a look at other Chess variations (I heard, Xiangqi has a river). Not in May though. I'm fully expecting to lose a week, but nevertheless, I want to have something more to show at the end of May.