My system for random chess pieces

Over the last few weeks I have been trying to come up with a system for memorising random chess pieces and past games. I wonder if anyone would be kind enough to critique my system so far…

I have come up with 64 Countries that represent each square. If I cant think of a Country, I think of a capital city. Here is an example, square a1 is always America, d1 is Denmark. All 64 squares have a country or capital in the same way as above.

Each piece has a famous person linked to it. I make all the black pieces famous black people and the white pieces well known white people. Here is an example, the black king is Barack Obama and black knight is Oliver Skeete( the showjumper with dreadlocks ). All my black pawns are represented by the All Blacks rugby team.

My white king is David Beckham. All my white pawns are represented by the Boston white sox. I am not a baseball fan, however the white in the title of the team automatically helps me know what colour I am thinking of in terms of pawns.

Here is an example of how I remembered some random pieces a chick at work put on my pocket chess board for me ( yes, she does think I am a nerd and no, I got no action from her :slight_smile: ). The black pawn was put on square c4. For me, c4 is always Monte Carlo. So i imagine the All Blacks doing the haka ( a traditional dance form of the Māori of New Zealand) so loud the buildings Monte Carlo are falling down.

I have a question, do you think I should have two famous people for each bishop, etc. My black bishop is Richard Pryor. As each bishop is placed on a different square, I have not so far got them confused. Richard Prior in Mexico(b6) creates a very different image to Richard Prior in France(h7).

Does anyone see any errors with my system or can anyone think of any changes to make it easier? . I normally start at (a1) and work my way through each square thinking of the associated images if any for that square.

Sorry about the long winded explanation :slight_smile:

Thanks,

Lewis

I’m just kind of wondering why you wouldn’t have another person? You would only need two more people. And anyways, a pawn could be promoted to a bishop or, more likely, the girl at work could make a mistake setting up the pieces.;-).

Thanks, I am still adapting the system and am open to ideas. I will add in a few more characters and see how it goes. Much appreciated.

Thanks Josh for your detailed reply. I never really considered doing that. It makes sense to try other methods, I will take what you said on board. Thanks.

I’m not really sure why you would want to memorize pieces in random positions. That wouldn’t improve your chess game. You would be better off memorizing the positions of the pieces in a match between two grandmasters. I’m pretty sure that the book The Invisible Gorilla: And Other Ways Our Intuitions Deceive Us has a chapter devoted to that (great book).

But…

I thought about the process that I would go through to memorize chess pieces in random positions. I would group the squares on the board. For memory right now I think in terms of PAO, so that is the first way I decided to tackle this problem. Therefore, I would group the board into groups of 3 like this:

***—+0
—**+
**—+
—**=
***—=-
—***=-
***—/-
—***//

That gives you 21 locations of three squares and 1 location of one square. In this system you would only have one group that wasn’t in a straight line–the 19 group.

I would make each grouping a location along a memory journey. I would assign each piece a PAO. So the white king could be Groucho Marx smoking a cigar and the white queen Marge Simpson juggling puppies. If you already have a PAO for cards, then you could just transfer that to the chess pieces and have at least 20 to spare (that’s if you assigned each pawn a separate PAO, but I think that I would only use about 4 PAO’s for the white and 4 PAO’s for the black and just use the one that popped into my head first). Then, I would memorize the 3x1 groupings as if they were cards.

Theoretically, I would say that using this method the world record for memorizing a chess board would be around 20-30 seconds–maybe even lower. [Time based on the record for cards and the similarity in method.]

This is a simpler system than the one you created since there are less things to store in long-term memory and 1/3 the number of locations. It should save you time and improve your accuracy once you’ve learned the system.

I might have to try this out myself.

Edit:

Unlike memorizing cards you will have blank spaces. You could create a blank PAO, but this would be used between 32 and 64 times per board which is very inefficient.

I’m sure that there are many ways to take care of this problem. The first one that comes to mind is a PAO-PA-P (pronounced pay-o-pap) solution. If one location has a person doing something to something else, then all three positions are filled. If a location has a person doing something either to nothing or to something that isn’t already in your PAO system, then only the first two positions are filled. If you only have a person (or a person doing something that isn’t in your PAO to something that isn’t in your PAO), then only the first position is filled. If you only have a person and an object (or a person doing something that isn’t in your PAO to something that is), then the first and third positions are filled. If you’ve destroyed the location, or placed a random object at the location or something else, then you have an empty 3x1.

Whew. Problem solved.

I’ve thought about this. First it depends on what your goal is.

  1. Memory trick to impress your friends
  2. Improve your playing ability by remembering key positions.
  3. Improve your playing ability by remembering entire games that illustrate important concepts.
  4. Attempt to use memory techniques to play blindfold chess.

There is the school of chess training, generally attributed to Russian grandmasters, that says in order to be a master level player you have to “know” 300 key chess positions. You have to know them fully, understanding everything about the position, and recognize the position instantly. The idea is extended to “knowing” key games between master players. GM Ziatdinov has written some about this:

http://www.jeremysilman.com/chess_instruction/ziatdinov_archive.html

I think using a memory technique to help remember key positions or key games is certainly a good use of your time if you want to become a better player. For example, endgame knowledge is largely about knowing key positions, and usually there are only a few pieces on the board. You could use the techniques to remember a lot of openings as well. So using memory techniques alone, you could play two-thirds of the game as well as a master player (with a lot of work, as there’s a lot to remember).

So one question when creating this system is, do you want your system to be general enough that you can use it to remember positions and games?

In order to remember games, you basically need to remember a series of moves. That requires a source-square and a destination-square, and rarely for a promotion you need to know what kind of piece was promoted to (queen, rook, bishop, knight). So for remembering a full game, it’s enough to have images for each square (basic images, PA, however you want to do it), and also have images for promotion piece types (QRBN).

Here’s an example. Using a Dominic system, square a1 could be Andre Agasi, and his object would be a tennis racket. Square f6 is Frank Sinatra with a microphone. So if a bishop moves from a1 to f6, you just remember Andre Agasi holding a microphone. This should be straight forward, but potentially it requires a lot of locations. If a game is 60 moves, you need 120 locations. Games can sometimes go 100 or 200 moves. If you’re wanting to remember a lot of games, locations add up quickly, so I see the number of locations required as one problem that needs to be addressed. If you could store more moves per location that would help.

For memorizing positions, I considered a few different approaches. One approach is the straightforward way of having 64 locations, and storing images based on the piece in the square. This also uses a lot of locations. If you want to remember 300 key positions that’s 19200 locations.

You can approach it from the other direction. Instead of 64 squares, there are a maximum of 32 pieces in any position, and as few as 3 pieces in the endgame. Since a lot of the key positions that GM Ziatdinov says are important are endgame positions, this might be a reasonable choice.

As for piece encoding, you could go with something straightforward, like the black king is always Barrack Obama. One idea I thought of was to give each specific piece an identity that sticks with it. So the white kingside rook is a different image than the white queenside rook. I thought that white could translate to W or L (for “white” or “light”) and black could be B or D (“black” or “dark”). Then the white kingside rook could be LKR, which could be the word Laker, and the image would be a memorable player for the Los Angelos Lakers, say Magic Johnson. So if the white king rook moves from h1 and later ends up on a1, your image would be Magic Johnson holding a tennis racket.

You could improve this by using a chaining method. The issue with chaining is always that you need a proper set (meaning no duplicates, otherwise you would remember incorrectly). If each piece has a distinct personality (as discussed above), then you can use chaining without any duplicates. Obviously you can’t have two pieces on the same square, so it’s impossible to duplicate squares. So if you avoid duplicates with the pieces, you can just create a chain and remember an entire position per location. This is obviously slower to recall because chaining is a serial process to recall. For example, to remember what piece is on d5, you have to go through half of the chain one by one. You could combine approaches and use chains to remember rows of the board, and then to recall what piece is on d5, you start at a5 and follow the chain to d5. That would require 8 locations per position.

I am not so sure that the slow recall for the chaining method is a deal breaker. Let’s say you are trying to memorize a key endgame position for KRP vs KR like the Lucena position. It’s only 5 pieces on the board, so your chain is short. And the memorization technique will only be required while you are learning it. After a while you won’t have to rely on the memorization technique, although you could always fall back to it if you forget. Because the most important positions will be the most frequently used, your brain will sort of “cache” them so you recognize them quickly, and recalling a full position with a lot of pieces will happen less frequently.

Finally, if you are intending to use a memory technique to help with blindfold chess, then the speed of recall becomes important. You essentially need to recall the current game state (position, side to move, castling rights, etc), and although not necessarily required, you probably want to remember the moves of the game as well (at least I would). For fast recall, I would want an image or location for each square, and then images for each piece type.

I’m not sure how well this will work in practice, because as you move pieces around during the game, and while thinking about potential moves to play, there is going to be a lot of repetition. In a very short period of time, the e4 square might have numerous different pieces occupying it. In theory, your mind only recalls the most recent connection between e4 and the image for the piece occupying it. But I know for myself, my first location in my house is the trash can in the kitchen, so it gets used frequently. If I’m going to the store, it’s usually not a problem to remember what item was attached to the kitchen trash can (low stress, I can take my time creating the visual image, the image isn’t being reused). But if I’m going through my busy day at work, I quickly attach some image to the kitchen trash can, and then later that day without thinking about it I need to remember something else and I attach another image to the trash can, and the next day I couldn’t tell you what image was attached to the trash can. Playing a blindfold game, to me, seems closer to the stressful busy work day than the stress free trip to the grocery store.

I have not actually used any of these systems yet. I also began thinking about it recently. So let me know what your thoughts are. Maybe we can combine some of these methods and come up with a really good system.

I like your idea of using the dominic system, I will definitely look into adopting that idea.

My goal was just to use this as a memory trick more than as a way to learn chess. My chess is decent, but certainly not great.

I remember watching this on UK television a few years ago now. Around the 4 minute mark, memory and in particular chess memory is discussed.

I like the idea of using chess notation. There is a description here.

I think a system built on algebraic notation would be ideal because that is how chess books list famous games. If one were very familiar with a mnemonic system based on that notation, it would be easy to pick up a book and memorize old, famous games just from the notation.

It seems like one would only need 64 images for the board, and five images for the pieces. A 60-move game would take 30 loci, because in algebraic notation, it appears that each turn holds two moves. Example:

  1. e4 c5
  2. Nf3 d6
  3. Bb5+ Bd7
  4. Bxd7+ Qxd7

Since the quantity of data in each locus changes, I don’t know if a compound-image system (PA or PAO) would be ideal for that kind of notation.

One idea is to create a grid, similar to the one described here. The letters could be the first letter of an object or person, and the numbers could be taken from the Major System.

a1 = atl-atl
a2 = anteater
f5 = fly
h2 = Han Solo

Then the 5 pieces (K, Q, B, N, R) could be people.

The locations e4 and c5 might be two objects/people interacting in one locus. “Bb5+ Bd7” would have more elements to the image, but could also go in one location. The “+” could be an image modifier, like a guillotine.

I made a wiki page where people can compile ideas on different ways to memorize chess:
https://artofmemory.com/wiki/Memorizing_Chess_Games

Just brainstorming out loud… :slight_smile:

On second thought, you could only use a single chain to remember a single position, not a full game (because of potential repetition of moves). You could use a chain for as long as there no duplicate moves in the game, meaning most games could be remembered with only a few loci, each containing a chain of moves.

Another problem is, using a chain method, you could only remember a single game before encountering many duplicates which will create confusion.

I’m thinking each game needs some keyword image, where every move’s image is combined with a key image.

For example, let’s say I’m remembering a game between Kasparov and Karpov, and I come up with some keyword image, say a hammer. For move 1. e4, ED in Dominic system is Ellen Degenerous. 1…e5 is Emelio Estevez. So the chain is Ellen Degenerous holding a hammer, linked to Emelio Estevez holding a hammer. And so on. Or if Emelio Estevez’s object is a hockey stick, then maybe the first image is Ellen Degenerous holding a hammer and a hockey stick.

Do you think this idea is reasonable? My primary concern is minimizing the number of loci as I think in order to apply this to serious study would require memorizing a lot of games. Do you think I’m going overboard on minimizing loci here?

when it comes to poetry, people often learn its grammar - rhyme scheme, scansion, verse etc. I would think of it the same way with chess. At least learn the opening used in the game. Also, learning basic tactical motifs and the historical significance can help. a lot of times there will be a wiki article on that game or at the very least a few germane comments on the chessgames.com page. openings are now notoriously broad and ramified with the improvement in computers and software. every opening has an ECO code that tells you the first few moves. below, i’ve posted the first few. it goes from A00 to E99. in the long run, it’s more practical to learn games by openings and variation (sicilian defense, kere’s attack or french defense, winawer variation) than ECO i think. most games start either 1. e4 or 1. d4 and in turn there’s a handful of responses to those.

so i’d code games based on the defense used and have preset images for those.

for 1. e4 maybe have images for the open game (1. e4 e5), french defense (1. e4 e6), sicilian (1. e4 c5), caro-kann (1. e4 c6)

for 1. d4 black most of the time responds either 1. d5 or 1. nf6. you’d want images for the queen’s gambit accepted, the main declined lines (slav, semi-slav, tarrasch) and the nf6 defenses (nimzo-indian, queen’s indian, king’s indian, gruenfeld). that’d be a starters kit. still, it’s a lot better to learn a few openings and memorize games than to have most of your games have the same initial 4-10 images.

ECO code sample:

A00 Uncommon Opening
1 a3, b3, d3, g4, etc.
A01 Nimzovich-Larsen Attack
1 b3
A02 Bird’s Opening
1 f4
A03 Bird’s Opening
1 f4 d5
A04 Reti Opening
1 Nf3
A05 Reti Opening
1 Nf3 Nf6
A06 Reti Opening
1 Nf3 d5
A07 King’s Indian Attack
1 Nf3 d5 2 g3
A08 King’s Indian Attack
1 Nf3 d5 2 g3 c5 3 Bg2
A09 Reti Opening
1 Nf3 d5 2 c4
A10 English
1 c4
A11 English, Caro-Kann Defensive System
1 c4 c6
A12 English with b3
1 c4 c6 2 Nf3 d5 3 b3
A13 English
1 c4 e6

That is an interesting idea. I searched around and here are some pages with links to more information:




Right now I am working on using memory for chess. I will fill up the wiki page soon.

There is a book called 100 Endgames you must know. I am halfway to memorizing all the positions … just the positions. Earlier, I had gone over Keres’ Practical chess endings … which had 333 diagrams. Next week, hopefully, I would have memorized it all.

I have got the book by Ziatdinov, and it will be my second memorized book.

Things to remember: Openings is different from Whole Games is different from middlegame Positions is different from Endings.

==============================================

I started with the Endings. So, here is how I learned the positions:
Knight vs Pawn. (the pieces are White king, White knight, Black King, Black Pawns).
Each square is named according to major system (a=1, b=2 … etc).
So, the number 24252684 represents 24-25-26-84 = b4-b5-b6-h4 … ie. White king is on b4, White knight is on b5, Black king is on b6 and Black pawn in on h4.

I have divided endings into specific patterns, and hardcoded the piece-order.


Game:
I translate each half move into a 3 digit number. The first digit is for the piece and the other two represent the square to which it moved.
So the rules for square is simple - d4 =>44, e5 => 55.
The rule for Pieces is following:
King = 1
Queen = 2
Bishop = 3
Knight = 4 or 5.
Rook = 6 or 7
Pawn = 8 or 9.

The encoding for knight, rook and pawn allows for two digits each, because, either one of them can occupy a square. For instance, it could either by Nbd7 or Nfd7. I use the lower number to indicate the ‘lower’ alternative … in this case, 447 would be used to describe Nbd7.

Say, a game consists of 30 moves. Or 60 half moves. Then in my system, I would memorize 60 * 3 = 180 digits. The standard PAO for playing cards encodes 6 digits per locus. So, I would need 180/6 = 30 loci to memorize a game.

Middle game position: You can use FEN notation for random positions. If it can arise out of an opening, and works out to be shorter or not significantly longer than FEN, then use the whole game method.

Openings:
I have a scheme. Firstly, I classify the opening into subcategories, before I start memorizing them … So, I would either be in Maroczy Bind or Accelerated Dragon mainline, or Accelerated Dragon Sidelines …

For each Subcategory, I create a table with a maximum of 1000 rows.

Each row contains a packet, with the following encoding:

  1. Number of ‘RAW’ moves if different from 1 (it will start with 0).
  2. RAW moves
  3. list of opponents moves, and row number

For example, say I am in a row that corresponds to the following moves (I am black):
1.e4 c5 2.Nf3 g6 3.d4 cxd4 4.Nxd4 Nc6 5.Nc3 Bg7 6.Bc4 O-O 7.Bb3 a5 8.a4

Then I want to play 8…Ng4 and depending on White’s response (9.Qxg4 OR 9.Nxc6), I want to do different things - which are present in row 83 and 101.

So, I will have the following entry in my row:

474-174-083-436-101

474=Ng4
174=Qxg4
083 = 83rd row
436 = Nxc6
101 = 101st row.

================================================================