FAQ for Dead Reckoning

Version 1.6
30-Jan-2007

Q1. What is Retrograde Analysis?

A1. Retrograde Analysis (retro) is a genre of tricky chess problems focusing on the legality of the position. "A position is legal if it can be reached through a legal chess game (no matter how weird)." (That's taken from the excellent Retrograde Analysis Corner, which is the basic online intro to the field.) So these problems are like detective stories: you're trying to look at the evidence to figure out what's happened. They can be quite fun: and at the top end can be extremely challenging to design and solve. This site contains a lot of different chess problems, with particular focus on the theme of Dead Reckoning.

Q2. What is Dead Reckoning?

A2. A weird thing happened. I stumbled across a chess rule that no one knew about before. No, really. The International Chess Federation updates the Laws of Chess every few years. Obviously, they don't change the key stuff: like how a bishop moves. But they do fiddle around at the margins, and in 1997 they introduced a rule which has little impact on the real game, but does have consequences for chess problems. It wasn't until late 2000 that anyone noticed these consequences. It happened to be me.

Q2. What is Article 5.2b?

A2. It says that if the position is such that neither player can possibly checkmate, the game is immediately drawn. These are called dead positions. Sounds simple enough but the consequences can be surprising.

Q3. Didn't it used to be called A1.3?

A3. Oh you've come across it before. But the new version of the Laws in 2005 moved some things around. I haven't got round to changing all the references in this site. It means the same thing.

Q4. What kind of positions are amenable to Dead Reckoning analysis?

A4. More than you might think. About 200 different retrograde problems have been composed to date. These fall into three basic classes of position, corresponding to different reasons why a position might become dead, to wit:

  1. Insufficient Material remaining for a checkmate,
  2. Stalemate,
  3. Blocked Position so that although the pieces can still move they can never achieve a breakthrough.

But in fact, it's can also be used for forward stipulations, problems where there's no Retrograde Analysis at all.

Q5. What does the text below the diagrams mean?

A5. This is the stipulation. Almost all of it is standard notation & terminology used by most chess problem enthusiasts. A quick summary follows, divided into three sections. The main activity asked by the stipulation may be forward or retro. A forward activity may nonetheless require retro analysis (e.g. is that e.p. legal?). A novelty of DR is that to determine the history of the game, reasoning about the future is often required!

GENERAL ELEMENTS OF A STIPULATION

  • (m+n) (e.g. (3+7)) Checksum: White has m pieces, Black has n.
  • a) b) c)... Twinning. Some problems pose the same question for slightly different variants of the same diagram. The questions in the twins maybe the same or different. The "a)" is sometimes omitted from the first twin.
  • DR (stands for "Dead Reckoning").
  • RV (Retro Variants: see Q6 on conventions)
  • AP (A Posteriori: see Q6 on conventions)

COMMON FORWARD ACTIVITIES

  • #n (e.g. #2) Mate in n moves. Also known as direct mate. By convention, White has the move unless it can be proved otherwise, in which case most commonly Black moves first, and then whichever move he makes, White must mate in n.
  • h#n (e.g. h#3) Helpmate in n moves. Black and White alternate playing legal moves, but cooperating, with the goal to mate Black with White's nth move at latest. By convention Black has the move unless it can be proved otherwise, in which case White moves first, and mates with his nth move.
  • h#n.5 (e.g. h#2.5) Helpmate in n.5 moves: i.e. White moves first, and then Black and White move, until White mates with his n+1th move.
  • h#n.0 Same as h#n.
  • =n Stalemate in n moves. Again by convention White has the move. White must force stalemate on Black. A checkmate is not good enough! Stalemate is also known as "pat".
  • h=n Help stalemate in n moves. Black and White cooperate to stalemate Black. Mate or stalemating White is not good enough!
  • sh#n (e.g. sh#20) Series helpmate in 20 moves. Black moves 20 times in series, and then White moves once, mating. Black may not check White until the final move.
  • csh#n Consequent series helpmate. As with sh#n, but in addition every position must be legal bearing in mind who has the move. No history from one position to the next is assumed. This is known as the Mauldon-Caillaud convention.
  • n solutions. There may be n separate answers to the problem, each beginning with a separate key move. This is particularly common for helpmates, less so for direct mates.
  • * (e.g. h#2*) Set play. Solve, then solve again assuming the other person has the move. E.g. h#2* = (a) h#2 (b) h#1.5. #3* = (a) #3 (b) #2.5
  • v (e.g. #2vv) Try. Solve a direct mate problem, then find a move which nearly works, except for some singular defence which defeats it. One v per try.

COMMON RETRO ACTIVITIES

  • Whose move? Except in the starting position for the game, equivalent to "Who didn't move last?"
  • Last move? What was the last move to reach the current position? Includes what piece came from which square and captured what other piece if any. Known as:
    • Type A if you deduce using your retro skills who has the move,
    • Type B if you are told in the stip who has the move,
    • Type C if White is in check (not such heavy retro skills required in this case!)
  • Castling rights? Has the relevant King and Rook have moved before?
  • En passant status? Was the last move the relevant Pawn double step?
  • SPG. Shortest proof game. Find the shortest sequence of move from the starting position to the diagram.
  • PG n.0, PG n.5. Proof game. Find a sequence of moves from the starting position to the diagram taking exactly n or n.5 moves.

Q6. How do you know from the diagram whether a possible castling or e.p. capture is legal?

A6. There are default conventions on castling and en passant which help you decide whether a player is allowed to castle or capture e.p. in an ambiguous position, where one doesn't know the precise history of the game. These conventions apply by default even if nothing mentions them in the stipulation. However, these conventions simply don't affect most of the Dead Reckoning problems, here, since Dead Reckoning itself is there to disambiguate the positions.

The default castling convention says that (given a certain player to move in a forward stipulation) castling is permitted unless it's provably illegal. If you castle, then obviously you can assume for purposes of any retro analysis that the castling must have been legal. If you don't castle, then you can't make any assumption either way. In the case of mutually exclusive castling, the first one to do so gets away with it, and excludes the others.

The default en passant convention says en passant is permitted only if you can prove it's legal: i.e. that the immediate prior move was the relevant Pawn double step. If you don't capture, then you can't make any assumptions for retro purposes that the en passant was legal or illegal.

There are two other conventions which are not default, and would usually be indicated by an acronym in the stipulation: RV overrides the default conventions, and says that every possible legal combination of castling and en passant rights must be considered as a separate problem. AP extends the default conventions, and says that en passant is legal if justified by a later castling move which would only be legal if the e.p. was legal. (If this last one sounds weird, you are right.)

Q7. Where are the solutions?

A7. Various places. Published articles have solutions associated with them. Small DR diagrams can be clicked to expand to a larger diagram for ease of viewing, and these below these diagrams are links to solution pages, which are being updated gradually to cover all the DR positions. There is also still an answer file accessible from the home page which gives the minimal answer without explaining anything.

Q8. What do "DR Illegal" and "DR Restricted" mean?

A8. A position with a specified person to move is DR Illegal if only under A1.3 is illegal. A position with a specified person to move is DR Restricted if only under A1.3 is there a unique last move.

Q9. The new stuff looks very different from the old stuff on this website. What's going on?

A9. I am in the middle of reorganizing. The "What's new?" area points to the new format that I will be using, and I will gradually adapt all the older material to this kind of approach. But the older material remains logically consistent. Please tell me what you like and what you don't like, not just in terms of content but also for style and organization.

For new compositions, I'm using one gif image per composition, and many compositions/page. Earlier I was using one gif image per square, and assembling 64 images to make one board/page. The larger older images are still visible: if you click on a smaller DR diagram it will expand. Try this one:

Grouping related images by page make senses, but several experiments with jigsaw technology for smaller diagrams have not given me satisfactory results, so I am using single images. I estimate that each board image takes perhaps a second to download on a 58k modem, so is probably acceptable for most users.

The main problem now is the quality of the smaller images. The original EPD2DIAG gif images are shrunk on the fly from 314x314 to 180x180. I am looking for a tool which will allow me to generate directly good single board gifs with size approximately 180 pixels. Any suggestions welcome.

These grouped images are probably easier for printing, but I should say that I have not yet looked at putting in pages breaks at the right point. This is not something which HTML does very well.

Q10. Is that all there is here?

A10. Well some people like it! Through DR, I got interested in other kinds of chess problem. You will find an increasing number on this site, and that's where I'm more active now. But I decided to keep the site name for now because it's a hassle to change it. There is still a ludicrous amount of stuff to upload, but I am increasingly getting the stuff published in magazines first whenever I can, but it all gets mirrored here afterwards. I am happy to receive mail from anyone who likes chess problems. Listen: if you're still actually reading, then you might be interested in moving on to this.

Thanks for your time.
Andrew Buchanan.

Back to home page