Option Chess by Paul Bonham

by ChessBase
2/17/2014 – As you know we have been on a search for a chess variant that keeps the game as recognizable as possible yet dramatically weakens the play of even the most optimized computer engine. The goal is to weaken the play of such engines so that they cannot defeat most humans, and won't be able to do so for decades or centuries to come. The latest variant is by a Canadian software engineer.

ChessBase 17 - Mega package - Edition 2024 ChessBase 17 - Mega package - Edition 2024

It is the program of choice for anyone who loves the game and wants to know more about it. Start your personal success story with ChessBase and enjoy the game even more.

More...

Option Chess

A chess variant by Paul Bonham

Many people, e.g. Frederic Friedel of ChessBase, have considered the problem of how to modify the rules of chess play in a way that keeps the game as recognizable as possible (minimal changes from the current rules) yet dramatically weakens the play of even the most optimized computer engine that knows and handles the modified rules. The goal is to weaken the play of such engines so that they (a) cannot currently defeat most humans, and (b) won't be able to defeat most humans without a paradigm shift in computer technology that may be decades or centuries away. Azlan Iqbal’s Switch-Side Chain-Chess and Dr. Kenneth Regan’s Tandem Pawn Chess are very creative and fascinating ways to challenge computer engines as never before… but is there a perhaps even simpler way?

One of the best ways to stymie computers in general is to force them to make hard inferences from limited data. This is something humans do rather constantly and have had to do throughout history. It could be argued that our brains have evolved to become optimized for these types of decisions. In today’s world, we humans make decisions about careers, about marriage, about where to live, and we do it mostly based on a fuzzy kind of logic called intuition. We get "feelings" about something. Optimized as we may be, we still make bad decisions, but the fact that we are here in a hostile universe on a hostile planet is testament to the fact that overall, we as a set of thinking machines (whether or not we have any kind of spirit or soul) do okay. We are still ahead in the extinction game.

This made me ponder: what if a highly-tuned computer chess engine was presented something it couldn't calculate or even assign a numerical value? What if for one or more of the options available, there is simply nothing to calculate? What if an engine has to "learn" to become intuitive?

This is the approach I took to creating a minimally-changed chess variant that would see computer engines, reprogrammed to handle the new rules, plummet back into the ranks of mediocrity in terms of performance versus human opposition. I would give the engines something they can't calculate: a choice between doing something now versus saving it to be done at an unspecified future juncture of the game. The farther into the future that juncture gets pushed, the less can be known about it. And it doesn't take many plies into a chess game's future for the known 'value' of that unused option to become... completely unknown.

In essence I will agree that this really amounts to an explosion of growth in the search tree. But it is an explosion of such magnitude that is it useless for engine programmers to pursue any of their normal CPU or memory or bandwidth or concurrency optimizations to try and compensate. Within just a few plies, it outstrips all imaginable compensations. It demands either radical new technology... or something more imaginable, a change from brute-force programming into AI programming.

I call my variant Option Chess. The option that is introduced is something familiar to chess variant enthusiasts: an option to play a double move in place of a normal single move. But there is a limit to the use of this option, and a double move has restrictions. The engine author would have to be careful that the engine didn't immediately want to play double moves as soon as it can, one right after the other. In other words, the strategic concept of holding double moves for the best moment(s) would have to be somehow coded into the engine, and that is something that isn't just brute force numerical calculation. Each engine, just like each human, would have its own "preferences" that would trigger making a double move at one particular juncture of a game. Chess would become less about pure logic and calculation and more about personality. Yes, computer chess engines would really take on "personality"!

Hereinafter, a double move option shall be described as an "option".

I decided after some consideration that the option should not become available until after Black's eigth move. Thus all standard chess openings can continue to be played to a depth of eight plies. Beginning with White's ninth move, the option exists for each player, and each player shall have twelve such options available to use between their ninth move and their 48th move inclusive. That's 40 moves with twelve chances to use a double move.

Also, after Black's 48th move, each player shall be awarded an additional four options to whatever number of options they still have remaining. This is to ensure that the endgame phase should not run out of options, because we want to affect the endgame as much as the middlegame. Tablebases: begone! Endgame Authors: prepare to write new and exciting books! Ditto for those who compose chess problems: imagine a forced mate in four in which both players have at least one option remaining!

Here's an interesting problem position which in standard chess is rather non-descript and has no mating or winning lines.

A normal chess engine gives Black about a pawn edge, with the following best line: 1.... Rxd2 2.Rxd2 Qe7 3.Qxe7+ Bxe7 4.Ng5 Nf5 5.Ne6 Bf6 6.Kg2 Bxg7 7.Nxg7 Nxg7 8.Kg3 Nf5+ 9.Kg4 Nd6 10.a4 Nc4 11.Rd3 Rg8+ 12.Kf3 Rh8 13.Kg3 Nd6.

But in Option Chess, it is Black to move and mate in three! Can you solve it after reading up on the rules of Option Chess? (Note the green tokens on both sides of the board!) The solution is given at the bottom of the page.

Before describing double moves and their restrictions: how do the players and indeed the spectators of a game keep track of how many options are available to each player? In physical over the board play, each player is given twelve "tokens" to place in plain view in front of them but not on the chess board itself. These tokens would be small disks, green on one side and red on the other. They could be made from checker pieces, painted green on one side and red on the other. Or they could be custom manufactured, or even printed on a 3D printer. They would initially all be green side showing up, to indicate this represents an available option, as shown here:

When a player decides to use an option on their turn, s/he must first place a token in front of the chess clock, visible to the opponent and still green side up. This indicates the player on the turn is committed to a double move, and this choice becomes irreversible once the token is released. The opponent must be notified by this manouevre. Playing a double move without such indication would be treated as an illegal move. This next diagram shows the proper precursor to the play of a double move:

The player then makes the two moves according to restrictions described below, taking as much time between each move as s/he likes. Only when the second move has been made does the player flip the token over to red and then hit the clock. The now-red token indicates a used up double move. It can be left where it is or placed into a container for used up tokens. The next diagram illustrates this action:

And now, as promised, the restrictions on double moves:

  1. Neither player can make a double move before his or her move nine (this allows for standard openings up to Black's 8th move inclusive).

  2. The first move of a double move may not give check.

  3. The first move of a double move may not capture material.

  4. An option may be played when the king is in check and a token is available, and either the first or the second move may be used to eliminate the check.

  5. Castling is a single move, not a double move.

  6. The same piece may be moved on both moves of a double move (meaning "ricochet" bishop, rook, or queen moves are possible, as well as lengthy knight moves) provided that the intermediate square be empty, even for knights, that neither move is a capture and that the first move does not put the king in check. Also, if the king is the piece moving twice, the king may not move through check. Finally, every double move must alter the position; a player may not move a piece from a square then back to that same square.

    For this case, there is a special extension of the "en passant" rule: if the twice-moved piece could be captured on the first square it stops on (which is illegal if one is double-moving the king), then after it's been moved the second time, the opponent may capture it "en passant" just as is done currently with Pawns making a two-square move through the capturing square of an opposing pawn. The en passant move must immediately follow the double move or the opportunity for en passant capture is lost.

  7. For OTB play, neither player may be more than a few steps away from the board on their opponent's time, in case the opponent wishes to execute a double move. Players may only leave the game area on their own time (this may require some lengthening of time controls for Option Chess).

  8. Consecutive double moves are allowed as long as the player has tokens remaining.

  9. A player's remaining double move tokens may not be hidden from plain view.

  10. In OTB play placing a double move token green-side up next to your side of the clock indicates you are on a double move, and doing this is irreversible once you release the token. Each move of the double move is subject to the same rules as a single move with respect to touching a piece and releasing it. For a software or online version of the game, the player would click a green token button which would then turn red, and the player's move would not be completed until s/he had made a legal double move.

Some notes:

Restriction (6) will change everything when it comes to calculations involving passed pawns reaching promotion squares. Attacking the promotion square once isn't enough if the pawn can promote and then a second piece can capture the promotion-square defender! But of course, that can't happen if you used all your available double moves earlier in the game...

Restriction (7) requires a bit of an explanation. Without it, this could happen: your opponent steps away from the board. You initially play a single move and hit your clock. But while your opponent is away, you might still be analyzing on his or her time and realize after five more minutes that you have an excellent second move. You could then surreptitiously place your green double move token by the clock, switch the clock back to your time, and instantly make the second move, then hit your clock and flip the token. This means you made your second move on your opponent's time, at some risk of being caught.

I hope you have enjoyed my presentation and it gets you thinking about playing Option Chess! I am considering authoring a rudimentary Zillions of Games engine to play Option Chess. By the way, time controls for Option Chess should generally be long. I can’t imagine such a thing as a Blitz Option Chess game…

Solution to the problem given above:

Black to play and mate in three

First let’s examine two options that do not work: 1...Qh3, Re1+. At first glance, it appears the queen cannot be captured because only one move of a double move may capture. Because of the check the black rook must be captured, while a non-capturing, non-checking move may being played first. However, this is mistaken. White can play 2.Ng1, Nfxh3, and all of Black’s threats are gone. So White didn’t have to capture the rook, he could block it instead, which allows capturing of the queen.

Here's the second option that doesn't work: 1.... Qh3, Qg2+. This is a legal move, but it allows the special en passant capture of 2.Nxh3e.p. Also take note that 1.... Qh3, Qxf3+ is an illegal move because if a single piece is moved twice, neither of its moves may be a capture.

The solution begins with 1…Qh3, Bxf2. Note that White’s king has not been put in check, but there are mate threats on g2 and f1. And in this case, Black’s queen cannot be captured. White cannot play 2.Ng5, Ngxh3 because a piece being moved twice cannot capture at all, not even on the second move. White could try something desperate like 2.Rd1, Rg1, but this is taken care of by 2…Qxf3+. White could try 2.Rd1, g8=Q but this runs into 2…Qxf3+ and if now 3.Rg1, Qg2, Black has the incredible 3…Re1,Qxg2#.

Even the remarkable 2.Rd1, Rxe3 doesn't quite work despite removing the powerful knight. Black cannot play 2.... Re1+, Qf1+ because both moves are checks. But Black has immediate mate anyway with 2.... Be1, Qg2#. The Be1 move is cute, uncovering the rook to support the Queen's mating move. And White cannot play 3.Nh4, Nxg2 because the second move is a capture making the move illegal.

White’s best chance seems to lie with 2.g8=Q, Qff7+. This attempts to avert mate by chasing the Black king with checks. But alas, there is a white pawn on a3 that prevents either rook from getting in some checks and contributing to this effort. And so Black replies 2...Ka6, Qxf3+.

Now White is reduced to desperado measures and Black still has one Option token remaining: 3.Qg2, Qb7+ trying to stop the one-move mate on g2. But Black uses that last Option token to play 3…. Ka5, Qxg2#. This is mate because the other White queen can’t make the ricochet move 4.Qg7, Qxg2 because the rules stipulate moving a piece twice disallows either move from being a capture.

Just in time, Black uses his last option to secure mate. The very first Option Chess problem, and a very fascinating one! Anyone wishing to discuss this problem or any games of Option Chess can contact me directly: pwb (at) twitoftheyear (dot) com.

About the author Paul Bonham

  • Born in Toronto, Canada
  • Diploma from DeVry Institute in Electronics Engineering
  • Self-taught in C and C++, Python, SQL
  • Briefly played tournament chess in Canada between 1989 and 1995
  • Arrived in Boston MA in December 1996 to work a three-month TN Visa contract
  • Met and married Julie Maxfield from Concord, New Hampshire
  • Continued working in USA and eventually acquired US Permanent Resident status
  • Twice worked a year at Microsoft in Redmond WA as contract software engineer
  • Amateur guitarist/bassist/keyboardist, recorded 21 original compositions
  • Interests include AI, highly concurrent software, game programming, computer graphics, chess variants, poker
  • Inventor of FET Bender and Ultrasonic Muffler

See also

2/2/2014 – A new, challenging chess variant
Ever since desktop computers can play at its highest levels and beat practically all humans, the interest of the Artificial Intelligence community in this game has been sagging. That concerns Dr Azlan Iqbal, a senior lecturer with a PhD in AI, who has created a variant of the game that is designed to rekindle the interest of computer scientists – and be enjoyable to humans as well: Switch-Side Chain-Chess.

2/13/2014 – Ken Regan's Tandem Pawn Chess
"Some years ago Frederic Friedel of ChessBase asked for a game that was reasonably like chess but would give humanity a chance against today's computers," writes IM Ken Regan, Associate Professor at the University of Buffalo, NY. He has produced a new chess variant in which two pawns are strapped together initially. The aim is to make a game that is harder for computers to play.


Reports about chess: tournaments, championships, portraits, interviews, World Championships, product launches and more.

Discuss

Rules for reader comments

 
 

Not registered yet? Register