## Solitaire abstract game

Marrakesh (1978), a game by Joli Quentin Kansil, combines Backgammon bearing-off with the trick-taking of card games. The author was a competitive player of Backgammon and Bridge and designed also Bridgette (1970), which is a kind of Bridge for two. JQK has published many other games, among which at least Marrakesh and Bridgette are enduring classics, in my opinion.

Original copies of Marrakesh are very difficult to find. Nevertheless, if you enjoy Backgammon or you like games of bluff, you should not miss this great game. Fortunately, Marrakesh is easy to play with a Backgammon set, two regular decks of cards, some additional markers, and some additional dice—components available in almost every gamer's collection. On the other hand, it is even better to have a dedicated Marrakesh set, and I designed my own board and used commercial photofinishing on an aluminum sheet, as I did with MeM, described in

Marrakesh is essentially a game for two. In this form, it is fast and thrilling. Luck does play a large role, and there can be rapid changes of fortune amid plenty of opportunity for bluffing or outguessing your opponent. I have played Marrakesh in this form, and I find it enjoyable. However, I read a post in BoardGameGeek about Solitaire Marrakesh, and I decided to give it a try. The solitaire version plays exactly the same as the two-player version, except that the AI opponent makes completely random moves.

Unexpectedly, the AI was remarkably strong! I lost the first three games soundly, as well as the fifth, although I pulled back in the fourth and sixth games. The total score over all six games was 285 points for me and 301 points for my AI opponent. Six games is a small sample, but after many more attempts the AI still plays a good game. Remember, I am trying hard to win, using my best understanding of the strategy, whereas the AI is moving completely randomly. Is it possible that random play is a reasonable strategy, if not always a winning strategy? Perhaps I am just not very good at Marrakesh. Either way, Marrakesh is an engaging solitaire abstract game.

The rules of Marrakesh, below, may seem complicated. However, the use of the ovals corresponding to tricks is very logical, as are the rules for matching cards and bonus cards. The genius of Marrakesh is how the various elements of the game fit together to construct an intricate mechanism that ticks along like clockwork. Once you understand the game, it is simple and straightforward.

The reworking below of the rules of Marrakesh is based on the scan of the 1984 rules in BoardGameGeek. My version below is more fleshed-out than the original, and hopefully brings out the logical structure of Marrakesh better. However, please feel free to refer to original rules, too. To simplify the presentation, I have assumed that the reader is familiar with the process of moving and bearing off pieces in Backgammon. Whereas Backgammon moves depend on dice throws, Marrakesh uses numbers on playing cards for the process of moving and bearing off pieces, not pips on dice. Dice are used in Marrakesh, but only to determine the initial setup of the board.

Here is my Marrakesh board design:

Original copies of Marrakesh are very difficult to find. Nevertheless, if you enjoy Backgammon or you like games of bluff, you should not miss this great game. Fortunately, Marrakesh is easy to play with a Backgammon set, two regular decks of cards, some additional markers, and some additional dice—components available in almost every gamer's collection. On the other hand, it is even better to have a dedicated Marrakesh set, and I designed my own board and used commercial photofinishing on an aluminum sheet, as I did with MeM, described in

*AG17*. I use six red discs and six blue disks for the pieces and five black cubes for the markers. It is nice to have six blue dice and six red dice, although you can get by perfectly well with fewer dice, which are re-rolled several times. Two decks of cards shuffled together, with some of the cards stripped out, complete the set. The design is shown in the diagrams below.Marrakesh is essentially a game for two. In this form, it is fast and thrilling. Luck does play a large role, and there can be rapid changes of fortune amid plenty of opportunity for bluffing or outguessing your opponent. I have played Marrakesh in this form, and I find it enjoyable. However, I read a post in BoardGameGeek about Solitaire Marrakesh, and I decided to give it a try. The solitaire version plays exactly the same as the two-player version, except that the AI opponent makes completely random moves.

Unexpectedly, the AI was remarkably strong! I lost the first three games soundly, as well as the fifth, although I pulled back in the fourth and sixth games. The total score over all six games was 285 points for me and 301 points for my AI opponent. Six games is a small sample, but after many more attempts the AI still plays a good game. Remember, I am trying hard to win, using my best understanding of the strategy, whereas the AI is moving completely randomly. Is it possible that random play is a reasonable strategy, if not always a winning strategy? Perhaps I am just not very good at Marrakesh. Either way, Marrakesh is an engaging solitaire abstract game.

The rules of Marrakesh, below, may seem complicated. However, the use of the ovals corresponding to tricks is very logical, as are the rules for matching cards and bonus cards. The genius of Marrakesh is how the various elements of the game fit together to construct an intricate mechanism that ticks along like clockwork. Once you understand the game, it is simple and straightforward.

**Marrakesh Rules**The reworking below of the rules of Marrakesh is based on the scan of the 1984 rules in BoardGameGeek. My version below is more fleshed-out than the original, and hopefully brings out the logical structure of Marrakesh better. However, please feel free to refer to original rules, too. To simplify the presentation, I have assumed that the reader is familiar with the process of moving and bearing off pieces in Backgammon. Whereas Backgammon moves depend on dice throws, Marrakesh uses numbers on playing cards for the process of moving and bearing off pieces, not pips on dice. Dice are used in Marrakesh, but only to determine the initial setup of the board.

Here is my Marrakesh board design:

You can see that the main part of the board is half a Backgammon board, in which each player controls six "points." The players each have three ovals beside the board that are used to keep track of pieces they "bear off" from their side of the board. I use the five stars across the middle to hold five

Each triangular point is a space that can hold one or more pieces. The pieces on a point are stacked end to end, so that you can easily see how many there are. The red pieces can only occupy the six points on one side of the board, the top in our diagrams; whereas the blue pieces can only occupy the six opposite points, the bottom in our diagrams. The points on each side are numbered mentally from 1 to 6, counting from points closest to the ovals. The numbers on the cards correspond to the numbers of the points and/or the number of points a piece can move down, from a higher-valued point to a lower-valued point, as in Backgammon. The Queens have no point value and do not permit any movement of pieces. The Queens are effectively zero-numbered cards.

The players decide who will be Blue and who will be Red. Blue throws the six blue dice, Red throws the six Red dice, and the players distribute their six playing pieces on their side of the board according to the pips on the dice. In my set, the five null chips are placed on the stars in the centre of the board. If Red throws 4-5-5-5-6-6 and Blue throws 1-1-2-4-4-6, the starting position is as follows:

*null chips*, for which I use black cubes. The original rules specify three null chips, but sometimes more are needed, and five fits with the design of the board. The players each have six pieces in their colour, which should be flat and stackable, like Backgammon pieces or checkers—the diagrams use red and blue pieces. The players each also have six regular dice in their colour, although you can get by with fewer dice that are shared or rolled several times. Lastly, a special deck of cards is needed, consisting of the Ace through Six from two regular decks together with four Queens, one of each suit—again a total of 52 cards, although of quite different composition from a regular deck.Each triangular point is a space that can hold one or more pieces. The pieces on a point are stacked end to end, so that you can easily see how many there are. The red pieces can only occupy the six points on one side of the board, the top in our diagrams; whereas the blue pieces can only occupy the six opposite points, the bottom in our diagrams. The points on each side are numbered mentally from 1 to 6, counting from points closest to the ovals. The numbers on the cards correspond to the numbers of the points and/or the number of points a piece can move down, from a higher-valued point to a lower-valued point, as in Backgammon. The Queens have no point value and do not permit any movement of pieces. The Queens are effectively zero-numbered cards.

The players decide who will be Blue and who will be Red. Blue throws the six blue dice, Red throws the six Red dice, and the players distribute their six playing pieces on their side of the board according to the pips on the dice. In my set, the five null chips are placed on the stars in the centre of the board. If Red throws 4-5-5-5-6-6 and Blue throws 1-1-2-4-4-6, the starting position is as follows:

Once the initial setup is determined, the deck is shuffled and six cards are dealt one-by-one and face-down to each player. It does not matter who deals. The undealt portion of the deck is placed face-down close by; the top card may be drawn as a bonus card during play. The players look at their cards, but keep their values hidden from the opponent. The six cards in each player's hand will be played to six "tricks"—each trick will correspond to one of the six green ovals. The play of each trick will result in pieces or null chips placed in one of the ovals, one trick per oval. Each player has three ovals, corresponding to three tricks. A player's pieces are "borne off" from the points, which means they are moved from the points to the ovals when a trick is won. A player is trying to bear off all her pieces. With three ovals, a player has three chances to bear off all six pieces for

The

The leader selects a card from her hand and places it

When the hand is complete, with the play of up to six tricks, the setup is repeated by again throwing the six dice each and placing the red and blue pieces and null chips back on the board. The players are dealt another six cards each from the top of the deck, without shuffling in the used cards. A full game lasts for 12 rounds in this manner, where the deck is shuffled again with all used cards only every three rounds.

The winner of a trick is determined by comparing the

At the end of the round, all cards played to tricks and any cards remaining in the players' hands are gathered up into a discard pile, where only the top card is visible, before the next hand is dealt. Every three hands, as mentioned above, the remaining cards in the deck, together with used cards in the discard pile, are all shuffled together again before dealing.

The winner of the trick moves or bears off her pieces according to the numbers on the cards played to the trick. The winner of a trick may also draw a bonus card and bear off or move further piece(s). A player must utilize the numbers on the trick cards and any bonus card to move or bear off pieces whenever possible. To bear off a piece means to move it from a point to the specified oval for this trick, where the point number matches the number on a card. The two cards played to a trick and the bonus card, when drawn, all contribute to bearing pieces off to a single oval. A piece is moved from a higher numbered point to a lower numbered point, as in Backgammon, except that pieces are restricted to their side of the board. As in Backgammon, if there are no pieces on points equal to or higher than the value on a card, then a piece may be moved or borne off instead from the next highest point that

Depending on whether the two cards played to the trick do not match in suit or number, match in suit only, match in number only, or match in both suit and number, there are four possibilities:

*game*. In the case of*gammon*or*backgammon*, explained below, a player needs fewer than three ovals to bear off all six pieces.The

*leader*to the first trick is the player with most pieces on her 1-point. If this is equal, the leader is the player with most pieces on her 2-point, and so on. If the players’ distribution of pieces is exactly equal, they both have to re-roll their dice, and place their pieces again.The leader selects a card from her hand and places it

*face down*in front of her. The other player, the*receiver*, selects a card to play*face up*. Then, the leader’s card is then turned face up and the two cards are compared to see who has won the trick. The winner of the trick may take some of her pieces or a null chip to place in one of her ovals. (Null chips can sometimes also be placed in an opponent's oval—see below.) The first trick that a player wins utilizes the first oval nearest the winning player; the second trick uses the next oval; the third trick utilizes the third and final oval. The winner of a trick leads to the next, and the process is repeated for all six tricks. The leader always plays a card face down; the receiver always plays a card face up. The cards played to tricks are not collected by the winner, but are left face up in front of the players for the duration of the round.When the hand is complete, with the play of up to six tricks, the setup is repeated by again throwing the six dice each and placing the red and blue pieces and null chips back on the board. The players are dealt another six cards each from the top of the deck, without shuffling in the used cards. A full game lasts for 12 rounds in this manner, where the deck is shuffled again with all used cards only every three rounds.

The winner of a trick is determined by comparing the

*suits*of the two cards. Spades is the highest suit, followed by hearts, diamonds, and lastly clubs, the same order as Bridge. However, if a club and a spade are played to the same trick, the club wins.*If the two cards have the same suit, the receiver always wins.*At the end of the round, all cards played to tricks and any cards remaining in the players' hands are gathered up into a discard pile, where only the top card is visible, before the next hand is dealt. Every three hands, as mentioned above, the remaining cards in the deck, together with used cards in the discard pile, are all shuffled together again before dealing.

The winner of the trick moves or bears off her pieces according to the numbers on the cards played to the trick. The winner of a trick may also draw a bonus card and bear off or move further piece(s). A player must utilize the numbers on the trick cards and any bonus card to move or bear off pieces whenever possible. To bear off a piece means to move it from a point to the specified oval for this trick, where the point number matches the number on a card. The two cards played to a trick and the bonus card, when drawn, all contribute to bearing pieces off to a single oval. A piece is moved from a higher numbered point to a lower numbered point, as in Backgammon, except that pieces are restricted to their side of the board. As in Backgammon, if there are no pieces on points equal to or higher than the value on a card, then a piece may be moved or borne off instead from the next highest point that

*does*contain a piece or pieces. The number on a card may be utilized multiple times to move various pieces—see below.Depending on whether the two cards played to the trick do not match in suit or number, match in suit only, match in number only, or match in both suit and number, there are four possibilities:

*No match in suit or number*

*For example, in the diagram above, Blue leads because Blue has two pieces on the 1-point, whereas Red has none. Suppose Blue leads 6♠️ and Red plays 2*♣️.*Red wins the trick and bears off a piece from the 6-point. Then, Red has no piece to bear off from the 2-point, so Red moves a**piece two spaces, from**the 5-point to the 3-point. (Otherwise, Red could move a piece from the 6-point to the 4-point or from the 4-point to the 2-point.) The diagram below left results.**Match in suit, not number*

*four*times to move and/or bear off pieces, as if a double were thrown in Backgammon. If the bonus card matches the card the winner (i.e., the receiver) played to the trick in both suit and number, the winner (i.e., the receiver) uses the number on the bonus card

*six*times to move and/or bear off pieces.

*For example, from the new position, suppose Red now leads 3♦️ and Blue plays 5♦️. Blue wins, because receiver always wins a suit-match. Blue cannot bear any pieces off. Blue moves a piece from the 6-point to the 1-point and a piece from the 4-point to the 1-point. Blue now collects a bonus card and draws 5♣️, which matches 5♦️ in number, not suit. Therefore, Blue now has 5 to move four times. As in Backgammon, if you have no pieces on points on or above the number to move, you can bear off from the next point below the number to move. Therefore, Blue bears off a piece from the 4-point, then from the 2-point, and lastly two pieces from the 1-point, resulting in the position above right. Note that all four blue pieces borne off are stacked on Blue's first oval.*

*Match in number, not suit*

*four*times, according to the number on the two cards, as if a double were thrown in Backgammon, and then draws and discards a bonus card. If this card differs in suit and number from the card the winner played to the trick, its number is used once to move and/or bear off a piece. If the bonus card matches the card the winner played to the trick in either suit or number (not both), the winner uses the number on the bonus card

*four*times to move and/or bear of pieces, as if a double were thrown in Backgammon. If the bonus card matches the card the winner played to the trick in both suit and number, the winner uses the number on the bonus card

*six*times to move and/or bear of pieces.

*Suppose Blue now leads Q*♥️

*and Red responds with Q♠️. Red wins, and the cards match in number (i.e., both zero) but not suit. Red does not move or bear off pieces due to the Queen pair, because four zeros is still zero. However, Red receives a bonus card and draws 2♦️. The bonus card does not match Q♠️ in number or suit, so Red gets to play 2 only once. Red moves a piece from the 6-point to the 4-point. Red has not borne off a piece, despite winning a trick, so Red places a null chip on the next oval, the second. The resulting position is shown below left.*

This is the first example of the use of the null chips. A null chip is used whenever a player wins a trick but does not bear off any pieces. Null chips are used in three different situations. In the first case, a player is unable to use the card values that result from winning a trick to bear off any pieces. Perhaps the card values simply do not permit bearing off, even when the player moves pieces on the board, as in the example here. The player must place a null chip on the corresponding oval instead.

*Match in suit and number*

*six*times, according to the number on the two cards, and then draws and discards a bonus card. Regardless of any suit and/or number match between the bonus card and the card the winner (i.e., the receiver) played to the trick, the winner (i.e., the receiver) uses the number on the bonus card

*four*times to move and/or bear off pieces, as if a double were thrown in Backgammon.

Here is one point where the inexorable logic of the Marrakesh mechanism falters, in my opinion. If the bonus card does not match in suit or number, I would restrict to just once moving or bearing off according to its value. If the bonus card matches in suit or number I would allow use of the number on the bonus card

*four*times. The bonus card cannot match in both suit and number again because there are only two copies of each card (and only one copy of each Queen). I usually play the variant where a non-matching bonus card gives only one move, even though this reduces slightly the prevalence of backgammons—see below.

*Suppose Red leads 3♣️ and Blue replies with 3♣️. Blue wins, because the receiver always wins when the two cards of a trick match in suit. However, the cards also match in number. Blue could play the value 3, six times, although Blue only needs two 3's to bear off the remaining two pieces. Blue would now draw a bonus card and compare it with his 3♣️. However, all his pieces have been borne off and in these circumstances the bonus card is not drawn. A bonus card is not drawn if it cannot be used because all six pieces are already borne off.*

*Lastly, suppose Blue leads with A♠️ and Red responds with 6♦️. Blue wins, but cannot bear off any more pieces because all are already all borne off. Blue instead puts a null chip in Red's third oval, and the round finishes. The final position is above right.*

This was the second example of the use of null chips. Again, a null chip is used whenever a player wins a trick but does not bear off any pieces. In the second case, a player is unable to use the card values that result from winning a trick to bear off any pieces, because all the player's pieces are already borne off. The player places a null chip on the opponent's next oval instead.

The process of deciding how many times to count a card value and how to use bonus cards may seem complex, but it is completely logical, aside from the one point I mentioned above, where the trick cards match in number and suit. A summary of the information is given in the table below, with the variant I suggest in purple. The leftmost column shows matching possibilities of the two trick cards, whereas the columns to the right show matching possibilities of the bonus card.

"-" means no match, "S" means match in suit not number, "N" means match in number not suit, and "SN" means match in suit and number.

Trick |
Values |
Bonus |
_ |
S |
N |
SN |

_ |
x1 |
No |
NA |
NA |
NA |
NA |

S |
x1 |
Yes |
x1 |
x4 |
x4 |
x6 |

N |
x4 |
Yes |
x1 |
x4 |
x4 |
x6 |

SN |
x6 |
Yes |
x4 [x1] |
x4 |
x4 |
NA |

If a player bears off all pieces in the first two ovals, with the first two tricks the player has won, then that player has achieved a

Now, the six cards in each player's hand at the start of a round correspond to six tricks, and each trick corresponds to exactly one oval. Whenever a trick is played, pieces or a null chip will come to occupy one of the ovals. Pieces cannot be borne off to an oval occupied by a null chip, and once a player has all three ovals occupied by pieces or null chips, no further pieces can be borne off. If a player does win another trick, then the player places a null chip in the opponent's next oval, which is a third example of the use of null chips—null chips are always needed to mark an oval when a trick is played and no pieces are borne off.

I do not think the 1984 rules deal explicitly with this third example of the use of null chips. Nevertheless, the logic of Marrakesh is inexorable. If your three ovals are already full of pieces and/or null chips, and you win a trick, then an oval must be occupied for this trick, even though all your pieces are not borne off. The only oval that can be occupied is an opponent's, and because you are the winner of the trick, it must be a null chip that goes to the opponent's oval—even if this seems a little unfair, as you still have pieces on the board, not borne off.

When a player has achieved a gammon, then the player's third oval is not used, and only five tricks are played. If both players achieve a gammon only four tricks are played. Likewise, if a player achieves a backgammon in the first oval, the number of tricks is reduced by two. If both players achieve backgammon in their first ovals, only two tricks are played! At the end of the round, if there are any gammons or backgammons, cards not played to tricks are simply discarded face up on top of the discard pile.

At the end of the round, both players score. Points are earned

If a player bears off all six pieces, that player score 3 points for game. If a player bears the six pieces off for game in specific patterns in the three ovals, as shown in the table below, then that player scores higher for game, as indicated. For example, with the first two ovals occupied with one piece, and the third oval occupied with four pieces, the player scores 9 points instead of 3 points for game, for achieving the Bogart pattern. The base score for a gammon is 6 points. However, if a player achieves a gammon in one of the two patterns below, the player scores 9 points instead of 6 points for gammon. Backgammon always scores 12 points, and the opponent does not score defensively for any null chips accompanying the backgammon. The X's in the table simply indicate unused ovals because all pieces have been borne off; the 0's represent null chips.

*gammon*. A gammon must involve the first two ovals. A player has not achieved a gammon, for example, if he has a null chip in the first oval, even if all pieces are borne off in the second and third ovals. Blue in the game above has won a gammon. If a player bears off all six pieces in a single oval, with a single trick won, then that player has achieved a*backgammon*. For a backgammon, a player is permitted to have null chips in the first or even the first and second ovals, as long as all six pieces come off in one play in a single oval. If a player achieves a backgammon, any null chips in her ovals are ignored for scoring—see below.Now, the six cards in each player's hand at the start of a round correspond to six tricks, and each trick corresponds to exactly one oval. Whenever a trick is played, pieces or a null chip will come to occupy one of the ovals. Pieces cannot be borne off to an oval occupied by a null chip, and once a player has all three ovals occupied by pieces or null chips, no further pieces can be borne off. If a player does win another trick, then the player places a null chip in the opponent's next oval, which is a third example of the use of null chips—null chips are always needed to mark an oval when a trick is played and no pieces are borne off.

I do not think the 1984 rules deal explicitly with this third example of the use of null chips. Nevertheless, the logic of Marrakesh is inexorable. If your three ovals are already full of pieces and/or null chips, and you win a trick, then an oval must be occupied for this trick, even though all your pieces are not borne off. The only oval that can be occupied is an opponent's, and because you are the winner of the trick, it must be a null chip that goes to the opponent's oval—even if this seems a little unfair, as you still have pieces on the board, not borne off.

When a player has achieved a gammon, then the player's third oval is not used, and only five tricks are played. If both players achieve a gammon only four tricks are played. Likewise, if a player achieves a backgammon in the first oval, the number of tricks is reduced by two. If both players achieve backgammon in their first ovals, only two tricks are played! At the end of the round, if there are any gammons or backgammons, cards not played to tricks are simply discarded face up on top of the discard pile.

At the end of the round, both players score. Points are earned

*offensively*for bearing off your own pieces and*defensively*for restricting the opponent from bearing off pieces. A player’s score for the round is the sum of offensive points and defensive points.If a player bears off all six pieces, that player score 3 points for game. If a player bears the six pieces off for game in specific patterns in the three ovals, as shown in the table below, then that player scores higher for game, as indicated. For example, with the first two ovals occupied with one piece, and the third oval occupied with four pieces, the player scores 9 points instead of 3 points for game, for achieving the Bogart pattern. The base score for a gammon is 6 points. However, if a player achieves a gammon in one of the two patterns below, the player scores 9 points instead of 6 points for gammon. Backgammon always scores 12 points, and the opponent does not score defensively for any null chips accompanying the backgammon. The X's in the table simply indicate unused ovals because all pieces have been borne off; the 0's represent null chips.

*Offensive scoring*

Type of achievement |
Name |
Pattern |
Points |

Game |
- |
- |
3 |

Game |
Casablanca |
2-2-2 |
6 |

Game |
Rabat |
3-1-2 |
6 |

Game |
Bogart |
1-1-4 |
9 |

Game |
El Ayun |
1-4-1 |
12 |

Gammon |
- |
- |
6 |

Gammon |
Gibraltar |
3-3-X |
9 |

Gammon |
Tangier |
4-2-X |
9 |

Backgammon |
- |
6-X-X, 0-6-X, 0-0-6 |
12 |

If neither player achieves at least game, the players total the point values of all pieces they have left on the board. The player with the lowest total scores 1 point for Pips. Neither player scores for Pips if the total is tied.

Defensive scoring depends on the null chips and singletons occupying the opponent's ovals. If the opponent has 1 null chip, the player scores 1 for Chips; if the opponent has 2 null chips, the player scores 4 for Chips. As mentioned above, null chips accompanying a backgammon, in the first or second ovals, are ignored for scoring. If the opponent is restricted to one piece or a null chip in each oval, instead of scoring for Chips, the player scores for Fez. The scoring patterns for Fez are shown in the table below.

Defensive scoring depends on the null chips and singletons occupying the opponent's ovals. If the opponent has 1 null chip, the player scores 1 for Chips; if the opponent has 2 null chips, the player scores 4 for Chips. As mentioned above, null chips accompanying a backgammon, in the first or second ovals, are ignored for scoring. If the opponent is restricted to one piece or a null chip in each oval, instead of scoring for Chips, the player scores for Fez. The scoring patterns for Fez are shown in the table below.

*Defensive scoring*

Name |
Pattern |
Points |

Little Fez |
1-1-1 |
4 |

Common Fez |
1-1-0, 1-0-1, 0-1-1 |
6 |

Royal Fez |
1-0-0, 0-1-0, 0-0-1 |
9 |

Grand Fez |
0-0-0 |
18 |

In the round finished above, Blue scores 9 for a Tangier and 9 for a Royal Fez, a total of 18. Red does not score.

The highest possible score is 30, with 12 points for Backgammon or El Ayun and 18 for a Grand Fez. Joli Quentin Kansil calls this a

A full game consists of 12 rounds, in which the pieces are distributed and the cards are dealt in exactly the same way each time. A short game may be played with only six rounds. During the play of a round, the extra bonus cards drawn are discarded face up on a discard pile once they have been used. At the end of a round, all 12 cards dealt for the round are discarded face up on top of the same discard pile, along with any bonus cards. The discard pile should be squared so that only the top card is visible. The deck is shuffled again with all the discards only after every three rounds. Otherwise, the cards are dealt from the top of the deck without shuffling and without including the discards.

The winner is the player with the highest score after 12 (or six) rounds. If the score is tied, the winner is the player who scored in most rounds. If this also is tied, the winner is the person who scored highest in any one round. If, lastly, this is tied, the game is genuinely drawn. When scoring, it is best to record the scores for the round as well as the running totals, in order to facilitate checking in case of a tie.

The highest possible score is 30, with 12 points for Backgammon or El Ayun and 18 for a Grand Fez. Joli Quentin Kansil calls this a

*Marrakesh*.A full game consists of 12 rounds, in which the pieces are distributed and the cards are dealt in exactly the same way each time. A short game may be played with only six rounds. During the play of a round, the extra bonus cards drawn are discarded face up on a discard pile once they have been used. At the end of a round, all 12 cards dealt for the round are discarded face up on top of the same discard pile, along with any bonus cards. The discard pile should be squared so that only the top card is visible. The deck is shuffled again with all the discards only after every three rounds. Otherwise, the cards are dealt from the top of the deck without shuffling and without including the discards.

The winner is the player with the highest score after 12 (or six) rounds. If the score is tied, the winner is the player who scored in most rounds. If this also is tied, the winner is the person who scored highest in any one round. If, lastly, this is tied, the game is genuinely drawn. When scoring, it is best to record the scores for the round as well as the running totals, in order to facilitate checking in case of a tie.

**Good play**

The comments below are applicable both to the two-player game and the solitaire. I would guess that Marrakesh requires a similar level of skill to regular Backgammon, or maybe a little less because it lacks the grand-scale strategic motifs of Backgammon, such as the back game or the running game. Marrakesh certainly entails a great deal of luck, but it is interesting to me how it is apparently possible to make rational decisions about the best moves, even though this rational behaviour does not affect the winning percentage greatly—at least, in my experience of the game. A happy side effect of this property of Marrakesh is that you can always credit your own cunning, rather than luck, for a win.

For solitaire play, I have the AI play cards to tricks randomly, and always move to bear off the maximum number of pieces. If there is any choice in moving pieces on the board, I always have the AI move from the highest points. Thus, the play of the AI is completely random. However, it would be a mistake to say that the AI has no strategy. The AI is playing a consistent random strategy. The random strategy might even be effective now and then in the two-player game. In games such as Marrakesh, where two human players are trying to bluff or outguess each other, random play has the benefit of unpredictability.

The 1984 rules of Marrakesh that I have used as a guide contain a series of suggestions on good play by great Mid-Twentieth Century Bridge player, Oswald Jacoby. The reader can refer to the original file, but my notes here are a summary, and sometimes an elaboration, of the key points.

*Numbers to play*

Play card values where your opponent has empty points. Even if your opponent wins the trick, he will find it more difficult to bear off. On the other hand, play to your own points that have pieces on them, because if you win it will be easier to bear off. Jacoby gives the Marrakesh axiom, "When in doubt, lead low!" If your opponent wins the trick, the lower numbers will make it more difficult to achieve a gammon or backgammon.

*Play of the Queens*

The Queens are zero-valued cards and, in accord with Jacoby's Marrakesh axiom cited above about leading low, are the most useful defensive cards. Utilize Queens carefully, they are powerful cards!

*Suits to play*

The table below shows who wins when various combinations of suits are led by the two players (or by the human player and the AI). When the leader plays Spades or Hearts, the leader has two chances to win, but only one with a lead of Diamonds or Clubs. When the receiver responds with Spades or Hearts, the receiver on the other hand has three chances to win, and only two chances with a play of Diamonds or Clubs.

Spades and Hearts, therefore, are termed

Jacoby gives the example below left, where Blue has 6-1 to play. He recommends bearing a piece off the 6-point, but then moving a piece from the 6-point to the 5-point, rather than bearing a second piece off the 1-point. With this play, Blue spreads her pieces over more points. With pieces left only on the 6-point and 4-point, Blue risks a null chip with the next trick she wins. The move would have extra benefit for Blue if Blue had a 5-card in hand.

On the other hand, a point which Jacoby does not mention, the bearing off of only one piece rather than two leaves the player more open to her opponent scoring a Fez. I would play to take the two pieces off. Either way, this example indicates the type of thinking necessary for careful play in Marrakesh.

*major suits,*whereas Diamonds and Clubs are*minor suits.*Of course, in the two player game, the receiver might play a Club if she suspects the leader is about to play a Spade. All things being equal, it is better to play a major suit. Of course, it depends on the six cards actually in your hand.*Play to avoid null chips*Jacoby gives the example below left, where Blue has 6-1 to play. He recommends bearing a piece off the 6-point, but then moving a piece from the 6-point to the 5-point, rather than bearing a second piece off the 1-point. With this play, Blue spreads her pieces over more points. With pieces left only on the 6-point and 4-point, Blue risks a null chip with the next trick she wins. The move would have extra benefit for Blue if Blue had a 5-card in hand.

On the other hand, a point which Jacoby does not mention, the bearing off of only one piece rather than two leaves the player more open to her opponent scoring a Fez. I would play to take the two pieces off. Either way, this example indicates the type of thinking necessary for careful play in Marrakesh.

*Play to make scoring patterns*

The diagram above right gives one more example by Jacoby. Usually, it is best to bear off as many pieces as you can each turn, but as we saw above, there are exceptions. Suppose Red has led 2♣️ and Blue responds with 4♣️. Blue wins the trick, and Blue wants to aim for the 2-2-2 Casablanca scoring combination. If Blue bears off from both 4-point and 2-point, the bonus card will probably give Blue another piece off, spoiling the chance of a Casablanca. According to Jacoby, Blue should bear off the 4-point and move a piece from the 3-point to the 1-point—then hope to bear off exactly one piece with the bonus card.

An alternative consideration, again which Jacoby does not mention, is that bearing off fewer than the maximum number of pieces reduces your chances of a gammon, even though you are increasing the probability of achieving a scoring pattern. In this particular example, I think Jacoby is correct. Nevertheless, sometimes a fine balance exists between the two options to maximize your score.

There is more detail in Jacoby's own words, and I refer the interested reader there. Luck, I suspect, will often defeat even the most sophisticated Marrakesh strategy. The joy of Marrakesh is that the high luck factor does not seem to matter much. I highly recommend Marrakesh as a two-player game, and also as a solitaire. Marrakesh is a masterpiece from one of the late Twentieth Century's great game designers and promoters. ◾️