1982
In 1982 I was thirty five and I had been indulging in inventing abstract games for a couple of years and playing them at the games club Fanaat of the University of Twente, then still called the 'Technische Hogeschool Twente'. I was inventing up against the late Martin Medema of whom mindsports features the ultra capricious game Explocus. He was the inventor of the notorious multi player game Atlantis that gave rise to Mu and Storisende. It was the year in which my then still seven years old son Demian, who often would come along to Fanaat, invented Congo. We had a great time and in great company, including Ed van Zon, the co-inventor of Emergo, with whom I would later start mindsports.
If my memory serves me well, I had by that time already invented Havannah, the hexagonal Draughts variant HexDame, the contact capture variant Bushka (I was a huge fan of Kate Bush), territory games like Medusa and Phalanx, the rotational chess variant Rotary, the fairy variant Chakra (with Ed) and some miscellaneous stuff.
A simple design goal
By that time Martin had started to drift into thematic design with hidden information and random events, while I went in the opposite direction. Thus I had started to consider chess variants in a more fundamental way, which eventually led to a question:
What is actually needed to make a chess variant and what can be omitted?
You need a king, obviously, a piece of absolute value, regardless of whether the goal is checkmate or the actual capture of it. Would you need pieces? With only two kings I found it hard to imagine how to capture, so I decided to have a uniform set of pieces and the rook seemed basic enough. But would these rooks be able to capture one another? Since Chess is all about chasing and capturing a king, and not necessarily about slaughtering each others pieces, I decided against it.
So here I was, a king and rooks that could only block rival rooks. In my mind they started swarming all over the place even before I had begun considering how many of them there would be and on what board size and with which starting position. It looked like there would be a lot of blocking and shifting and no permanency. Kings played hide and seek and on the face of it they seemed hard to pin down.
The castle
That's when I decided to lock the kings up in their respective 3x3 castles. That strongly suggested to put eight rooks around them and thus the initial positions were settled. The king would not be allowed to leave the castle, so you would always know where to find it. Inside its domain it would have the king's move or the knight's move at it's disposal, for maximal flexibility.
A bump in the road
Ed and I used a 10x10 Draughts board to try the game because these are common in the Netherlands. You see an uncheckered version on the left.
In the first try-out Ed moved each of his four rooks that were on the edges of his castle one step outward, forming a 3x3 diagonal square around the king as shown. Then he started moving his king around on the remaining five vacant squares. The point was clear: without mutual capture between rooks the game was stuck. What to do?
The wall
Back to the drawing board. I kept my trust in the concept but concluded that some form of mutual capture would be inevitable, though it would have to be very restricted to prevent an all out mutual slaughter. Then a real life picture rose in my mind, a somewhat medieval one, of attackers on the outer wall of a castle who were shooting at defenders inside, and vice versa. As it turned out it would be exactly the right solution.
***** ***** *****
The rules of Chad
The diagram shows the Chad board with the pieces in the initial position. The areas covered by the pieces are called the castles Each castle has twelve adjacent squares that together constitute the wall.
White begins. Players move, and must move, in turn.
The king is confined to his 3x3 castle. He may go and capture using either the king's move or the knight's move.
Note: It's customary to look at the king in terms of the squares it does not cover. In the center it covers the whole castle, on the side it does not cover the square on the opposite side, and in the corner it does not cover the other corner squares.
The rook moves as the rook in Chess, unhindered by castles and walls.
If it ends its move inside the opponent's castle, it is promoted to queen.
The queen moves as the queen in Chess, unhindered by castles and walls
A king facing an opponent's rook along a rank or file is in check. A king facing an opponent's queen along a rank, file or diagonal is in check. Kings may not move into check and if they are in check they must get out of it by moving, interposition or capture of the piece that gives check.
Barring one particular situation, pieces other than the king cannot capture other pieces and thus only block one another. The mutual right of capture between pieces exists, and only exists, between an attacking piece that is on the opponent's wall, and a defender inside its own castle.
Checkmate is a win, stalemate and 3-fold are a draw.
***** ***** *****
Either side to move can checkmate in 2
Basic heuristics
The penultimate rule is the defining and crucial one! It is illustrated in the diagram on the left. Black's castle shows a white rook on the wall facing a defender inside. Now both have the right to capture.
But in this specific situation only White can capture because the black rook is pinned! The position shows one of the basics of attack. What can Black do? His only option is to move the defending rook towards the pinning one. But this leaves a white rook on the wall attacking three squares inside the castle - a thorn in Black's side.
The white rooks illustrate a basic attacking pattern. It appears in a variety of forms in almost all attacking concepts.
If it is White's turn he can checkmate in two:
1. H7xH9 - ad lib; 2. H9-I9 checkmate
However, if it were Black's turn he also could checkmate in two.
A related basic concept is the promotion sacrifice. It derives from the fact that an attacker, once it is inside the castle (and thus automatically a queen), can only be captured by the king. A king on the side leaves one square unprotected and a king in the corner three. Sacrificing a piece to force the king to the side or into the corner, to clear the way for a second piece to promote on an unprotected square, is a common heuristic. Getting a queen is worth the sacrifice of a piece anytime!
It's worth noting that exchanges are possible: an attacker on the wall captures a defender inside the castle and the king captures the attacker. But the attacker has to get on the wall first, so unless the attacked piece is pinned, he is exposed to capture himself. Reversely, if a defender captures an attacker, which inherently will be on the wall, then it can not be recaptured. Add the fact that the king can attack an invader using the knight's move but can never be attacked by the knight's move itself, and it will be clear that all manoeuvring around and inside the castle is very tricky indeed. It's good to realise that all attacks eventually draw from defending forces, so if one goes for checkmate, it should be driven home. If it fails, 3-fold is one's only hope!
The irony
I was satisfied with the game, very satisfied. At Fanaat it hit the ground running. There were two exceptionally good players, Wim van Weezep and Mark Waterman, who after a few weeks had only each other for opponent because they wiped everyone else off the board. In retrospect the game didn't meet modern design criteria like finitude and drawlessness but it was 1982 and we didn't care, and actually I still don't. Chess variants almost inherently have forced cycles which puts an end to finitude. I'm sure that Chad on the elusive 'hypothetical top-level' that inventors often refer to, will be draw prone. But no-one has as yet played it on that level, and probably no-one ever will. Down here on Earth it's a recreational game and in that quality it is finite and decisive enough.
A few weeks later I was fumbling around with a new game of breathtaking simplicity that Fanaat had acquired, called "Isola". It was my first encounter with a game that had its playing area sinking away square by square, a procedure that came to be known as the 'Atlantis effect'. There were only two opposing pieces and every turn you had to move your own piece with the king's move and push out a vacant square. The game crawled its way at snails pace to its predictable conclusion: someone would get stuck.
I like simplicity but this, I thought, carried it too far. It was a perfect kindergarten game but I wondered for whom Fanaat had bought it. So contrary to my usual approach I actually tried to complicate things a bit.
I introduced a 'jumper' that could remove squares by jumping over them. But having only one of them seemed too restrictive, so I added a second one to have them combine their efforts. Then I thought it would be nice to disallow the main piece to be adjacent to an opponent's piece. Then I suddenly realised I had a chess variant! Just like that and totally unintentional. The irony immediately struck me: it was far simpler than Chad. This game had a 7x7 board and two non-royal pieces per side, that's one third of the squares of Chad and a quarter of the pieces. As a bonus it did away with promotion and mutual capture of pieces. Someone Up There was making fun of me!
The event changed both my view on and my approach of inventing games. I became more of a hunter than an inventor, trying to discover rather than to design and less preoccupied with the outcome. Intended goals are all fine, but they can also hamper the associative process. In my case associations tend to go in every direction and a lot of it is unplanned and unforeseen. Too much focus on an intended result can make you disregard ideas and possibilities that present themselves along the way. Paying attention beats thinking! I've had a lot of 'accidental' inventions since then, like Dameo, Symple, Starweb, Multiplicity and Lox, to name a few of the more prominent ones.
***** ***** *****
The rules of Shakti
In the initial position the board is covered with 45 tiles. Each player has a king and two warriors. During play the number of tiles is bound to be reduced due to the 'atlantis effect'.
All play is on the tiles.
White begins. Players move and must move in turn (unless they cannot: in a stalemate position a player must pass).
The King, if not in check, may move to the first tile he sees in any of eight directions as shown in the diagram.
If in check, the king is restricted to adjacent tiles. Anticipating on the warrior's move, the diagram on the right shows that pieces giving check from a distance therefore need no protection. It follows that the king can only capture an unprotected warrior on an adjacent tile.
The evil stare rule:
Kings may not see one another along the same rank, file or diagonal, with no tiles in between, so neither player may effectuate that situation. Thus a king may protect a piece against capture by it's counterpart. A king protects a warrior it sees, as in the diagram, where the black king, in check, must move to the indicated tile.
The Warrior:
A warrior too may move to the first tile it sees in any of eight directions. If both are vacant, a warrior may also move to the second tile, removing the first. The removal is compulsory, but of course the player may not, in doing so, put his own king in check. The diagram shows the warrior's options. If he chooses the second target-tile in any direction, the one jumped is removed. The black king is not in check because the vacancy-condition is not fulfilled.
Warriors are strictly king oriented and cannot capture one another.
Stalemate is not a draw in Shakti: if a player cannot move, his opponent may move instead.
In case you wonder, the corner tiles have been removed to make it harder to create a little fortress in a corner and play for a draw. Draws remain possible of course and at a 'hypothetical top-level' they may be common because the game's theoretical truth can hardly be other than a determined draw. But in a recreational context the game is far from trivial.
***** ***** *****
Chad problems
White checkmates in 6 moves
Thanks to Chris Huntoon for checking the Chad problems.
1. D7-H7
threatening 2. H7xH9 ++
... I8-H8
on H9xH7 follows 2. G3-G9 and then on ... H7-H9 3. G9xH9 Q++ and on ... I9-I10 3. E5-E9 checkmating next move on H9 or I9 as the case may be. But Black can postpone the inevitable.
2. H7xH8 Q+; I9-J9
on I9-I10 follows 3. G3-G10 and then on ... H9-H10 4. H8-I8 ++ and on ... I10-J9 4. G10xJ10 Q++
