After Martin Medema had conceived the fabulous China Labyrint, various applications emerged, the most surprising of which was the discovery of its one-to-one relationship with the I Ching, the Chinese Book of Changes, eventually resulting in the I Ching Connexion.
Martin was experimenting with games where a transcendental lay-out was used as a playing area an idea that would in my opinion be very well suited for a videogame in an 'enchanted' castle. But is was long before the videogame explosion - we were still in the cardboard age.

The basic idea
At a certain point I saw the line patterns of the set suddenly as liberties. Those are the moments to stop everything else and start looking very carefully. A simple and new idea.

Can you see it? Liberies explicitly marked on the pieces itself, instead of implicitly given by the board? Groups interconnected by matching liberties and sharing the remaining ones. Capturing pieces or groups by taking the last liberty and ... and reversing the piece to show the same pattern in the captor's color, the captured group uniting its captors in one new group. It looked very promising indeed!

One of the implications was a double set of double-faced pieces. All 64 of them? One of the pieces of the set didn't even have liberties. Reversing a piece with no liberties renders the a piece with no liberties - it would forever be oscillating. Oscillation didn't seem a workable prospect, so the piece was out.
But interconnecting two pieces with one liberty each would also result in oscillation. This sprouted the first explicit rule: creating an oscillating group would be illegal.

There was one sobering fact: the average number of liberties of the set was three. That ain't much. How, for instance, to employ a piece with with one liberty? Pondering that question, the answer presented itself quite naturally: have it commit suicide by connecting up with an opponent's group, thereby reducing that group's liberties by one. Suicide as a weapon for killing groups - I must have been ahead of my time.
Sorry, you need a Java enabled browser to view the Dominions Player!

The actual game
From that point on, zooming in on the actual game was a matter of technique. It seemed to demand a 'growth type' system of placement, that is: adjacent to the pieces already on the board - it was very hard to visualize, otherwise, which pieces would fit in between the loose groups. Moreover, how would it end? Unlike Go, where the board eventually fills up till no more territory is available, the number of pieces in 'Dominions' was limited. The game could in fact be played on any flat surface, so I hadn't as yet considered a board, let alone its size.
Extending from the position on the board did pose a problem: why wouldn't players simply extend from their own groups and puzzle their way down to a draw? That's why I implemented the rule that a piece placed should take at least one of the opponent's liberties. This would ensure local conflict.
The first implementation behaved far too rigid however, suggesting placements with only 'blank to blank' contacts should be allowed, as long as the contact would be with at least one opponent's piece. The fact that this would create pieces that were initially disconnected from the 'beam structure' unexpectedly turned to the game's advantage. It allowed a final touch of freedom of placement: the obligation to move adjacent to an opponent's piece would exclude extensions from a piece or group still disconnected from the beam stucture. That way new groups that would start up with 'blank to blank' placement, would demand immediate attention by the very fact that they could grow independently from the opponent's groups.

Eyes & suicide
In all fairness, it took more than a couple of games to tie it all together. We worked with fairly large pieces, starting without a board. This seemed to drain some tension from the endgames and I eventually felt that a restricted playing area would introduce additional challenges. That turned out to be a good choice and the current size provides enough room for the game to unfold and yet face restraints that induce their own 'border play' tactics.

A naturally evolving concept was 'eyes', vacant positions with liberties around them that demanded a piece that was no longer available to one or both players. Implicitly, a position could be an eye for one player, but not for the other. An eye would of course mean unconditional life for a group that had a liberty bordering on it.

A dominant characteristic of the game is suicide as a weapon. There aren't too many liberties to begin with, so trimming an opponent's group down by placing suicide pieces is a rule rather than an exception, and battles over the 'last liberty' are frequent.
Dominions is an altogether unusual Go variant, again illustrating that Go as a generic concept is far bigger than Go as a 19x19 square boardgame.