willemien, what is "natural" for the purpose of go rules? For the sake of simplicity, let us forgo the grid, the choice of playing stones on the vertices (instead of the facets or lines), the number of players, Black first etc. and let us assume the game shall be a mental competition etc.

Alternation:

This allows for a balanced interaction between the two players and creates the greatest density of turn-driven interaction. Balance is natural and greatest density is natural for the sake of mental competition, which we have presumed. That the game is turn-driven and not a game of simultaneous actions is not natural though but yet another rather arbitrary choice of game design. Also simultaneous actions would mean balanced interaction and greatest density (of a different kind than turn-driven interaction).

Play of one stone:

The units of expressing plays are the stones. At a turn, a player could play none, one or several stones. Any number of played stones at a time greater than one would be somewhat arbitrary but also highly dynamic. Being arbitrary is not that natural while being highly dynamic is natural for the sake of mental competition. Hence it is far from obvious whether one or several is more natural. Playing none is unnatural though because then progress by means of stones placement is halted. So at least we can say that playing one or several stones is more natural than playing none. - However, a game to be a mental competition should normally come to an end to allow for a comparison between the two players' skills under controlled conditions. Oh, wait - should it? It would also be possible to have a live scoring system that adds to the score at, say, every turn. Then one could have an eternal game with ongoing evaluation of the players' mental competition successes. Therefore also playing no stones during a turn could be natural. - Thus, to say which number of played stones per turn is natural, it requires a context of game procedure, game end and scoring.

We learn that "natural" is not necessarily a characteristic of each single game aspect considered alone for itself. Rather "natural" is or is not a property of a game's entire rules design as a whole. This, however, requires a fresh approach to possibly assessing whether a ruleset is natural. What makes a system (like a set of rules) as a whole "natural"? Supposedly if its components fit well together in some optimal, clear, easy to understand manner.

Now go ahead and evaluate which ruleset is or is not natural as a whole...!

the problem with go is that to much is natural.

It is as natural to count territory as it is to count area. And therefore there is a lot of discussion between the merits of one above the merits of the other.
(one of the big advantages of territory counting is you don't need to count as far)

all kinds of repeated position have something "looks natural" over them. and therefore are also open to discussion. (non is obviously better than an other, they all have some merit)

It goes a bit back to an old adagio.

Chess was invented, Go was discovered.
(but we still need to learn to recognise the best rules)

_________________
Promotor and Librarian of Sensei's Library

Go is a game on the board; therefore keeping prisoners next to the board is already unnatural; hence territory scoring is unnatural. Another unnatural aspect is the introduction of different kinds of play: real (the game's move sequence) and hypothetical (the analysis game trees).

What does it matter for the rules whether repeated positions look natural?

If you don't want to count far, then use either 1) prisoner counting or 2) putting away pairs of stones counting or 3) arranging all stones on the board symmetrically counting. You don't even need to count; just determining the remaining colour determines the winner; this makes each of these methods easier from the view of remaining counting effort than Ing fill-in counting and - after the rearrangement - about 70 times as short as the numerical counting step of Japanese rearrangement counting. If you want to include the rearrangement, then Point-by-point-half-counting is as fast as Japanese rearrangement counting.