This is not the case, just consider how those rules would score unclosed borders (or empty board).
This is what I wrote: in J89 the enable rule anchors hypothetical play to real go and strategies, and trying to do hypothetical play WITHOUT such anchor is what asks for trouble in the various other project-back ideas.
The point of pass stones is to areafy a territory phase, but they only work if that phase has the equal moves requirement. You cannot normally use pass stones without that requirement (would make little useful sense and could also raise pass fights).
Did you mean to drop pass stones altogether, and revert to pure area? (In that case, what you wrote still seem to give an incorrect compensation in the missed last dame example, as well as allow pass parity manipulation with plays inside territory.)
To clarify: the button or Taiwan rule only tries to correct the rounding/granularity problem, and otherwise leave area strategy untouched - which is natural and fine. What I wrote about pass stones was that IF you use them (instead of pure area), THEN you have a much simpler / straightforward alternative by just omitting them from the first phase.
Per above, Jasiek’s NAJ rules. Isn’t OGS open source? If you care enough, you could probably implement it and send a PR. They probably just have other priorities.
I’m not that inclined to debate the nuances of Jasiek’s NAJ rules. So I’m omitting that. I think we just understand them differently. Perhaps we can revisit them some other time.
The problem with the button or other parity corrections is that they are derived on the assumption that the number of passes is balanced.
So they will always break in situations like your counter example where one player can force an uneven number of passes.
You can see this by running the scoring math on the senseis page linked above.
My example rules are written without assuming pass stones and correct the parity bit for both cases of pass parity. Stated in the most simple form: White gets 1 extra point when the play parity (stones played = stones on the board + prisoners) is odd.
The reason this works is much more apparent if you use a version with equivalence scoring and pass stones because then the “rounding to an odd number” is visually apparent and the reason for why my correction is what it is becomes obvious. (Since play parity + territory = board parity. And with the equivalence scoring rule, we are adding in the white stone half the time, which consistently rounds things to an odd number.)
The pass stones are not necessary, but they make it cleaner. There are dozens of other possible variables you can use to “unround” the final result in the same way without any strategic problems that the naive correction has.
Again, all I’m doing is calculating whether or not we dropped that 1 extra bit of info and rounded the score, if so, then I give white an extra point.
Unless I messed up the math, the result should distinguish every play sequence that area scoring distinguishes in the exact same way, but it should also be able to distinguish sequences where the rounding would otherwise let one player make a 1-point mistake without penalty.
If you can find an example where this is incorrect, please let me know.
As a starting point, “pay white 1 point if play parity is odd” should not have a problem with your counter-example for button scoring since this rule correctly accounts for the pass parity change that left the play parity unchanged.
Does this clear things up?
The one thing I haven’t thought through yet is odd seki. And I expect my proposed rule to get things wrong. So if we want to also adjust for those, the issue is that the odd seki shifts the expected parity around.
Without an odd seki, we correct the score when the play parity is odd. That’s because with an odd number of scorable-intersections, plays + territory is forced to be odd. If play parity is odd, then the territory parity was even. And since area scoring rounded the score back to an odd number by virtue of that extra black stone, we pay white the point to undo the rounding / extra stone.
With an odd seki, the scorable-board parity is now even. So plays + territory will also be even.
I am not certain of the math from here on out and would appreciate a second pair of eyes, but here is a first cut at attempting to handle this case:
Since the scorable-board is even, area scoring is now rounding up to the next even score. Thus we need to distinguish between the situation where plays and territory are both odd vs the situation where plays and territory are both even.
Conceptually, if we gave black that odd seki point and then added 1 play to the count, this would put us back in the “odd scorable intersections”. Black’s score would be 1 higher, and it would be corrected back in whites favor when the new play party was odd (i.e. when this hypothetical move swapped the play parity from even to odd.)
IOW, we are saying that the presence of an odd seki means that white’s komi was too high. So we reduce white’s komi by 1 when an odd seki exists, and we still pay white the extra 1 point whenever there are an odd number of plays.
Alternatively, on an even board, we compensate black by 1 point when the number of plays is even because it was white who got the last competitive play.
This feels “off” stated this second way. So I’d like to work through some examples and have a second set of eyes double check things.
But I think that think this means that since white “normally” gets the last play, the rounding now favors them because they will be the ones able to use that last play to take the game back to even instead of 1 point in black’s favor. i.e. Instead of black turning a tied game into winning one by claiming the final dame, white can turn a lost game into a tied one by doing the same.
The reason it seems off is because an odd number of plays still rounds up an odd territory score to an even area one. And an even number of plays still implies even territory parity. But since we are effectively saying that an even board means our “even” 6.5 komi should actually be and “odd” 5.5, the score still works out correctly and all of the reasoning an heuristics are good.
I don’t know if an even board really should have an odd komi and if it really should be 1 point lower than komi on an odd board. But there is a symmetry here. On an odd board, the territory komi was even. And we are now saying that on an even board, the territory komi should be odd. (And the area komi would have to be even.)
I think this is correct, but I am unsure and would benefit from some examples to test this with and from some critical push-back on the idea.
Essentially we have a few different issues:
Is the parity correction accurate? For example, it assumes that the players cannot do anything that changes the “direction” of the rounding.
Out of all of the possible ways of calculating it, is there a “cleanest” or “easiest” one that is the best to use in practice?
How to handle people who want to use equivalence scoring and whether equivalence scoring is preferable under this type of dynamic komi arrangement.
For now, I want to focus on the first one. And since “play parity” is the easiest way to specify it, I think that’s what we should stick to using for now, even though that might not be the ideal way to implement the rule in practice.
In light of that here is the correction as it currently stands:
If there are an even number of unclaimed neutral points, white gets 1 extra point of komi when the number of board plays was odd. And, tentatively, if there are an odd number of unclaimed neutral points, black pays 1 fewer point of komi when the number of board plays was even.
(This effectively assumes that “correct” territory komi on an even board is one point lower than “correct” territory komi on an odd one. If it is actually 1 point higher, than you do not need the qualifier about the number of unclaimed points. If territory and area komi are equal on an even board, instead of 1 different because of parity differences, then the rules is that no adjustment is done if the number of unclaimed neutral points is odd.)
Does all of this odd seki stuff make sense to you?
Your ideas still seem unclear even in the simplest example: pure area with no pass stones, there is 1 dame left, and it is B’s turn.
If B plays the dame, then it’s bDame wPass bPass and B should pay compensation, and what you wrote is ok. If B overlooks the dame, then it’s bPass wDame bPass wPass and B should not pay compensation (and button/taiwan is ok). But in your terms it’s odd passes with W passing last.
If B plays the dame but tries to be tricky (as you yourself noted earlier), it goes bDame wPass bInTerritory wPass bPass, and B should still pay compensation (button/taiwan also ok). But in your terms it’s odd passes with B passing last. If W also plays into territory it becomes no-pass-go, where the result depends on SHAPES, so unpredictable and random.