Stone removal and scoring updates

[anoek]

Hello OGS!

Following up on our Auto-score improvements from a few weeks ago, and the mild mutiny objecting to the auto-sealing feature therin, we’re following up with a number of updates and improvements to scoring.

Scoring & Stone removal phase changes

First off, auto-sealing has been removed. Replacing it is a system which attempts to notify players of points that still need to be sealed before the game can be scored as the players likely intend. We have various checks and thresholds to try and not warn about dame, but I’m sure we’ll need to make some tweaks to the system based on some wider real world exposure, please let me know if you see weirdness with the new system.

---

Secondly, we’ve incorporated David Wu’s scoring algorithms for figuring out what’s dame and not. With this we’ve removed manual marking of dame, which means no more blue squares during the scoring phase. The expectation is that David’s algorithms are smart enough to score finished games correctly, including detecting implicit dame like T8 in the image below. (When area scoring is used, T8 is considered a point for black of course.)

---

Third, a minor visual change for removed stones has been made, now stones marked for removal during the stone removal phase will have a red X going through them as well as semi-transparent

---

Fourth, during the stone removal phase, if a stone that is part of a stone string that is definitely alive is clicked, such as O4 in the image below, the group will not be marked for removal and instead a message will be displayed informing the player that the system thinks the group is definitely alive so is soft-refusing to remove it.

As the message says, if for some reason the stone string really should be removed, this protection can be overridden by shift-clicking or long pressing.

---

Finally, the last change in the stone removal phase is the way stone strings are marked or unmarked for removal has changed to remove the string of stones clicked as opposed to recursively marking all stones. The reason for this change is to primarily make changes less surprising for beginners, and it’s my hope that the automatic stone removal improvements will largely make making large changes largely unnecessary, but let me know how this feels in practice. We want to strike the balance between being quick and convenient to fix any scoring mistakes that might still be present, and protecting beginners from inadvertently marking the entire board as dead and getting the board state into a mess.

---

Score marking in Analysis mode and Reviews

Marking scores and removed stones is now possible in both analysis mode and in reviews. In addition to white and black scoring positions, you can also mark the board with custom colored squares. The calculator button will automatically mark all scoring positions, and the erase tool by the calculator will clear the score markings from the board.

---

Thanks in advance to folks who test the new systems out on the beta server. There were a lot of moving parts to change out, so please let me know what issues you find, and also any feedback you have on the new systems.

– anoek

[yebellz]

Just a small UX design suggestion

[Eugene]

Is this some kind of joke?

How can it be that a post from yebellz refers to “scoring a game correctly”?!

(A compendium of OGS's terrible scoring system confusing beginners - #262 by qnpnpmqppnp)

[yebellz]

Right, to properly implement the Japanese rules scoring, we would of course have to handle some potential situations with Clippy informing everyone that both players have lost.

[benjito]

If I recall correctly, it also depends whether you’re using the '89 ruleset or the XP ruleset.

[Feijoa]

This looks like a dramatic improvement! Follows the rules of go:

No surprise invasions or random dame in the middle of territory:

Even handles weird cases:

[GreenAsJade]

Woo hoo “dramatic improvement” … that’s what we like to hear

[anoek]

Thanks for the testing @Feijoa!

[BHydden]

F2 hurts

[qnpnpmqppnp]

Yeah I’m still not a fan of this “scoring correctly” wording. Scoring it “correctly” would be to just apply the scoring rules of Go to the situation as is.

I much prefer “scoring as the players likely intend”, as used in Anoek’s announcement.

Nitpicking aside, all of this sounds like nice improvements. Looking forward to see it implemented!

[gennan]

@Feijoa

Indeed this seems like a great improvement, but those red squares might be more confusing than helpful in this position.
If I understood correctly, those red squares indicate an open border. But if I were to indicate open borders in this position I’d say that black has open borders at (roughly) J9, J7 and J5, while white may have and open border at G10. So something like this:

The red marks in the actual image would be the most favourable border expansion that black might achieve if white keeps passing indefinitely, so it’s not very impartial.

[qnpnpmqppnp]

I don’t think the intent is to be impartial, rather pedagogical.

Your second example is closer to a realistic outcome, but it relies on a lot of assumptions/guessing and in my view seems altogether harder to understand (what’s special about J5, why not H5?). I prefer the first one.

[gennan]

@qnpnpmqppnp

what’s special about J5, why not H5?

Indeed there is ambiguity when trying to indicate open borders. It is inherently fuzzy.
Perhaps this fuzzyness might then be indicated something like this (with a “thicker” border)?

[teapoweredrobot]

@gennan

this fuzzyness might then be indicated something like this (with a “thicker” border)

I think this would be best, if possible!

[gennan]

I do feel that it is a bit cluttered with so many red squares. Either way, I feel those red squares are not very intuitive and need some additional text explanation about what it means, which suggests it’s not a great UI design.

Another suggestion is to draw lines connecting points where the influence of the nearest black and white stones balances out, something like this:

Or perhaps connect boundary stones where the “influence” of a player drops below some value, something like this:

Or perhaps the complement of that, showing arrows through the bottlenecks where the opponent still has a way through those sector lines, something like this:

[GreenAsJade]

@gennan

I do feel that it is a bit cluttered with so many red squares.

I dunno, I think the sea-of-red-squares suggestion is not so much “cluttered” as “suitably indicating the problem area”

So far that’s the easiest to explain and simplest suggestion so far. It says “you have a problem around here”.

[gennan]

@GreenAsJade

It says “you have a problem around here”.

Does it though?

@gennan

image655×657 71.4 KB

I’d counter that many of those red squares are not really a problem, like H6, H8 and G4 are basically neutral points that have little to do with unclosed borders.

Another alternative might be to indicate the full extent of the problem, floodfilling the whole contested area with red squares, like this:

A benefit of this method is that it highlights the problem very clearly without offering assistence on how to fix it.

[BHydden]

All of these screenshots are neglecting to include our helpful friend Clippy who is actively explaining the marked squares indicate a border that needs sealing.

[gennan]

@BHydden

our helpful friend Clippy who is actively explaining the marked squares indicate a border that needs sealing.

If that is the explanation text for the red squares, then I think an effort should be made to have those red squares actually marking border closing intersections (like in my first image, or the one below[*]), and not the boundary where floodfilling fails.
If the explanation doesn’t actually match the implementation and resulting marking, then I still expect many questions to be asked.

---

[*]

[ArsenLapin1]

@gennan

Another alternative might be to indicate the full extent of the problem, floodfilling the whole contested area with red squares, like this:

I like this floodfill suggestion the best. It really highlights the problem “if you think black had territory here, you need to do something about it”.

I see two issues with only marking some “border” intersections as red:

• it’s pointing out the “solution” before the problem, telling the player where to play to fix their territory even though they didn’t even know they had a problem in that territory
• it’s too authoritative, seemingly telling the player where to play even though the exact location of the red squares is pretty arbitrary