Coordinate System Proposal (corner-relative)

[quantumf]

Well, if you're discussing a purely local issue on the side instead of in a corner, the distance from the corners doesn't matter much, so coordinates are irrelevant locally, only the line (2nd, 3rd, 4th, etc.). Similarly, for a central fight, if you're talking only about the local situation, coordinates don't matter at all, because they could be shifted a few lines to one side or the other without changing local aspects of the fight.

Considering these globally is a different story, of course, but I think in most cases it's easy enough to reference other parts of the board generally, like "this isn't playable because white's ladder breaker in the corner is too good" or "this will live locally, but you'll undermine the corner, leading to a net loss." If you're trying to analyze a whole side or board verbally, I think you'll lose your audience long before the position is fully described unless you have a board or diagram to reference.

I kind of agree but do wonder if our tendency to discuss matters in a largely local fashion is at least to some extent a consequence of an inadequate system for discussing the board in a (shared) global way

I suppose if there were one coordinate system to rule them all, so to speak, we'd already be using it because its superiority would be unquestionable. As it is, the sunk cost of having implemented the current system and most westerners being used to it leads to a transition cost that is greater than the added utility of a new system.

Don't underestimate our ability to be entirely wrong about something. The absence of a good, shared system doesn't necessarily mean that one does not and can not exist. Up until five years ago, the Internet was largely the domain of techies, and there was no effective way for (remote) families and friends to share their day to day lives effectively with each other

[YeGO]

I don't see any niche for it either ...
... Now this isn't always an option, of course. Not all online go places allow you to add labels or variations in all circumstances.

Not all online situations allow for diagrams either. Your examples just demonstrate situations where utilizing the visual aspect of a diagram is better than using any coordinate system. I fully agree that using a diagram is a much nicer alternative than relying on coordinates to describe a board position, sequence, etc. I am not all suggesting that we replace diagrams with text-only descriptions.

However, sometimes using a diagram isn't feasible or convenient, as it requires the ability to create, manipulate and/or view a picture.

For example (as mentioned earlier), one particular niche for corner-relative systems is that they may be much easier for the visually-impaired to work with than traditional, absolute coordinate systems.

Also, coordinate systems could be useful in casual, text-based communication mediums where sharing a diagram or even a link to a diagram may not be possible or inconvenient.

Code: Select all

Alice: Hey Bob, did you see catch that pro game today?
Bob: Yeah, that was quite an interesting opening.
Alice: Oh really, what happened? Got a link to the kifu?
Bob: Don't know where/if it's been posted, but here are the first few moves ...
3-3 D-D
3-C D-4
K-K 4-K
C-K F-K
K-3 K-6

What was that famous old game where black opened on 33, DD, KK? The player is just on the tip of my tongue, but I'm drawing a blank.

Again, coordinate systems certainly aren't meant to fit every niche, but they can be useful in some.

[Bantari]

When discussing a position or sequence without a board (or other visaul aid) handy, I see 3 possible scenarios:

1.We discuss a global, full-board situation.
I honestly believe that in 99% of the case, this is too complex for most of us to really visualize just based on coordinates. This is why blind 19x19 go is so hard, even for the pros. But even if we do manage to visualize a full 19x19 board, it would be much easier to follow any given sequence by making the coordinate system have a nuifor origin (i.e. the center or one corner as it is now.) Having each corner sort-of described by its separate local coordinates, even if then finally blend to some global system, is more confusing than it is helpful. Thus - the proposed system fails.

2. We discuss a local (corner?) situation.
In this case, the coordinate system we already have, with single origin in single corner, is sufficient. One could say that for small positions it is not distinguishable from the new system - all situations are local and the stones/moves are described with respect to the A1 corner. So the proposed system is not really needed.

What is important to me, however, is that the current system is more continuous. Imagine we discuss a corner situation which spills accross the tengen, some complex joseki, or maybe just a long running fight or a ladder. Using the current system, we just continue expanding our values - we talk about J8, K9, and then L11, and M12. In the proposed system we talk about JJ, KK, and then 99 and 88 - we increase the alpha values at first, and then as the position grows we have to remember to start decresing the numbers instead - but the flow of the play is the same. I find it highly counter-intuitive. So to me, the corner-relaive system fails in this situation as well.

3. We discuss local (corner?) position with some global ramifications (letter breaker, etc.)
In this case, like in case #1, it is my assertion that a system which one specific origin in one specific corner makes it much easier to visualize than the proposed system. The reason is that, like with for example - the ladder - it is easier to visualize along a continuous straight progression of coordinates (f7, g8, h9, j10, l11...) than to count up and then count down again in a confusing manner (f7, g8, h9, jk, kj, ...) At least - I would find it much easier. So here also the corner-relative system fails.

I actually had to work hard to figure out the above jk,kj sequene, and still not sure it is right.

Or another example: It is just much easier to visualize (and calculate) that a stone on M7 can break a ladder on H3 (current system) than a stone on 8G can break a ladder on H3 (corner-relative system.)

It shows to me, that in general, the corner-relative system - while might be equivalent to the present system in local small-scale corner situations - completely fails when sequences spill across the 4 quadrants and the coordinate counting needs to be adjusted. Especially when the tengen is not involved... Very very confusing.

In addition, I find it much easier to visualize distance between points when they are expressed in the present system. For example, i instinctively know that a4 and q16 are in opposite corners, or at least - not very close. It is much harder to immediately make this connection between DD and 44 (or between D4 and 4D.)

Granted - some of the above might be just training and habbit issues, but like it or not, this is what we are mostly used to - since most of the world around us is structured like that. Chess coordinates, isles in supermarket, blueprints of a house, etc. of course, there are other coordinate systems used as well, but these are usually for very specific situations and much harder to visually adjust to in general.

PS>
Yup, slow week.

[YeGO]

1.We discuss a global, full-board situation.
I honestly believe that in 99% of the case, this is too complex for most of us to really visualize just based on coordinates. This is why blind 19x19 go is so hard, even for the pros.

No doubt that it is very difficult to play blind 19x19 go, but this is just one situation where any coordinate system might be used to describe the global situation. The difficulty of substituting a visual interface with coordinates is not really an argument against corner-relative systems, but rather all coordinate systems as a whole.

But even if we do manage to visualize a full 19x19 board, it would be much easier to follow any given sequence by making the coordinate system have a nuifor origin (i.e. the center or one corner as it is now.) Having each corner sort-of described by its separate local coordinates, even if then finally blend to some global system, is more confusing than it is helpful. Thus - the proposed system fails.

I think this broad rejection of corner-relative systems reflects a failure to grasp that others might have a vastly different perspective than your own. You might personally find it much easier to use "A1" coordinates, but that certainly does not mean that everyone in the world shares that preference.

As I mentioned earlier, Audouard's corner-relative system was specifically designed by a visually impaired go player who found that a corner relative system was much easier to work with than absolute systems like "A1".
http://senseis.xmp.net/?AudouardCoordinates

Another user has even converted the game records of Shusaku into Audouard's corner-relative system and shared a script that dictates these records:
https://gist.github.com/Iktomist/1417181
https://gist.github.com/Iktomist/1439360

I also personally find it much easier to work with a corner-relative system. This opinion is due to my personal experience of finding it confusing and unnatural to refer to three-quarters of the board relative some other arbitrary corner. Personally, I find it much easier to convert between a specific point on the board (in a physical, virtual, or mental visualization) and the corresponding corner-relative coordinates.

Perhaps this preference is of a small minority, but it's not reasonable to claim that one type of system is objectively better or easier than another.

Alternatives exist to serve different needs and preferences, which is all that I am proposing. Your insistent hostility to this alternative seems to suggest that you are under the impression that I am advocating for universal adoption (and forcing others to change systems against their wishes), which certainly is not my goal. It would be completely unreasonable for me to insist that you use a corner-relative system (since you have a clear preference against it), just as it would be unreasonable for you to insist that I or others must conform to using an absolute system.

And people do have different perspectives which results in different needs and preferences. In the remaining response, I will try to further explain my personal perspective to illustrate.

2. We discuss a local (corner?) situation.
In this case, the coordinate system we already have, with single origin in single corner, is sufficient. One could say that for small positions it is not distinguishable from the new system - all situations are local and the stones/moves are described with respect to the A1 corner. So the proposed system is not really needed.

In the case of a position or sequence fully confined to one corner, it seems that the most common practice is already just to use plain numerical coordinates relative to that corner, since a whole-board absolute system like A1 is not needed either. For example, discussions like "following white's 3-4 approach to black's 5-3 stone, black's play at 4-6 initiates the Taisha joseki" using corner-relative coordinates (regardless of the actual corner) seems to be much more common than using absolute "A1" coordinates instead.

However, a difference between the proposed system and A1 system is apparent when it is desired to have such a discussion in a context where disambiguating the corner/orientation may be desired for clarity (e.g., discussing a specific sequence within a game). For example, if the sequence happened to have played out in the upper-right, the discussion remains unchanged, e.g., "At move 11, following white's 3-4 approach to black's 5-3 stone, black's play at 4-6 initiates the Taisha joseki", whereas with the A1 system the discussion would become "At move 11, following white's R16 approach to black's P17 stone, black's play at Q14 initiates the Taisha joseki". If the sequence happened in a different corner/orientation, the corner-relative discussion only changes slightly, with letters being substituted for numbers in order to distinguish which corner/orientation, e.g., "At move 11, following white's C-4 approach to black's E-3 stone, black's play at D-6 initiates the Taisha joseki". Compare to the same discussion based using "A1" system which would substitute "C16", "E17", and "D14" instead.

The proposed system allows one to continue using just a slight modification to the familiar corner-relative terminology, while also being unambiguous about the specific corner under discussion.

What is important to me, however, is that the current system is more continuous. Imagine we discuss a corner situation which spills accross the tengen, some complex joseki, or maybe just a long running fight or a ladder. Using the current system, we just continue expanding our values - we talk about J8, K9, and then L11, and M12. In the proposed system we talk about JJ, KK, and then 99 and 88 - we increase the alpha values at first, and then as the position grows we have to remember to start decresing the numbers instead - but the flow of the play is the same. I find it highly counter-intuitive. So to me, the corner-relaive system fails in this situation as well.

3. We discuss local (corner?) position with some global ramifications (letter breaker, etc.)
In this case, like in case #1, it is my assertion that a system which one specific origin in one specific corner makes it much easier to visualize than the proposed system. The reason is that, like with for example - the ladder - it is easier to visualize along a continuous straight progression of coordinates (f7, g8, h9, j10, l11...) than to count up and then count down again in a confusing manner (f7, g8, h9, jk, kj, ...) At least - I would find it much easier. So here also the corner-relative system fails.

I actually had to work hard to figure out the above jk,kj sequene, and still not sure it is right.

Or another example: It is just much easier to visualize (and calculate) that a stone on M7 can break a ladder on H3 (current system) than a stone on 8G can break a ladder on H3 (corner-relative system.)

It shows to me, that in general, the corner-relative system - while might be equivalent to the present system in local small-scale corner situations - completely fails when sequences spill across the 4 quadrants and the coordinate counting needs to be adjusted. Especially when the tengen is not involved... Very very confusing.

I think that the potential difficulties of working with coordinates across quadrants is definitely a valid point. As you mentioned, counting across the sequence "... H J K 9 8 7 ..." seems to be a counter-intuitive drawback. Along a similar vein, some may struggle with intuitively seeing the table shape of stones at J9 99 9J KJ. However, I think these drawbacks reflect the general difficulty of completely forgoing a mental, physical, or virtual visualization in an attempt to work with the coordinates directly, which does not seem to be the right way to utilize any coordinate system.

For example, in a game played on a physical or virtual board, I would not attempt to analyze a ladder by first converting the positions of the stones involved into any coordinate system and then just work with the coordinates to understand the position. Instead, I would mentally visualize the continuation of the ladder and how it may be influenced by potential ladder breakers and other complications. If a virtual/physical diagram was not available, I would resort to a mental visualization, where I would first place the stones in the appropriate positions in the mental picture and then visualize the continuation extending from that position. In general, ladder analysis is more complex than simply checking just one diagonal, so coordinate arithmetic would be a poor substitute for actually mentally visualizing the whole board picture and potential continuations/disruptions.

Thus, personally, I still find the corner-relative system easier to work with in situations like this, since I find it easier to convert them into the appropriate locations for a mental visualization, whereas in working with "A1" coordinates I would struggle with intuitively locating coordinates such as "M7", "R17", "O14", "G15", etc., which occur further away from the bottom-left origin. A benefit of the corner-relative system for me is that it avoids these larger coordinate components, which are less familiar to me since so much existing go discussion uses plain numerical corner-relative coordinates confined to a 9x9 corner of the overall board.

Revisiting the table shape mentioned earlier, I still do find it easy enough to visualize "J9 99 9J KJ" as a table shape surrounding tengen, however, I can see how this may be stumbling point for others not familiar with this system. The key, I think, is not to just treat and decipher the coordinates as a jumble of characters, but to plot the mental picture that each point expresses. On the other hand, if I had to use absolute coordinates in dealing with a table shape, such as one in the upper-right at "R15 R17 P17 P16", looking at just the relative relationship between the characters would reveal that it is a table shape, but the specific absolute location of it becomes harder for me to pinpoint due to the unfamiliarity of mentally plotting points relative to the opposite A1 corner.

In addition, I find it much easier to visualize distance between points when they are expressed in the present system. For example, i instinctively know that a4 and q16 are in opposite corners, or at least - not very close. It is much harder to immediately make this connection between DD and 44 (or between D4 and 4D.)

The letters versus numbers distinction helps me to determine which corner the point is in. A letter indicates counting from the left or bottom, and a number indicates counting from the right or top. Hence, when I see pair of points like "DD and 44", since one is letter-letter and the other is number-number, I can immediately make the connection that they are at opposite corners. Similarly, the D4 and 4D pair is readily apparent to me as points in opposite corners since for each component, the letter-number distinction is flipped.

Further, the corner-relative nature of the coordinates helps me to quickly see that they are all just 4-4 hoshi points, whereas with absolute coordinates, my ability to intuitively see Q16 as a hoshi point would be limited by my ability (or lack thereof) to intuitively associate "Q" and "16" with the 4th line. I struggle with recalling the correct associations (for larger letters/numbers), and find it cumbersome to have to remember completely different associations for other board sizes (e.g., K10 becomes a 4-4 point on the 13x13 board). On the other hand, the 4-4 points are "4-4 D-4 4-D D-D" in corner-relative coordinates across all boards sizes from 9x9 and up, making it easier for me to carry this familiarity across different board sizes.

Perhaps other do not share these perspectives, but that is what makes the corner-relative system easier for me to work with.

Granted - some of the above might be just training and habbit issues, but like it or not, this is what we are mostly used to - since most of the world around us is structured like that. Chess coordinates, isles in supermarket, blueprints of a house, etc. of course, there are other coordinate systems used as well, but these are usually for very specific situations and much harder to visually adjust to in general.

It is not my aim in this response (or overall thread) to convince you to use this, or any coordinate system for that matter. That decision is yours and depends on what works best for you. It is also not my aim to convince you that this corner-relative system would be easier or better for you. I fully agree that corner-relative systems are not for everyone. Perhaps even the vast majority strongly prefer absolute coordinate systems, finding corner-relative systems to be confusing and useless. However, it is not reasonable to dismiss the system as universally and objectively worse, confusing or useless, as some players have personally found corner-relative systems to be useful and easier to work with.

Ultimately, my aim here is just to illustrate how different people may bring different perspectives to this issue by explaining my own.

[YeGO]

I must have I overlooked this comment earlier, but it sounds very interesting...

There is another historical coodinate system which I believe hasn't been mentioned so far. The four corners are labeled North, South, East, West. Then the coordinates are relative to the respective corner. Tengen is separate.

This system is used in the book "The game of Wei-Chi" from 1929, which is based on older Chinese texts. I'd love to know when this coordinate system was in use and what happened to it. (Also in the book, tengen is marked with a yin-yang symbol. I'd love to know what happened to that as well.)

I am curious about this old Chinese coordinate system as well. I have not seen that book and do not know where to find it.

However, I did find the below comment on Sensei's Library that seems to allude to something similar. John Fairbairn mentioned that Audouard's corner-relative system closely resembles an ancient Chinese system that is mentioned in old texts.

John F. Although Pierre Audoard sounds like a truly remarkable person, and he may well have devised this system independently, I think it needs to be recorded that it appears to be identical to the ancient Chinese system first seen in Carefree & Innocent Pastime. But that raises an interesting question about motivation, along the lines of great minds think alike, perhaps. It also means that lots of problems and game commentaries using this system already exist in the old Chinese literature. Personally I found it a bit of a curate's egg, but it strikes me as usable and I wouldn't take too much convincing that it's a better system than the usual x,y coordinates.

Unfortunately, I have no idea where to find/view a copy of this "Carefree & Innocent Pastime" book, but I would be interested in finding out more. Apparently, as mentioned on its SL page http://senseis.xmp.net/?WangyouQingleJi, this is a very old book (from 1100 AD!), so it seems that corner-relative coordinate systems have certainly existed for quite a long time!

[YeGO]

There is an interesting discussion on the OGS forums about various alternative coordinate systems, motivated by the scenario of recording moves on pen and paper (where a kifu sheet is not available).

OGS Forums Thread: Tournament Shorthand