A picture is worth a thousand words, so here is the proposed system used to relabel the 19x19 GnuGo ascii board


This idea is based on similar systems that also label the board relative to the corners
http://senseis.xmp.net/?AudouardCoordinates
http://senseis.xmp.net/?Rokirovka%2FCoordinates

However, the proposed system has some benefits:
Coordinates are consistently 2 (or 3 if using hyphens) characters for sizes up to 19x19
First moves in the "polite triangle" are naturally labeled (e.g., 33, 34, 44, 35)
Letter vs number distinguishes sides and corners of the board
No additional/special characters are needed disambiguate the corner

Perhaps this idea has been proposed before (since it seems fairly natural), but I haven't found this exact system on Sensei's Library.

The system is also straightforward to "easternize" by substituting the letters with 一二三四五六七八九十

The natural labeling of the "polite triangle" first moves makes the conventions of the system easier to remember

Looking at it, it seems unnatural to me.

Personally I prefer the column, row both being number/number as Japan does it, but I can live with the normal Western representation with letter/number. This just seems completely confusing.

Hmm...
What problem are you trying to fix?

_________________
- Bantari
______________________________________________
WARNING: This post might contain Opinions!!

The purpose isn't necessarily to fix a "problem", but rather offer an alternative that provides a different perspective.

Like other corner-relative systems (such as Audouard and Rokirovka), the aim is to conceptualize the coordinates as distance from the nearest corner, which is often how players think and talk about moves (e.g., terms like "3-4 5-3 enclosure" are often used regardless of which corner the moves are actually played).

Some find this type of system easier to visualize. In fact, the Audouard system was developed and used by a professional player who had severe sight loss and found this system to be more accommodating than the traditional absolute coordinate systems.

My specific proposal is perhaps just yet another corner-relative system, but I believe it may have some advantages over the other corner-relative systems (see the benefits that I listed above).


"Natural" was probably the wrong word for me to use since that is highly subjective and not quite what I intended. What I meant is that I thought that this proposal seemed like a logical way to setup a corner-relative system, and that I wouldn't be surprised if the same exact thing had been proposed before. However, I would gladly claim credit if this were indeed novel.

Whether or not a corner-relative or absolute coordinate system is more "natural" is of course highly subjective and a matter of personal preference, but some players do find corner-relative systems to be helpful (see above).

Alternative diagrams to explain the above system (Western and CJK)

Enough advantages to outweight either:
(a) re-writing all the books, software, websites, and other info already in existance, or
(b) use multiple systems concurrently, thus adding to the general levels of confusion?

Just wondering...

PS>
Sorry, don't mean to be hostile or anything here. Its just that to me, personally, a change has to have a clear purpose and the advantages of its implementation needs to outweight the problems such process will invariable generate - especially if you propose to change an established system.

Conceptually, I don't see nothing wrong with the system you propose. But I also see no real advantages. You are right, it might improve our "distance from corner" perception, but this is at the cost of numbers-and-letters-mixing which I (and others) find messy. And this evens out. So I am stuck with the question: what other advantages would this new system offer to go through all that hassle?

_________________
- Bantari
______________________________________________
WARNING: This post might contain Opinions!!

I am not at all suggesting a universal change and adoption of the proposed system.

As I said above, the proposal is merely an alternative. One that people could choose whether or not to use based on their personal preferences.

As I also said above, I believe the proposal may have some advantages over the other corner-relative systems, such using a consistent number of characters (for up to 19x19) and avoiding special characters to disambiguate the corner.

I think any comparison between the proposal (or any corner-relative system) versus absolute coordinate systems is inherently subjective, and advantages/disadvantages between the two classes (corner-relative vs. absolute) are mostly a matter of personal preference.


Again, I'm not all suggesting that we need to re-write all of the go media out there. This system is merely an option that people can choose to use if they find it convenient.

Also, for a lot of the digital media (software, websites, e-books), translating between different coordinate systems is relatively simple to implement as a user option in how the content is presented.


It seems that there already are multiple systems concurrently and widely used. There seem to be at least three (or two if you count the numerical systems with and without CJK letters as one) widely used absolute coordinate systems (i.e., "Q16", "4-16", or "四16" to describe the top-right star point), and there already are at least two alternative, corner-relative systems that have been proposed (see "Audouard" and "Rokirovka" that I mentioned above). There does not appear to be a universally accepted coordinate system.

Ultimately, I don't think offering a new alternative really adds much general confusion. In fact, if corner-relative systems help some players visualize the moves/board position better, perhaps such systems help decrease confusion.


As I mentioned above, corner-relative systems in general may be helpful for some players that find it much easier to visualize than absolute coordinate systems. For some, such as severely visual impaired players that must rely on coordinates being read to them, using such a system may be essential to playing the game. That in itself seems to be a fairly compelling reason to warrant the existence and discussion of such systems, rather than simply dismissing them.

Personally, I find absolute coordinate systems to be more of a hassle since it's much harder for me to figure out where coordinates point to (particularly those using letters larger than K) without referring to a labelled board. I don't really feel mixing letters and numbers (as already done in the western system) is messy or costly. However, I can see that other players may have a different point of view on these issues.

The Sensei's Library article for the Audouard coordinate system (one of the similar corner-relative system that inspired my proposed system) explains the motivation of Pierre Audouard, who devised that system and is a visually impaired go professional:

I see no reason to make any changes when we have something which works and has worked for a long time. Certainly any change would have to add utility. If you are wanting to make a coordinate system then why not set tengen to be the origin? That way all the hoshi points have the same absolute value of their coordinates, and 4-3 and 3-4 points are merely reflection in the coordinate system.

But as I said, I am not in favor of any such change.

_________________
Still officially AGA 5d but I play so irregularly these days that I am probably only 3d or 4d over the board (but hopefully still 5d in terms of knowledge, theory and the ability to contribute).

Again, I'm not suggesting a universal change, but merely offering an alternative idea that people can choose to adopt if they find it useful.


I believe the general concept of corner-relative systems have been useful to some players that find it easier to visualize, including those with severe visual impairments who rely on read coordinates. This general concept is not new, and the proposal is really just a minor tweak on the idea.


The aim of this and similar corner-relevant systems is to base the coordinates relative to the nearest corner, which reflects what people often do when talking about positions (and as you have done when referring to the 3-4 points. Note that you didn't say "... and 16-17, 17-16, 17-4, 16-3, 4-17, 3-16")


Don't worry, I doubt any one will force you to use it, nor are there any broad sweeping changes being suggested.

In addition to corner-relative, how about a tengen-relative system with polar coordinates.

BTW, Qwerty won .

Hmm. Tengen-based coordinates? Sounds interesting Tengen is 0,0 and first line would be all 9s, with top and right side first line in positive and bottom and left side first line in negative. Bottom left star point would be -6,-6. Bottom right star point would be 6,-6. Left side star point would be -6,0. (Hope I got it right. I was thinking of coordinates for finding locations on the Earth's surface.)

Coordinates are for recording moves in printed format. In actual play coordinates don't really matter, as the player does not think about them. Rather, relative positions of stones are more important.

There is also the perception problem. Go players consider the first line to be the bottom and tengen to be the top. Think of stones setting up base near the valley bottom in preparation for a long climb up the mountain, with the tengen at the peak. Personally I consider a goban to be flat ground and tengen is just "center point".

Yes, Go players tend to perceive distances from corners, but if I say 3-5 point, do I mean 3,5, 3,15, 17,5, or 17,15?

I'm OK with the letter + number algebraic system of coordinates. Letters are on the bottom and numbers are on the side. The XY system may be confusing for some, as one has to remember that the first number is on the horizontal axis and the second one is on the vertical axis.

competing standards
http://xkcd.com/927/

If someone finds the board easier to visualize that way, why not? However, I guess that idea didn't seem to be very well received the last time it was brought up here:
viewtopic.php?f=10&t=10721

Overall, I'm quite surprised about the nature of the reactions to this idea, which is really just a small tweak upon the existing concept of corner-relative coordinates. I didn't expect the responses to turn into an overwhelming rejection of the broader concept of corner-relative coordinates, nor did I expect such snark in response to my genuine proposal.

Perhaps the community may just be overwhelmingly against the mere mention of alternative coordinate systems, even if there isn't even agreement over calling the top-left corner "1-1" or "A19".

Is this some sort of taboo subject that I've stumbled into?

But coordinates do have a point. Its about communication. And if we're not speaking the same language, they become useless.

Its difficult enough that Western coordinates label A-T (or sometimes S) across the top from left to right, and 1-19 from bottom to top, while Eastern coordinates number from top to bottom. And when referring to a point, which is referenced first, a row, or a column? Even on the same demo board, students and teachers aren't speaking the same language, one says row-column, the other says column-row.

Of course, a real board doesn't have any coordinates, so that explains why a 5-4 point could be in any one of 8 actual points.

Heh...
You like thinking about coordinates, we like discussing it.
And I still stand by what i said: what problem are you trying to solve?
There is no taboo. Rejection? Maybe... it just shows its not a very good idea? I dunno...

We can discuss anything here.
How about making the boards pink and the stones green and yellow? I might have the advantage to attract more children because of the bright colors. Maybe it will attract more girls? They did a test like that once, with a gun lying on the street. Large percentage of boys picked it up, but only a few girls. Then they painted the same gun pink, and the results were opposite. So pink boards might have some merits.

But seriously... if you write software, you can do it however you like, with spiral coordinates for all I care. But personally, I rather use what we have, the system is really not that bad, all things considered. And it is trivial to have a function which recalucates from one coordinate system to another, so each piece of software can do it in multiple ways, its just presentation.

But as xed_over said - its all about communication and speaking the same "language" - presentation in an of itself is meaningless unless it allows us to communicate or communicate better. Which brings us the full circle to my initial point: what problem are you trying to solve?

If you just try to introduce changes to an old and established and well-working system just for the sake of making changes, its not worth it, imho.

See - no taboo at all! I could talk about it all day.
Its just that its a bad idea. Unless you have good reasons.

_________________
- Bantari
______________________________________________
WARNING: This post might contain Opinions!!

Yes, and this creates confusion.
Now - instead of trying to fix that, you propose to make it even worse.
Is this the idea? Since we already have a mess, why not make it a bigger mess, just for the heck of it?


As I said - sure, if you write a software, implement as many systems as you want, and give the user a choice. Conversions are trivial, so why not? I could easily come with a handful of different coordinate systems.

As for visualizing... I dunno. I never tried to visualize based on coordinate designations, so maybe you have something there. But I would think that the easiest system for that would be some kind of descriptive notation, like they used to have in chess becore they switched to the algebraic notation.

Generally speaking:

Overall, I would say that the ideal to strive towards would be to have a uniform system, accepted everywhere, and used by everyone. Of course, no such system can be the "best" for everybody in every situation, but I think that this is a very minor problem, and the overall advantages would by far outweight it. I think that to have once globally accepted system is a much more desirable situation than having multiple systems for people to choose from. And to be honest, I do not really care which exact system it is - migth be the one you propose.

This is the whole idea behind the concept of "standards", and it is usually accepted as a very good idea.

However:

Sometimes the accepted solution has flaws, and a better solution is found (see chass with descriptive -> algebraic notation) - and then it is worth to make the switch. But it needs to be demonstrated that the new system solves some particular flaws in the old system, and that the flaws are serious enough to make the change. Or yes, to introduce alternatives, in your case.

But please understand - this is just my opinion. And I recognize that not all may share it, and that there are people out there who would welcome multiple systems just to have alternatives. They are certainly entitled to their own opinions, and they are probably not wrong. I just don't happen to share this sentiment.

_________________
- Bantari
______________________________________________
WARNING: This post might contain Opinions!!

Whole board coordinate have not, as far as I can tell, really taken on. People still mostly talk in corner terms (3-4) and then have to rely on other cues to figure out which corner (sometimes explicit, saying "top left", or contextual). I'm not clear on the reasons for this, I would speculate that its one or more of:

- the board is just simply too big and it's not plausible (for most people) to deal with the whole board in a single system in the same way it is in chess
- so far we only have lousy whole board coordinate solutions (presumably this is what you believe)
- we have too many conflicting standards, with differing origins, directions, numbering, lettering, and most people just can't be fussed with trying to learn one minor system as no one else will understand it anyway
- it's just simply not necessary for most people, most discussions are corner specific (although I would suggest that the causal relationship might be back to front here)

If there were no alternatives already in place, I'd happily adopt it. Your solution appears to be at least as good as the Audouard system, possibly better. But since there are already many options available, it's not clear why I'd bother learning this one.

As pointed out already, the whole point of a coordinate system is to communicate something, so it's really important that as many people as possible speak the same language. So you (or anyone) who cares about this has two options - lobbying for the increased adoption of an existing (capable) standard, or introducing a new one, and lobbying for that one to be widely adopted.

Communication occurs in many different forms, between humans directly (such as talking over a physical board), between a human and a machine (such as a person using go software to study games), between humans but with the interaction mediated by machines (such as people playing go on a server or discussing go in a forum), and even between machines (such as files and data being exchanged between programs, perhaps as part of mediating human interactions).

Having a consistent language is an important part of communication, but actually using a language that is appropriate to the context and facilitates clear/easy communication is perhaps even more important, and in different contexts different languages seem to be preferred. For example, over the board, most people already speak in corner-relative terms (albeit, usually ambiguously unless clarified in context), machines often use SGF which employs an absolute alphabetic system, and players on go servers often use "A1" since most servers/clients tend to label the board that way. Consistency need only be maintained in the context of each separate communication rather than universally. I think forcing a universal system across all of these contexts would only cause more difficulty.


I believe the primary goal of standardization is to clarify and consistently specify concepts/systems, rather than impose universality. It is rarely the outcome (or even the aim) of a single standardized system to achieve universal adoption. Often, this is simply due to the problem having different contexts that require different solutions, since a good solution is one context may be poorly-suited to another. See the examples discussed earlier.


The problem is that the commonly used absolute systems are difficult to use or seem unnatural/counter-intuitive to some players. Corner-relative systems provide a conceptually different alternative that might help address some of these issues. Also, I believe the existing corner-relative systems could perhaps be improved, and hence I proposed some small tweaks upon that idea.

I already mentioned the case of Pierre Audouard and his motivation to create a similar corner-relative system due to him finding traditional coordinate systems difficult to work with given his vision impairment. Perhaps you overlooked the above remarks, or dismissed the example, but I think it illustrates how a novel coordinate system can greatly aid the communication of board positions, where other systems were not suitable.

Personally, I also struggle with using the absolute coordinate system, since I find it difficult to remember the mappings of letters larger than K. Even the double digit numbers feel awkward as you are essentially counting from the opposite side. Too often, it feels like a distracting waste of time to manually convert in my head or refer to the board labels (if available), since numbers like 15, 16, 17 just don't immediately and naturally map to the 5th, 4th, and 3rd lines. Even if one does get used to mapping these instinctively, one has to start all over (and might even get tripped up) when dealing with other board sizes. The choices for convention (i.e., origin placement and row-column vs column-row) in traditional systems also seem somewhat arbitrary leading to typical first moves being labeled by not easily deciphered coordinates such as "Q16" for hoshi or "4-17" for komoku.

The proposed corner-relative system addresses some of these issues by reducing the letters used to A though K, which are mapped to 1 through 10, and setting convention based on the principle of labeling typical ("polite triangle") first moves using their typical corner-relative terms, e.g., "4-4" for hoshi and "3-4" for komoku. The entire top-right quadrant is simply just labeled relative to the corner using two numbers. In other quadrants, the coordinates are also just the distance from the corner, but one or two letters are instead used to disambiguate which quadrant you are in.
The first coordinate gives the column, which counts from the left if it is a letter, or from the right if it is a number. The second coordinate gives the row, which counts from the bottom if it is a letter, or from the top if it is a number. The CJK symbols offer another alternative if one finds that more familiar or easier to use than the roman letters.


The key aim of the proposal and similar corner-relative systems is to provide a way to unambiguously refer to any point on the board while still retaining familiarity of corner terms. In the proposed system, this is disambiguated by whether letters or numbers are used to specify the point


I've made this similar point as well. In a lot of situations, software simply makes the application of different coordinate systems a simple matter of presentation. Additionally, software also makes it a simple matter of translation between two people that prefer different systems. For example, imagine as go server where whenever you reference a point using your preferred system that is being displayed on your client, the server automatically translates the coordinates into the preferred systems of other users as chosen and shown by their client.


Since you seem to strongly think that it is "bad idea", I would be very interested to hear some more specific criticism about the actual proposal. Besides "the cost of numbers-and-letters-mixing which I (and others) find messy", I don't seem to see any other remarks regarding the specific system. It seems that your arguments have mainly focused on a general objection to any alternative coordinate system proposals based on the idea of universality.

I agree that the conventions for Western "A1" coordinates and numerical coordinates "1-1" seem a bit arbitrary. In the system that I propose, the convention of placing 1-1 in the top-right and using column-row is determined by the principle of aligning typical first moves (i.e., in the "polite triangle") with the typical corner terms used to refer to them. For example, hoshi (Q16, 16-4) is simply 4-4 in my system. Komoku (R16, 4-17) is simply 3-4.
Thus, remembering the convention of my proposal is aided by this mnemonic.


My proposed system writes all of the points in corner relative terms, but letters sometimes substitute for numbers to indicate which side/corner the point is in. For example, the eight 5-4 points are given by:
Top-right: 4-5, 5-4
Top-left: D-5, E-4
Bottom-left: D-E, E-D
Bottom-right: 4-E, 5-D
By mentally mapping A through K (excluding I) to 1 through 10, all of the above coordinates express 5-4, with the use of letters disambiguating which corner.