Life In 19x19
http://lifein19x19.com/

Centre-based Coordinate System?
http://lifein19x19.com/viewtopic.php?f=10&t=10721		 Page 1 of 2
Author:  Loons [ Mon Aug 18, 2014 2:35 am ]
Post subject:  Centre-based Coordinate System?
I think basing coordinates on tengen = 0,0 is a good and natural idea. With the first quadrant as black's top right (the normal first move).

At the moment, when we want to talk about a 4-4 stone-

On the board it might be : Q16, D16, D4, Q4 (unless Q is P instead).

Why not, intuitively : 6,6 ; -6,6 ; -6,-6 ; 6,-6 ?

(Travelling round counterclockwise).

I can imagine the point that edges are important, but I would argue that remembering where Q, R, 17 and F are is more confusing.
Author:  SoDesuNe [ Mon Aug 18, 2014 2:47 am ]
Post subject:  Re: Centre-based Coordinate System?
I don't find that intuitively at all ^^

Every board game I play (chess, Shogi, Go) uses numbers and letters as coordinates. I feel quite at home ; )
Author:  RBerenguel [ Mon Aug 18, 2014 2:49 am ]
Post subject:  Re: Centre-based Coordinate System?
http://senseis.xmp.net/?AudouardCoordinates
Author:  Loons [ Mon Aug 18, 2014 2:57 am ]
Post subject:  Re: Centre-based Coordinate System?
@Sodesune
Go (at least starts with) more symmetry than eg. Chess.

@RBerenguel
I had a cursory glance at that page before. It seems inelegant to me; prefixing a move with the quadrant, where quadrant 2 is named "a" and named clockwise. Then there are four corner origins.

Compared to just using cartesian coordinates around a centre origin (of heaven).
Author:  RBerenguel [ Mon Aug 18, 2014 3:30 am ]
Post subject:  Re: Centre-based Coordinate System?
Loons wrote:
@Sodesune
Go (at least starts with) more symmetry than eg. Chess.

@RBerenguel
I had a cursory glance at that page before. It seems inelegant to me; prefixing a move with the quadrant, where quadrant 2 is named "a" and named clockwise. Then there are four corner origins.

Compared to just using cartesian coordinates around a centre origin (of heaven).


I was on iPad so didn't feel like writing that much :) For me, it would be extremely more easy having coordinates centered at 0, too, being used to cartesian "mentally." I'd play much better blind through GTP, so far even the most stupid engine at lowest level beats me. It would also make coordinates "uniform" between different sized boards. So, I'd love if SGF worked like this instead, but... :(
Author:  hyperpape [ Mon Aug 18, 2014 4:21 am ]
Post subject:  Re: Centre-based Coordinate System?
"Negative six seven"
"D17"

"-6, 7"
"D17"

It might be a nice mental model, but it's not so hot for verbal or written communication.
Author:  DrStraw [ Mon Aug 18, 2014 4:43 am ]
Post subject:  Re: Centre-based Coordinate System?
RBerenguel wrote:
It would also make coordinates "uniform" between different sized boards. So, I'd love if SGF worked like this instead, but... :(


I don't think so. The 3-3 point would have different coordinates on each board size.
Author:  quantumf [ Mon Aug 18, 2014 5:10 am ]
Post subject:  Re: Centre-based Coordinate System?
DrStraw wrote:
RBerenguel wrote:
It would also make coordinates "uniform" between different sized boards. So, I'd love if SGF worked like this instead, but... :(


I don't think so. The 3-3 point would have different coordinates on each board size.


Not much worse than the current system - only one point (C3) is consistent across the sizes.
Author:  RBerenguel [ Mon Aug 18, 2014 5:19 am ]
Post subject:  Re: Centre-based Coordinate System?
DrStraw wrote:
RBerenguel wrote:
It would also make coordinates "uniform" between different sized boards. So, I'd love if SGF worked like this instead, but... :(


I don't think so. The 3-3 point would have different coordinates on each board size.


3-3 is not even well defined. Which 3-3?
Author:  jug [ Mon Aug 18, 2014 9:18 am ]
Post subject:  Re: Centre-based Coordinate System?
Quote:
It would also make coordinates "uniform" between different sized boards.

On DGS it has been asked to establish Audouard coordinates, but DGS also has even-sized boards, which does not have a center point.

I don't think it's better or worse ... just different.
Author:  RBerenguel [ Mon Aug 18, 2014 9:29 am ]
Post subject:  Re: Centre-based Coordinate System?
jug wrote:
Quote:
It would also make coordinates "uniform" between different sized boards.

On DGS it has been asked to establish Audouard coordinates, but DGS also has even-sized boards, which does not have a center point.

I don't think it's better or worse ... just different.


Agree. I'd rather have cartesian having a centerpoint in the board, when it's not available... well, inexistent point would work, too. I'd probably be happy enough if SGF used human coordinates instead of just-letter coordinates (ee instead of e5, for instance.)
Author:  Elom [ Mon Aug 18, 2014 9:45 am ]
Post subject:  Re: Centre-based Coordinate System?
I had the exact same idea a while ago, and I thought that it's main use would be for computer programming pro go databases into Go playing programs (to give the bot some PRO-level opening moves), but of course it could also be a more universal coordinate system than the Western alphabet based coordinates. It would be interesting to make a new file extension like .ten or .tng that way.
Author:  Bantari [ Mon Aug 18, 2014 11:39 am ]
Post subject:  Re: Centre-based Coordinate System?
I am not sure what is to be gained by changing the coordinate system. What is the advantage, other than changing things for the sake of changing things?

For humans:
I really think that the existing system is much easier. For a few reasons:
  • a1 is easier to say that "-10,-10" or whatever. This is why we have such letter-number system rather than using the same cartesian axes but with only numbers instead, so a1 could be "1,1". And by the way, both systems are conceptually identical, except we move the origin to tengen rather than having it in lower left corner.
  • the origin of the system (a1) is always visible, while it is sometimes hard to pinpoint tengen in late positions. And this can make it really hard to count, not to mention being mistake-prone when writing down the move coordinates. We make enough mistakes as it is, methinks.
  • people are used to counting along the edges - it is easier that way, than counting along some (sometimes covered) center-lines. At least, conceptually. In practice, you could write the numbers along the edges just like the present letters and numbers, but then - what would really be accomplished? We would be in the exact same place but with different labels.
  • mistakes are less obvious and so harder to fix. When somebody record a move as "12", you have no clue if he forgot a comma (so it should be "1,2") or if he forgot a whole number (and it should be "12,x" or even "x,12".) With the present system, when you see "12" you know that the letter was not recorded, and this makes things easier to fix. In the present system, the missing move is one of at most 19 candidate moves. Under the newly proposed system, the number of candidate moves is almost twice as large.
  • there are many boards and other equipment (like software, or game-recording pads) already in existence with the present coordinate system. Reprinting/rebranding/recoding/reworking it all will take ages, resources, and valuable inner peace. Or we will have two systems side-by-side for a while, which is not good. Why is that needed?
  • and so on... I could probably think of some more reasons, but you get the drift

For computers:
It takes a one-liner and a fraction of a micro-second to recalculate from one system to the other, so also not sure what is to be gained by switching. Are there any reasons to think that either the computer programs or the databases are obstructed by the system we have? And besides, how programs internalize their data has usually nothing to do with the way the data is presented to the user.

PS>
I think that, generally speaking, when you think about introducing changes to any (existing) system, there should be a clear reasons and a clearly defined problem that the changes are supposed to fix. And besides, since each change also introduces at least a short-term problem and cost, this cost should be less than the original problem which is getting fixed.

What problem are we fixing by switching to a different notation system?
Are people complaining they cannot possibly play Go because they can't understand where the point "a1" is, while they could easily figure out "-10,-10"???
Author:  Loons [ Mon Aug 18, 2014 1:10 pm ]
Post subject:  Re: Centre-based Coordinate System?
@Bantari
I can't help but point out neither 12 nor 10 would be on a 19x19 board. So nothing but single-digit numbers for coordinates on normal board sizes would be an advantage.

I was gripped by this while writing java code to rotate boards, so. I do think it is a more elegant and symmetrical system, and go is to me a very elegant and symmetrical game. Compared to labelling one of the axis with letters a-t (except for i, most of the time) and the highest magnitude number possible. I would still put coordinates along the edges of the board, were I to make one. Hey, I should.
Author:  RBerenguel [ Mon Aug 18, 2014 1:18 pm ]
Post subject:  Re: Centre-based Coordinate System?
Loons wrote:
@Bantari
I can't help but point out neither 12 nor 10 would be on a 19x19 board. So nothing but single-digit numbers for coordinates on normal board sizes would be an advantage.

I was gripped by this while writing java code to rotate boards, so. I do think it is a more elegant and symmetrical system, and go is to me a very elegant and symmetrical game. Compared to labelling one of the axis with letters a-t (except for i, most of the time) and the highest magnitude number possible. I would still put coordinates along the edges of the board, were I to make one. Hey, I should.


I also wrote code to rotate SGF "brute force way" (so, rotated the letters! Had my reasons) and sucks.
Author:  Bantari [ Mon Aug 18, 2014 4:40 pm ]
Post subject:  Re: Centre-based Coordinate System?
Loons wrote:
@Bantari
I can't help but point out neither 12 nor 10 would be on a 19x19 board. So nothing but single-digit numbers for coordinates on normal board sizes would be an advantage.

Ah, but you don't have "nothing but single-digit coordinates". In the present system you go from 1..9 and then continue with 10..19. In the new system you would go from 9..0 and then continue with -1...-9. Either way you need a second "sign" - be it a '1' or an '-'. You simply cannot represent 19 lines with only 10 symbols, you need some kind of addition, either more symbols or more (decimal?) places.

As for you pointing out that there would be no 12, point taken.
But (almost) the same could have been said about somebody writing '9' or '3'. There is still less definition to that under new system than under old.

Quote:
I was gripped by this while writing java code to rotate boards, so. I do think it is a more elegant and symmetrical system, and go is to me a very elegant and symmetrical game. Compared to labelling one of the axis with letters a-t (except for i, most of the time) and the highest magnitude number possible. I would still put coordinates along the edges of the board, were I to make one. Hey, I should.


As I said, for computers you use the systems which are best for what they do, translating from system to system is a trivial matter in this case. I can imagine a single program could use multiple different systems to internalize the data, depending on the task at hand, and then translating from one to the other for the clarity of algorithms.

In your example, rotation of boards, I can see where the digints-only tengen-based system would be more efficient and much easier to code. But then - have a simple (hard-coded even) conversion matrix and you can move back-and-forth between the system in a blink. I would assume that some less trivial tasks, like reading/writing existing sgf files, could be done much easier if you used the existing system. Just to see a different example.
Author:  Loons [ Mon Aug 18, 2014 5:00 pm ]
Post subject:  Re: Centre-based Coordinate System?
As you point out, it's trivial to convert between systems and wrt code it's syntactical sugar. To me it seems more clear where -5, 7 is than... E17, and its relationship to 5, -7 'P3'. Though it takes a moment to adjust mentally to edge=+/- 9 (not 1, a, 19, and t).
Author:  topazg [ Mon Aug 18, 2014 6:00 pm ]
Post subject:  Re: Centre-based Coordinate System?
It's funny, nowhere in this has anyone yet mentioned my irritation that Q is the 17th letter of the alphabet but only the 16th column on the go board. Obviously I know _why_ "i" is absent, but I still get occasionally line-jumped when mentally processing the letter co-ordinates from J onwards :P
Author:  Uberdude [ Mon Aug 18, 2014 6:18 pm ]
Post subject:  Re: Centre-based Coordinate System?
@topazg: Do you actually know the ordinals of the letters of the alphabet off the top of your head? I didn't before I played Go (would just count A-B-C-D in my head), now if you ask me what number T is I'd think of a Go board and say 19 oh yeah plus 1. Go makes me think K is 10 :) .
Author:  topazg [ Mon Aug 18, 2014 7:02 pm ]
Post subject:  Re: Centre-based Coordinate System?
Uberdude wrote:
@topazg: Do you actually know the ordinals of the letters of the alphabet off the top of your head? I didn't before I played Go (would just count A-B-C-D in my head), now if you ask me what number T is I'd think of a Go board and say 19 oh yeah plus 1. Go makes me think K is 10 :) .


Yeah, I have a few benchmarks I work from, like M and N are 13 and 14, R is 18, V is 22, but I messed around with letter cryptography and other things on a simple level as a kid, and they just sort of stuck :P
Page 1 of 2 All times are UTC - 8 hours [ DST ]
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group
http://www.phpbb.com/