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

Rank -> actual skill
http://www.lifein19x19.com/viewtopic.php?f=10&t=7122		 Page 1 of 2
Author:  Annihilist [ Sun Nov 04, 2012 10:54 pm ]
Post subject:  Rank -> actual skill
How much does rank actually correlate to skill? Are there grey areas or overlaps in different ranks, or is it fairly cut and dry? Thus, is it worth designating a handicap by the discrepancy in rank between competitors, if it is not always the most accurate measure of skill?

I ask as I much prefer even games, yet I don't like negating handicap settings of open games on KGS or the like. So this is more a discussion on whether rank really should be the gold standard of skill level, or if it's a more arbitrary and vague indication of general player knowledge or competency.

Thoughts?
Author:  Tami [ Sun Nov 04, 2012 11:31 pm ]
Post subject:  Re: Rank -> actual skill
I think it's very difficult to be certain about how rank and skill relate. Yes, the better you are at go, the higher your rank will be, but there are many other factors that could have an influence:

A might have greater go skills than B, but B might be better at handling tournament conditions, and so achieve a higher rank in over-the-board tournament play.

B might be better at playing OTB tournaments, but be uncomfortable with computers, and so might have a lower KGS (or other online) rank than A, who is good at using computers.

C might be better at go, tournaments and computers than both A and B, but might be afflicted with poor health, so that he or she can only play at their best 10% of the time.

I believe practical skills and personal skills (meta-skills) are very important, anyway. Again, if you want to get a higher ranking, then the priority is to work on your go skills, but it`s pretty likely that improving your meta-skills will give your performance a boost, too.

It might be worth remembering that your ranking is not a measure of your worth as a human being. It`s not even 100% reliable as a measure of your combined go and go-related skills. After all, people vary day by day, and grow (and decline).

As for not liking to play handicap games, just divide your time. You don`t have to play handicap games on every occasion. Also, if your partner agrees, there`s always the option of using a reverse komi instead, so that the game will feel like an even one, but will still be interesting for the stronger player.
Author:  Boidhre [ Sun Nov 04, 2012 11:36 pm ]
Post subject:  Re: Rank -> actual skill
Rank is just one number. You could have a 10k with the fuseki for a 5k, or one with the fuseki of a 15k. The former will be weak somewhere else, that's why they're 10k while the latter will be very strong somewhere else. Ditto life and death strength, midgame fighting strength, yose strength etc. It's not uncommon at all to have players with poor fuseki strength but strong fighting skills for their rank. The opposite is less common as far as I can see. It's highly imperfect, people even when accurately ranked vary in strength considerably day to day for various reasons like tiredness, level of distraction etc. Rank mostly just dictates the level of player you can generally beat or how many handicap stones you can take from players of certain strength. Does this correlate with skill and understanding of the game? Yes, but not absolutely perfectly.

The point of handicap stones though is to make things interesting for both players. While I play even games against stronger players quite often, it's them doing me a favour, I play a lot of handicap too lest the games become too one-sided and boring for them. That said, reduced handicap games can be quite interesting for both sides I've found while still giving the weaker player a good teaching experience.
Author:  Toge [ Mon Nov 05, 2012 2:08 am ]
Post subject:  Re: Rank -> actual skill
Rank is meant to be an expression of ability, measured as success against opponents. It's very difficult to say what this "ability" then consists of. Two players play a game, there can be only one winner.
Author:  SoDesuNe [ Mon Nov 05, 2012 3:02 am ]
Post subject:  Re: Rank -> actual skill
Rank measures progress and has little to do with skill (especially on internet Go servers).

There are countless examples on KGS, where players achieve a 7d-9d-rank by winning on time against high handicap. Unlike in tournaments many players just play for the fun on the internet, trying new things and so on and thus don't show their true skills.

EGF ranks are very different in almost each country. In Germany many players claim Dan-ranks but actually don't have enough GoR. In my experience czech dan-players are stronger than other equally high dan-players. If I remember correctly, you have to be awarded Shodan by a commitee in the Netherlands.

So, if you personally gain a rank, don't matter where, you personally have made progress. But you can't say your skill is lower, equal or higher compared to anyone claiming the same rank.
Author:  Falcord [ Mon Nov 05, 2012 3:13 am ]
Post subject:  Re: Rank -> actual skill
There is also the matter of how seriously you take your own rank, specially in online go servers.

For instance, when I started I took my rank very seriously. I was used to competitive videogames, and keeping my rank was important. Therefore, whenever I played a ranked game I focused really hard and played as if it was a tournament.

Nowadays I'm more relaxed. I use ranked games to experiment too, and if a game is not going in a direction I think will help me improve, I might surrender earlier than my past self would. This probably means my rank is more loosely related to my skill.
Author:  tapir [ Mon Nov 05, 2012 3:36 am ]
Post subject:  Re: Rank -> actual skill
skill = rank

If you believe you are much more skilled than your rank shows, because you had a cold during the last tournament, you are fooling yourself. Don't succumb to feel-good excuses, but obsess about your skills and the lack thereof, not about your rank.
Author:  Mike Novack [ Mon Nov 05, 2012 6:31 am ]
Post subject:  Re: Rank -> actual skill
Tami wrote:
C might be better at go, tournaments and computers than both A and B, but might be afflicted with poor health, so that he or she can only play at their best 10% of the time.


This is the one example with which I agree there might be a large discrepency. Surely there are parallels in go to somebody like Nimzovich in chess. However do take note of that later comment about not deluding oneself. Real examples of this sort of thing would not be common.
Author:  jts [ Mon Nov 05, 2012 7:27 am ]
Post subject:  Re: Rank -> actual skill
tapir wrote:
skill = rank

If you believe you are much more skilled than your rank shows, because you had a cold during the last tournament, you are fooling yourself. Don't succumb to feel-good excuses, but obsess about your skills and the lack thereof, not about your rank.

Particularly because people have such awful trouble understanding that generalizations like ranks are probabilities and as such are (gasp) a bit unpredictable. Even quite intelligent people who haven't studied stats and probability theory are likely to look at a string of randomly generated numbers and find patterns in them. (This string of tails is where the penny-flipper really was able to focus and put out the results his team needed, but the pressure got to him so later he snapped and had four heads in a row....)
Author:  SmoothOper [ Mon Nov 05, 2012 7:44 am ]
Post subject:  Re: Rank -> actual skill
It is pretty well known to not be true. The classic example is in professional honorary titles. Essentially once a 9 dan always a 9 dan, though in practice players do not always play with 9 dan skills, though generally the players are still revered for their teaching, promotional, and organizational value.
Author:  v00d00 [ Mon Nov 05, 2012 8:03 am ]
Post subject:  Re: Rank -> actual skill
Rank represents skills.
Author:  Fedya [ Mon Nov 05, 2012 8:18 am ]
Post subject:  Re: Rank -> actual skill
If you don't like your current rank, play a couple dozen serious games, and you'll get the rank you deserve.
Author:  Uberdude [ Mon Nov 05, 2012 11:56 am ]
Post subject:  Re: Rank -> actual skill
Maybe I am self-deluded and have an inflated belief in my own abilities (feel free to say so!), but I do think that (some) of my Go skills are better than my rank. And by my rank I mean 3d. My current GoR is a weak European 3d, but a decent British 3d.

I think I am relatively good at those wishy-washy strategic things like direction of play and judgement and techniques, but I am relatively weak at reading (and that's the singly most important skill in Go). I am also mentally weak, making many blunders, and poor at time management and playing in byo-yomi. I remember a quote from Go Seigen recalled in one of JF's books about how a large part of what lifted him above the other 9 dans was not his technical Go skill (as they were all pretty darn amazing), but his mental toughness, self-belief, desire to win etc.

So what evidence do I have that those skills are better than my rank? Well just in my last tournament I beat a 6d who said he was surprised I was 3d, I played more like a 5d. I responded that maybe I play like a 5d in main time and 1k in byo-yomi, so 3d is about the average. I was fortunate in that game that I didn't make any blunders and managed the byo-yomi ok, later in the tournament I lost to a 2d. But that's just one game, and every so often maybe a 3d does beat a 6d.

A less anecdotal example is OGS, where I am 7d (not that there are enough dan players there to make that particularly meaningful in comparison to other rating systems). But I do play better there than in real-life tournaments and this could be interpreted as more of a test of a players "true skill" (where such skill excludes things like coping under time pressure). It also makes reading skill less important as you can play out variations on a board. On OGS at the moment I'm current playing Alexander Dinerstein 3p (7d European) and the game is very close (one yose I played out I won by 1.5). One could argue that if he ends up winning even by 0.5 points he was just doing that to be cool and you can't make any judgement of my skill from that close margin, but I think if my skill was only 3d level I wouldn't even manage to make it a close game. Of course you could also argue that if other 3 dans put as much time and effort into playing as I do on OGS they'd be that good too...
Author:  tapir [ Mon Nov 05, 2012 1:07 pm ]
Post subject:  Re: Rank -> actual skill
Of course there are many legitimate arguments to be made about different strength in different situations, partial skills etc. Most players I know, myself included, will tend to use it as comfort to not face their record. Like losing on two stones against someone, but believing you are just as strong "if you could focus in the endgame" or "if you would bother to learn joseki" or whatever. I know the feeling, when someone nominally stronger seems suddenly within reach, but it is invariably better to keep silent and work instead of comforting oneself to complacency with fancy ideas about your hidden skill that fails to manifest in actual games. That is just the flipside of stagnation.

But uberdude is doing the opposite. His rank on OGS is just as real as his EGF rank and he is busy to explain it away. I mean, you are playing a close even game against the multi-time european champion and it wasn't even you who challenged for that game, but he challenged the top players on OGS, if I recall correctly. That is a testament to your strength and you shouldn't hide behind your EGF rank and "well, I am sometimes better than my real 3d rank". Obviously, you are. :tmbup:
Author:  TheBigH [ Mon Nov 05, 2012 2:03 pm ]
Post subject:  Re: Rank -> actual skill
I do know that online ranks are unreliable from 30k to about 15k. At those ranks you get sandbaggers, strong players with new accounts that haven't had a chance to rank up yet, people who've learned to pummel the some of the weaker bots but can't handle human beings, and weirdoes who play bizarre moves for fun. I once played a 20k who played out a non-working ladder from one corner to the other and after the game told me he thought the zigzag pattern was pretty.
Author:  oren [ Mon Nov 05, 2012 2:16 pm ]
Post subject:  Re: Rank -> actual skill
TheBigH wrote:
I do know that online ranks are unreliable from 30k to about 15k.


I think you can remove 'online' from that statement, and it would be more generally accurate.
Author:  Bill Spight [ Mon Nov 05, 2012 2:49 pm ]
Post subject:  Re: Rank -> actual skill
There are many go skills. (I want to say that there are at least two dozen, but I don't really know. ;)) Rating or rank is a kind of average of them, related to handicap. It is not transitive. That is, player A may be able to beat player B in an even game, player B may be able to beat player C, and player C may be able to beat player A. Rank depends upon the other players, not just in terms of skill, but in terms of their mix of skills. For instance, in the example of A, B, and C, we expect that they will have the same rank. But suppose that A moves away. Now B may be one rank stronger than C. ;)
Author:  xed_over [ Mon Nov 05, 2012 4:46 pm ]
Post subject:  Re: Rank -> actual skill
Bill Spight wrote:
But suppose that A moves away. Now B may be one rank stronger than C. ;)

But does he have A's shoes?
Author:  golem7 [ Wed Nov 07, 2012 5:12 pm ]
Post subject:  Re: Rank -> actual skill
To be exact: ranks don't represent skills, but results. More precisely, results in a certain environment (e.g. online play or live tournament) under specific rules according to a specific rating system.

So it's very hard to compare different ranks but in the end results are the only measure of go ability we have.

One could say, ranks measure the skill to win a game of go (to place your stones more efficiently than your opponent over the course of a whole game) under certain conditions (time settings, atmosphere etc.). But even with the same rank there can be big differences in specific skill areas (opening, joseki knowledge, fighting, endgame) that eventually cancel each other out. Also one might be more comfortable with some styles of play than with others.

All in all I believe if a rank results from playing games on a regular basis under comparable conditions then it's a fairly accurate representation of your skill (of playing in just these conditions).

But if you really want to know how strong someone is, you'll just have to play them yourself and find out.
Author:  Annihilist [ Wed Nov 07, 2012 5:19 pm ]
Post subject:  Re: Rank -> actual skill
golem7 wrote:
To be exact: ranks don't represent skills, but results. More precisely, results in a certain environment (e.g. online play or live tournament) under specific rules according to a specific rating system.

So it's very hard to compare different ranks but in the end results are the only measure of go ability we have.

One could say, ranks measure the skill to win a game of go (to place your stones more efficiently than your opponent over the course of a whole game) under certain conditions (time settings, atmosphere etc.). But even with the same rank there can be big differences in specific skill areas (opening, joseki knowledge, fighting, endgame) that eventually cancel each other out. Also one might be more comfortable with some styles of play than with others.

All in all I believe if a rank results from playing games on a regular basis under comparable conditions then it's a fairly accurate representation of your skill (of playing in just these conditions).

But if you really want to know how strong someone is, you'll just have to play them yourself and find out.
So do you think handicap settings should default to the discrepancy between the ranks of the players? Shouldn't an even game be the more preferable default between players who have not played each other before?
Page 1 of 2 All times are UTC - 8 hours [ DST ]
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group
http://www.phpbb.com/