Rank - as given by Igowin

    Keywords: Question

What about Igowin ranks?

Igowin is a popular, free, MS-Windows 9x9 version of a go playing program, The Many Faces of Go. More and more, players' first taste of our game comes from Igowin (kudos to programmer David Fotland for helping to promote go). One feature of Igowin is that it estimates what your rank is when you play it for several games.

How "accurate" is it? As long as you play only Igowin, its rank is good enough for your needs. But when you play other people, you will find that Igowin tends to overestimate your rank. (For example, on KGS I'm 10 kyu; based on AGA tournaments, I'm 11 kyu; but Igowin thinks I'm at least 6 kyu. Sometimes it mistakes me for a 3 dan.)

These comments may be interesting:

(the following one is copied from Jurgen):

  Confused: The rank display by Igowin
  is extremely unrealiable and tends to overestimate you by 8 to 15
  kyu. Going by my personal experience, if you can beat Igowin at
  even games reliably, your rank should be 20 - 25 kyu on KGS.

(this one's new) Jan: For what it's worth, I'm 12 kyu at the Go Club Utrecht, but playing Igowin, I can sometimes get to 3 kyu. More often, I just happily click away (possible bad habit: Playing as fast as your opponent?) and reach 6 or 7 kyu.

Monteo, 1 dan: However, on 9x9 board, Igowin Pro's rank estimates are quite consistent with GoQuest. It gives me 1dan, and sometimes I beat 2-dan players there. Another point is that pro 19x19 go players may be weak at 9x9, and that is why some go programs can beat 19x19 pro on 9x9 board. (For example, see [ext] http://teytaud.over-blog.com/article-mogotw-wins-the-first-ever-9x9-game-against-a-top-pro-as-black-38407487.html). If the program plays againt top 9x9 go players, such as 'spaceman' or any 8 dan players at GoQuest, the result might be different.

Herman: 19x19 skills transfer perfectly well to 9x9, and 9x9 is also played, for prize money, by professionals. They are definitely not weak at it. The reason computers beat professionals at 9x9 earlier than at 19x19 is because computers deal better with smaller boards. If GoQuest ranks line up with Igowin ranks, then in all likelihood the GoQuest ranks are as inflated as those of Igowin are.

Monteo (1dan) Three points are interesting. First, if computers deal better with smaller boards, human players should deal better with smaller board as well. But, advanced go teaching has long been focusing on 19x19, which can be seen from the number of advanced 19x19 go books and 9x9 ones. Second, I don't think the ranks promoted by GoQuest (GQ) and Igowin are inflated if we split go between the 19x19 and the 9x9 go--which share the same rules but have quite different arts of play. In 19x19 go world, we have a number of 9 dan players, in 9x9 world (e.g., GoQuest), we have only one 9 dan player and few 8 dan players. We need more 9x9 go game results to show that 9p 19x19 top pro can beat 9d 9x9 top player at GoQuest (GQ). Evidences at GQ show that pro players play on the larger board size (13x13) better than the 9x9 one, based on their ELO scores. Third, it remains to be seen whether Lee Sedol 9p can beat Ohashi Hirofumi 6p 19x19/9 dan GQ or not, on the 9x9 board size; I think he cannot.

Computers use a different process to choose moves than humans do. That it why they were always much stronger at smaller board play than they were at larger board play. It seems clear that actual playing strength at 19x19 and 9x9 in humans are not equal. The main reason for this might be that most Go players don't care very much about 9x9. You lose the depth of the game when you transfer it to a small board - although understandably 9x9 has its own charm and some people enjoy its challenges. So - not knowing, or having much experience at 9x9, most regular Go players are weak at. If they practiced a little they might well get stronger.

Grauniad: I presume Confused means that if you can beat Igowin at even games reliably, but not when Igowin has a handicap, then your rank is 20 - 25 kyu on KGS. I think this judgement is hard, and that you have to be better than 20 kyu on KGS to achieve this. In any case, we shouldn't expect ranking against Igowin to say much about ranking in general because 9x9 go is so different from 19x19 go!

Jasonred : I used to think this was harsh, but I estimate my own rank at probably 20 kyu-ish, maybe weaker... and I can beat Igowin reliably when Igowin has 2 stones... presumably, this is cause, as you said, 9x9 is too different from 19x19... Strangely enough, my weakness is FIGHTING, and running battles, and any other harsh battles, so why in the world can I manage this on Igowin???

Confused: Yes, Grauniad, that's what I meant. I started to play on Igowin and later on KGS. While Igowin is nice for training small scale infights and life and death, it doesn't touch a lot of subjects relevant when playing on a 19x19 board, like influence, tenuki, getting ahead and opening. In an even game against Igowin, it's fairly easy to win just by building along the center line and pushing. On the 19x19 board, I ended up losing games on the whole board while winning infights locally in the corner.

I had the impression that compared to other players around 20 kyu, people who played Igowin before were stronger at infighting while having a weaker game development. Working on those areas will give a very visible and quick improvement of the game.

Anonymous: I find the kyu translation from igowin to kgs to be pretty accurate. With mild effort I can fairly reliably defeat igowin when I go first (ie. I win at 9k and lose at 8k). Today, when I tried harder to look for expansion that igowin feels it must respond to, I've gotten it down to 4k (igowin has 2 handicap stones). I am currently 22k on KGS. (I have also developed that bad habit of moving too quickly from playing against igowin)

ilanpi First of all, the best way to measure your rating against Igowin is to look at your average rating (in the Statistics menu), not the rating it changes after each game. You will note its idiosyncratic use of fractional part, that is, 2.11K is stronger than 2.10K, so that improving from 2.99K leads to 1.00K.

After playing about 3,000 games with Igowin, I believe that its strength should be about 17K on KGS. I base this on being able to give it 3 stones (when I concentrate, which is hard when you play it constantly), so about a 13 rank difference, as estimated here [ext] http://jakobsen.dyndns.dk/edo/pages_danish/edo_open071203.htm and my current rank of 4K on KGS. Stronger evidence, in some sense, is that this is about the upper rating limit at which I can give 3 stones to human players on KGS.

Actually, I believe that my 9x9 strength is somewhat higher and maybe about 1K (whatever that means, basically, I have no clue about 19x19 openings, but this defect in my game does not affect my 9x9 game, therefore I should do better against opponents of my rating who usually know more than me about openings). My current average rating on Igowin is 1K, so my feeling is that its rating system might be fairly accurate from about 1K and up.

You may have noted that the previous two paragraphs are inconsistent, in the sense that they do give the same rating estimate for Igowin. I can resolve this by saying that I disagree with the 13 rank difference for 3 stones, I think it is bigger and consistent with me being 1K and Igowin 17K.

I should further note that the accuracy at the higher level implies that the Igowin rating system is inaccurate at weaker levels, since Igowin gives you 9K for beating it with black, but if you accept my above remark that Igowin is 17K this should only give you at most an 14K rating, since non komi 9x9 corresponds to about a 3 rank difference (see the above website and I do agree with this estimate).

This last remark also implies that the jump from 8K to 7K (I mean, the level where you switch from playing black to playing white) is very big and corresponds to about 6 ranks!! This is because you switch from playing non komi with black to non komi with white.

It is my completely subjective feeling that Igowin generally plays weaker moves when given 2 handicap, as opposed to when it plays black in an even game.

Finally, I've decided to stop playing Igowing for a while, because I'm getting used to its bad moves, and this is hurting my regular go game against humans (or Gnu Go, which is much stronger).

Spathever: I've been playing against igowin for some time and I've reached a point where I can play about 14-16 ranks consistently (2 stones handi). I tried GNU go and I managed to beat it in an even match when it was playing at its full power. Of course the pace was a bit different and I guess you can conclude that much of the outcome was because of the pace. Might be my personal style too, though...

revo: I must have a psychological problem with igowin. I'm currently a KGS 15kyu, but I think my real EGF rating could be around 9-12kyu (I rarely played on KGS the last months). But it's still very hard for me to beat igowin when it takes two stones, so I mostly play it between 5 and 8kyu. Once I managed to get a jigo at 3kyu, but it never happened again. Maybe the reason is that I don't take igowin games too seriously, but more as a "I'm bored, let's click around a bit"-game. But it somehow confuses me that there are 10kyu players who claim to beat igowin at dan level.

dlb: I am ranked 30kyu on KGS, but hover between 9kyu and 7kyu on igowin. The igowin program is somewhat predictable

isshoni: i m rated 13k on kgs, and igowin rates me 6,67 (max: just over 7). I've played igowin thousands of times though. In good streaks i ve occasionnally risen to 1d and even 2d. My level against Igowin seems to be more correlated with the age of the captain than is my level on kgs.

Imagist: My ranks are KGS 4k and Igowin 1k. I used to be 16k and Igowin 5-6k (last time I played Igowin). Maybe Igowin ranks correspond better to actual ranks as ranks get higher?

LukeNine45: I'm not sure-- I'm 3k or 4k on KGS and depending on the moment I could be anywhere from 4k to 6d on Igowin. My average is 1k right now, though. (later: I'm now 1k on KGS and 1-5d on Igowin, averaging at 2.5d)


Say, could someone add the Igowin ratings to the chart in this Wiki?

Igowin's ratings have no relation to real world ratings basicly, win one game - rating increases by 1; lose one game - rating decreases by 1 (*unless its an overwhelming win/loss)

  • 5 stones - 25-23k
  • 4 stones - 22-20k
  • 3 stones - 19-17k
  • 2 stones - 16-13k
  • 1 stone ("even", no komi, playing black) - 12-9k
  • 1 stone ("even", no komi, playing white) - 8-5k
  • 2 stones - 4-1k
  • 3 stones - 1-3d
  • 4 stones - 4-6d (can't remember if it goes to 5 or 6d)
  • 5 stones - Pro dan

LukeNine45: I second this. "Rating" as given by Igowin has little correlation with reality, even if you go with the average it gives. It over-estimates the ddk's by something like 8 or 9 stones and me by something like 3 or 4.


Rank - as given by Igowin last edited by 110.168.55.193 on September 24, 2016 - 01:59
RecentChanges · StartingPoints · About
Edit page ·Search · Related · Page info · Latest diff
[Welcome to Sensei's Library!]
RecentChanges
StartingPoints
About
RandomPage
Search position
Page history
Latest page diff
Partner sites:
Go Teaching Ladder
Goproblems.com
Login / Prefs
Tools
Sensei's Library