Wheb it comes to Go tournament systems in the UK, it doesn't get much better than the McMahon. Am ideal way to get even games while the bar, a crucial component, Still makes things fair. However, could we be using the McMahon execcively, and losing out because of that?

I will try to make this brief since my pandanet match is coming up, so:

1: Many people improve faster when given the chance to play strong opponents in serious games.

2: It puts too much emphasis on rank. The rank you put when entering a tournament for the first time suddenly becomes very important, and if you're a quickly improving kyu player? Man...

3: Ironically, too much McMahon causes the rating system to go hey-wild, especially when there are many handicap games.

I'll put this here for now and follow up later. But one idea: is it possible to implement layered-McMahon? One Mcmahon point for every three Kyu ranks, making a fairer tournament while simultaneously giving evenly-matched games.

_________________
On Go proverbs:
"A fine Gotation is a diamond in the hand of a dan of wit and a pebble in the hand of a kyu" —Joseph Raux misquoted.

Huh? Don't get it. In a MacMahon tournament, you win and move up, playing against stronger opposition. It's one of the best organisational methods for that.

_________________
Geek of all trades, master of none: the motto for my blog mostlymaths.net

In a 3-round McMahon, the strongest opponent a 10 kyu might gey to play is a 7kyu. In a triple-layered McMahon, that number becomes 1 kyu. In any case, winning one game lets a 10kyu play a 9-7 kyu opponent.

_________________
On Go proverbs:
"A fine Gotation is a diamond in the hand of a dan of wit and a pebble in the hand of a kyu" —Joseph Raux misquoted.

I have played some chess, and I wouldn't change MacMahon for Swiss. Playing your first game against a first seed is pretty demoralising

_________________
Geek of all trades, master of none: the motto for my blog mostlymaths.net

Of course, in 95+% of tournaments, a swise makes completely no sense. But maybe it is possible to use something midway in a some? Almost zero tournaments in the UK use something other than a McMahon. So I suggested the "triple-layered McMahon" as an example for something "midway" between McMahon and a swiss.

_________________
On Go proverbs:
"A fine Gotation is a diamond in the hand of a dan of wit and a pebble in the hand of a kyu" —Joseph Raux misquoted.

You can reset your rank if you improve quickly. I reset from 10k to 8k, 8k to 5k, 5k to 3k, and 1k to 2d.


Yes, but under your proposal when the weaker player plays someone 3 rather than 1 stone above them (or 6 rather than 2) then your opponent is playing someone 3 stones rather than 1 stone weaker so they don't improve so fast and could get bored beating you up. Swings and roundabouts.


[citation needed]


Evenly-matched? If a 10k who wins plays a 7k who lost rather than a 9k who lost then that game is less evenly matched and there will be fewer close games. Fairer? How so? Do you think lots of people who enter at 10k are really 3k? My experience of UK tournaments (and I've been to dozens) is that whilst there are occasionally new players with wrong entry grades, they and the organisers usually do a pretty good job of picking a sensible initial rating, and most people are solidly graded players anyway.

As an aside, if the ultimate goal of your rating system is to properly handicap games, then ideally a majority of your games would be handicap games.

Sadly I don't have too much to say for McMahon.

I remember playing club tournaments with 3 rounds and winning all 3 games only to turn out in the 3rd or 4th place in the tournament.

Realistically speaking in big tournaments like Euro go congress where you have 400+ players with 5 rounds if you are a 10 kyu who plays on high dan level you can win all 5 games only to end up on 10th place or something.

That seems fine to me. Otherwise people would have too much incentive to under report their rank.

I think McMahon is a pretty darned good system for Go tournaments. But you can get the unfortunate situation though in which a player just below the bar wins all their games, including against the winner of the tournament: http://www.europeangodatabase.eu/EGD/To ... n=10225083. So there Matt Crosby 2d (bar at 3d) beat Christian Scarff 2d (1 win), Chang Li 4d (dubious rank, 0 wins) and finally Alex Selby 4d (2 wins), whilst Alex beat Paul Christie 3d (1 win), then Alastair Wall 4d (1 win), then lost to Matt. So Alex did have slightly harder first 2 games, but not much, so it does seem rather harsh on Matt to not be able to win the tournament by beating the winner in his final game!

There's actually two players there from the second group that won all their games. There's no fair way of saying whether Alan Thornton should be above or below Matt Crosby.

Well, there's SOS, which they drew on (bad luck for Matt that Chang Li lost all games). But Alan didn't beat the tournament winner like Matt so I think it's understandable for Matt to feel more aggrieved than Alan to not win. With only 3 rounds this kind of unfortunate situation will arise occasionally, I'm not sure there's much one can do to avoid it. Other tournament systems without this disadvantage would probably have bigger other ones.

I manually applied this to the current OGS Simultaneous McMahon title tournaments. For 19*19 tournaments, one McMahon point = 2 kyu difference; For smaller boards, 1 McMahon point = 2.5 or 3 kyu difference.

Example

It is possible and it has been done. My own preference for tournaments in Europe is to make a layer by rating, and ignore the ranks, apart from displaying them at the end of the tournament. That's what the poorly written software I made does anyway, unless it crashes first.

_________________
North Lecale

I'd say that since Alan never got the chance to play Alex, its hardly fair to mark him down for it.

But yeah, with only 3 rounds, and such a large group of evenly matched players (I'm counting nine 2d players just below the bar), you're going to get strange results sometime.

I ran a 5 round tournament once (http://www.europeangodatabase.eu/EGD/To ... y=T090905B) where a player from just below the bar won all 5 games and got second place behind a top group player with 4/5. But in that case, they didn't even play each other...

Sometimes, fate conspires against us

I have actually had the same idea that you call layered McMahon and I think it would be better, but I have no way to know, as there ist no such tournament.

It just makes so much sense: you have McMahon, to guarantee appropriate opposition, then you have the top bar to give all strong players a chance for tounament victory, because standings are otherwise unfair against lower ranked players. Now, why should this be limited to the top. Why shouldnt someone in the lower ranks be given a fair chance to place above the guy who is ranked two stones higher and a real risk to place below who is two stones lower by making them play similar opposition(at least from the start, they might diverge if they really do better/worse then you, as expected) by basically having a bar every couple of stones? Its not like high-dans would sudenly need to play mid-kyus, so games would still be interesting for everyone.

The problem is that it is also true if you play at your 10 kyu level. You might well win your 5 games, meeting only players as strong as yourself, while the 7-9 dan players, all meeting each other, have 4 wins only.

To fairly get the first player right among 400, you need at least 9 rounds.
To fairly rank everybody according to their number of victories alone, you need 399 rounds.

To be valid, the McMahon system should be able to place above the bar only those players with a chance of winning. It can't really well cope with 6-dans declaring as 1-kyu. I suppose that another option for Elom to avoid the regular McMahon would be for him to organise a league on KGS / IGS / OGS for BGA members (and recognised by the BGA). Such a league could have stronger players in it, and could be submitted for rating with the EGD.

_________________
North Lecale

There is already the BGA online league: http://britgo.org/bgaleague.html. Maybe Elom would like to play in that? Its teams are usually geographic, but I'm sure Elom could find a place to play (I once played for Belfast when Cambridge had no team). There's also a youth team, for which he would probably be first board so get to play lots of stronger players.

Ironically, the year i mentioned this, Sandy Taylor was unfairly robbed of his win in the London open due to the McMahon System. So I've been vindicated? And maybe the solution is that when using McMahon, the final result should be based on your score combined with your SOS for a S+SOS score.