95.92.146.253: **Value of handicap stones for amateurs vs ranking systems**
(2010-12-23 22:43) [#8190]

This is something which has been bugging me lately, so I'll try to make it clearer in my head while writing this post.

Let's put apart rule sets for a minute as that's not what this is about.

If playing an even game, white gets 6.5 komi advantage.

If playing a 1 stone handicap game, white gets 0.5 komi advantage.

If playing a 2 stone handicap game, white gets 0.5 komi advantage but black gets an extra stone to begin with (although he can't choose where to put his first stone anymore).

Therefore, ranking tells us 1 handicap stone is equal to 6 moku.

But what is the reason for this to hold true for amateurs? Is there any place where we can look at even amateur game statistics so to check that each side wins around 50% of the time? In my opinion, most single digit kyus would rather play with -24 komi than a 4 stone handicap game. That's because the meaning of close game also changes depending on your rank.

More generally, in my opinion, 1 handicap stone is worth more than 6 moku.

Given that, doesn't it somehow correlate to the fact that the difference between the ranks isn't really 1 stone on most online servers? (this has been stated in many places around SL)

When playing online, you get to play more players around your own rank than if you play at a go club. Now, if we really consider the value of each stones to be bigger than the value of the komi, shouldn't 1 rank difference games be played with komi 6.5 and 2 handicap stones? 2 ranks difference with komi 6.5 and 3 handicap stones and so on?

Bottom line, what do you consider more difficult against someone 1 stone weaker: 0.5 komi for white or 6.5 komi for white and 2 handicap stones for black?

-- miguelaraujo

tapir: **Re: Value of handicap stones for amateurs vs ranking systems**
(2010-12-23 22:53) [#8191]

the european go rating counts no komi games as a half stone handicap only, if i recall correctly. also general belief is that a single move is worth twice the komi (that is 12+ moku) - if you ask me it has something to do with the introduction of komi in even games, while handicap games are still largely no komi handicap games (either no komi or 0.5 komi) - which makes the jump from no komi game to two handicap very large while the difference between komi and no komi is rather small. but in the local club nobody forces you to play komi games anyway.

HermanHiddema: **Re: Value of handicap stones for amateurs vs ranking systems**
(2010-12-24 09:15) [#8192]

Actually, the value of handicap stones is twice the komi.

Consider the difference between Black starting with a move, and Black starting with a pass. That should tell us the value of the first move, relative to not playing (i.e. passing). If Black passes, White gets to start the game. Effectively, Black passing is equal to reversing the colors, except for the fact that White still has the komi. So to see the value of that first move, we consider:

Game 1: Black starts, some normal game is played, Black scores X, White scores Y Game 2: White starts, exact same game is played, so White scores X, Black scores Y

Now add the komi (K) for White in both cases to get the final score:

Game 1: Black scores X, White scores Y+K, so result = X - (Y+K) = X - Y - K Game 2: White scores X+K, Black scores Y, so result = (X+K) - Y = X - Y + K

And the difference = 2K.

If we make the assumption that each handicap stone is worth roughly the same as the first move (it certainly seems reasonable to think that at least the first four corner moves are of equal value), then the value of each handicap stone is 2K. So removing komi as the first step in handicap is actually only half a stone handicap, and further handicaps are actually 1.5, 2.5, 3.5, etc.

95.92.146.253: **Re: Value of handicap stones for amateurs vs ranking systems**
(2010-12-26 02:47) [#8199]

That goes as my post suggests.

My question was a step further: can't that be a (big enough) reason for rank differences in online servers not being representatives of 1 stone?

As the majority of the games played are even or 1 stone apart, aren't the ranks being separated by half a stone differences, instead of 1 stone?

Shouldn't online servers change their handicap system to reflect (and correct) this?

-- miguelaraujo

HermanHiddema: **Re: Value of handicap stones for amateurs vs ranking systems**
(2010-12-26 11:52) [#8200]

I think most servers handle this correctly, the concept that the first handicap stone is not a full stone is widely known.

The KGS rating math page at http://www.gokgs.com/help/rmath.html tells us that 1 rank is substracted for each handicap stone, while an different adjustment is made for each point of komi. So the first handicap step, which just removes komi, is not treated as a full rank by KGS, but as some fraction of rank appropriate for removing around 7 komi (6.5 for Japanese, 7.5 for Chinese style rules)