I claim free placement improves TPK’s whole board thinking.
2 Likes
True. I ll have to add this “helps to make the review shorter”
What’s TPK? 
20+ kyu (Twenty-Plus Kyu), to distinguish from the DDK range 19k-10k.
4 Likes
I prefer Triple Pfigure Kyu
5 Likes
BTW do people on OGS generally understand DDK to only mean 19-10k? I’ve always used and understood it as 30k (or lower if such ranks in use) to 10k. I think OGS is the only place I’ve come across TPK.
4 Likes
(if kyu>99 not possible)
DDK: [99k; 10k]
TPK: [99k; 20k]
7 Likes
I think the usage of DDK is somewhat inconsistent. I think people on OGS use it both ways, or it depends on context. I think TPK is generally much less common (maybe a newer phrase) across the Go community.
2 Likes
I propose the following additional designations to have more ranges we can talk about:
- 2+ kyu: Two-Plus Kyu, or TPK
- 3+ kyu: Three-Plus Kyu, or TPK
- 10+ kyu: Ten-Plus Kyu, or TPK
- 12+ kyu: Twelve-Plus Kyu, or TPK
- 13+ kyu: Thirteen-Plus Kyu or Teen-Plus Kyu or TPK
- 20+ kyu: Twenty-Plus Kyu or TPK
- 30+ kyu: Thirty-Plus Kyu or TPK
2 Likes
Thousand-Plus kyu? Trillion-Plus kyu?
1 Like
Indeed, TPK seems not widely used or understood. I think I heard of it first here on the OGS forums a few years ago.
Before that, I would tend to distinguish between more experienced beginners and unexperienced beginners by calling the latter group “novices”.
But that also doesn’t fit exactly with TPK for me. By “novice” I mean more 30+ kyu, which in my mind are players that may not yet be able to play and score a game without some supervision. In my mind a 20k player can already do that.
2 Likes
I think most can, but iirc I have seen games in the 15-20k range occasionally that ended with premature passing.
1 Like
Yes, 15-20k players may miss some border gap (even much stronger players may do that occasionally, including myself), but I’d expect to see this only occasionally.
1 Like
I wouldn’t limit to 15-20k. I’ve seen it in even stronger games, e.g. 14k here (I’m not counting the fake 6k beginners).
Hi, I’m here to completely sidetrack this conversation.
TPK stands for Total Party Kill, that is the only thing it will ever mean. HTH, HAND, YHBT.
1 Like
My wife has been griping at me about this ever since I told her I asked xD.
2 Likes
I’m sorry if I’ve missed this, but just to clarify: I’ve always assumed that your rating is affected much less by handicap games than even games. Is this actually the case or did I make it up? 
I’d expect that this highly depends on the game result. Winning or losing a 50/50 handicap game or 50/50 even game would win/lose a normal amount of rating points (half of the value of the volatility parameter K in Elo-like rating systems).
An expected result in a lopsided even game will win/lose very few rating points. But a highly unexpected result will win/lose many points, in an Elo-like rating system up to twice the amount as winning/losing a 50/50 handicap or 50/50 even game.
1 Like
That’s a good question. I know in the French go federation we weight the games, with minus 10% per handicap stone, so that an even game is 100%, a 1-stone game is 90%, a 2-stone game is 80%, etc. Up to a 9-stone game which only counts for 10%.
I don’t think I’ve seen anywhere a mention of something similar for OGS so I don’t know if that’s the case.
1 Like
As @gennan said, it depends, because our current rating system adjusts the rating based on an estimation of the probability of one player winning.
So if the system thinks a player has a 90% chance of winning a match, and then that player does win, the rating system can basically go “oh, I was right” and the match won’t affect their ratings much.
But if that player loses, the system will interpret this as, statistically, their ratings being very wrong, so it will adjust their ratings very strongly.
Instead, if the predicted chance is 50%, the most the system can do is adjust both rating by a “moderate amount” and hope that in the long term doing this will have good effects.
If you think about it, the simplest rating system you could build (that is still based on match results) is just one where you bump a rating up a set amount for every win, and bump down the same amount for every loss.
So in a way, forcing handicap in all ranked games would make most of the complicated maths currently applied in the system essentially useless, and it would become almost* equivalent to the simplest and dumbest rating system possible.
I expect it would make significantly wrong ratings very slow to adjust, and increase rating volatility by never giving the system a good confirmation that the rank is in fact about right.
*“almost” equivalent because handicap doesn’t completely make chances even, so the system might still be able to take advantage of the small unevenness.
For example a typical handicap match might have a predicted winning chance of 53% versus 47% (I’m making these numbers up, I have no idea), and this may or may not be enough to help the system a little bit.
So IF OGS decided to force handicap in all ranked games, at the very least please make it something like Arsen Lapin proposed, a “handicap = rankDiff -1”, or even better, something like “handicap = roundDown(rankDiff/2)”.
I believe this would help the rating system (compared to a fully forced handicap) and also partially appease to people who really dislike playing with handicap or who believe handicap is just a different game than normal Go.
I imagine the “handicap purists” wouldn’t like this very much though.
(Another one of those situations where the “average of all the proposals” might end up making everyone unhappy, I guess)
1 Like