1s per move ranked games

One icon is missing on your picture.
I say that because I only play correspondence on lichess and they do have a separate icon/rating for that.


OGS had separate ranks before and it worked bad, they were often outdated.
With most settings, 1 person has similar rank on any of these settings. Current system uses it.
But, on few settings rank will be very very different from rank on most settings.
So perfect solution would be partially separate rank.
main rank and “other” rank.

Do you guys know if this has an impact on the effectiveness of glicko in predicting game outcomes for other players ?

Because if this guy is 1d in 1s games… good to him. He’ll probably feel outranked when playing other time settings and this is not fun at all.

1 Like

As you can see, there is no consensus among the mods on this subject. The argument is that the impossibility of finding a perfect solution should stop us from having a good solution, because that would be arbitrary. This is an abdication of the use of judgment. The idea that such games should be counted in the overall rank makes a travesty of the ranking system.

An additional trick used by some blitz trolls is to restart a game that has gone to scoring. One would play 4 sec. per move and then reject the scoring and restart if he had lost. It was nearly impossible for his opponents to react fast enough to such an unexpected event, even at 4 seconds. Consequently, he won nearly all his games. I warned him and annulled his recent games that used this trick, and he promptly disappeared, probably starting a new account with a VPN. He’s probably still out there somewhere cheating people.


If more than 20% or 25% percent of the games are won or lost by time, then this is too fast.


IF there is something that easy to automatically detect and mods ban for that:
-ban always should be replaced with technical impossibility for player to do it.
I believe this is true for any PC/browser/… game

1 Like

Please forgive me if I am misunderstanding your point. I think you are saying that the restart should be technically prevented if it leads to bans in some cases of abuse. (I think a thread on that may already exist, but I can’t remember.) A case might be made for that, perhaps, in view of its wide abuse for stalling with repeated restarts as well as cases in which losers restart and win because of tesuji advice received from the new autoscore system. The restart still serves some purposes, however, in some situations: (1) when a beginner (or cheat) insists that his dead stones must be captured, so the winner restarts, captures them, and wins anyway because his lead was so large; or (2) when the autsoscore makes a mistake in scoring, and the opponent tries to leverage that.

Similarly, the most hilarious cheat I ever encountered argued that score cheating was legitimate because the system allowed him to do it. Of course, the stone removal phase (scoring) allows for honest correction of the score, and dishonest abusers of the system deserve to be banned for repeat offenses after a warning.

I tried 10 (unranked) games and I won all of them by timeout.
Apart from the last game, in none of them did I continued despite having lost a position.

I think this is a clear proof that some of the players who accept such fast games can’t really play at this speed.


I don’t talk about something specific. I talk about any situation that looks like this:

imagine other site where its be possible to place stone inside eye of group of opponent in Japanese rules, game not automatically ends, stone not automatically disappears. (in real world its possible)
And imagine that mods ban those who do it.
I wouldn’t call that other site evil only if developer just don’t have time to fix this bug.

I’m late to the party but is it possible to know connection lag averages and maybe set a minimum based on that. I.e. a minimum duration average per move that the typical connection can keep up with could be the ranked floor.

Edit: I suppose plus some “reaction time” period.
Thinking about it if we are limited to integer seconds then the answer is probably 2 anyway …


It probably means something else, but


Ducking autocorrect…


Well said ! I had this feeling, but couldn’t put my finger on it.


The fix could be as simple as removing 1s from the list of options in the UI, no back-end chance needed?

But it’s not quite as straightforward as that right? 1s is fine if main time is big enough. In that we are back to calculations of per move time for a typical game.

Edit: like all the discussion in Unusual Time setting: Canadian Byo-Yomi Blitz

1 Like

I would consider a 1s increment almost the same as absolute time, not quite enough to finish a game even if it is in the late endgame (see tricks like resumptions/unexpected pass/playing far away from last move/stalling etcetera).

1 Like

5s+1s games can reach the counting phase, I linked some examples in the first message


Although Conrad’s example above suggests that even 4s is too short to deal with this but I think in most cases 4s should be ok.

It seems to me that removing 1s when main time is below some figure would be a reasonable start and improvement. Whether to go further can be discussed further!

1 Like

Or reach it often enough to be reliable for determining rank?

The question seems to be less about what is the minimum that can achieved with fast clicking and a good connection but rather what is the minimum where some element of Go skill is involved. Are they the same?

1 Like

An average time per move of 2s might be doable (assuming ~120 moves per player to go to scoring). So something like 2 minutes basic time and 1s increment (Fisher).

It may be possible, but if 90+% of those games end in one of the players timing out, I still consider it not really doable.

1 Like