Anti-escaping and Anti-stalling features

Hi All,

We’ve rolled out a couple of new features to help mitigate some annoyances that pop up from time to time. The first, our “Anti-escaping” change is that now when someone leaves a live game without first resigning, after 30 seconds the remaining player is given the option to end the game:

Because this is sometimes a misunderstanding by beginners, the person who left the game is also presented with a Toast letting them know that’s going on.

In this way, if you have reason to believe that your opponent is having some connection issues and there’s a chance they’ll come back, you can hang around and wait for them - otherwise if you think they rage-quit or just left because they didn’t know any better, you can claim your victory after 30 seconds instead of waiting around for 5 minutes.

The other annoying thing that happens from time to time is “stalling”. This is when an opponent continues to place stones instead of passing to end the game, or passing and then resuming from the scoring phase repeatedly.

Now, if the game is reasonably far along (specifically width * height / 3 width * height / 2 moves into the game, so move 180+ for a 19x19 match), and you have passed 3 times in a row, the server will ask KataGo for who it thinks is the winner of the game. If KataGo is at least 99% sure that one of the players has won by more than 10 points, we present the option to end the game in favor of whoever KataGo thinks is going to win:

In this way, if you have someone who is just playing useless moves at the end of a game to annoy you or something of that sort, we figure if you can afford to pass three times and still have KataGo say you’ve got a 99% chance of winning, then you can quickly claim that win and move on to your next game.

– anoek


30 seconds? It’s not enough to fix internet. At least 1 minute would be more useful.


We’ll see how it goes. No one is required to click the button right when it pops up, the logic is that most people will be happy to wait a little bit if they think there’s a chance they’re coming back. At the same time, if you’ve obviously won and you have the feeling they rage quit, 30 seconds is better than 1 minute.


I suspect this is a social experiment :stuck_out_tongue:.


Hmmm, a little bit is just a bit vague.
How long is a little bit?

Well that depends on the person in the context of the quote right?

How long would you be willing to wait if your opponent has left the game and is no longer online?

I think some people will be ok to give it a minute or two if their opponent has internet issues or other problems. Other people just want to play Go, and so will want to move on and keep playing.


I like these. They certainly seem better than my idea, which was to replace the “your opponent disconnected” notification with one that read " It appears as though your opponent has forsaken their honour and is running like a whipped dog!"

Pretty sure I don’t need to say from where I lifted that line :slight_smile:


From 224 games, here’s the breakdown of how long it took for folks to press the button (measured in seconds)


You know what I just learned through this experiment? Our player base is pretty awesome.

I know all the pessimists are going to say, “Hey anoek, look at that huge spike of < 45 seconds! It’s too short! People are instantly clicking it for a win!”. But I have bad news pessimists, I went and manually went through all 65 of those games, and guess what I saw? 59 were immediately recognizable as either a rage quit or abandonment, the remaining 6 I had questions and reservations about. However, upon further analysis two were very likely legitimate disconnects and either didn’t come back that day or didn’t come back for 15-20 minutes, 1 was a clear abandonment / rage quit as they immediately started a new game, another was a likely rage quit because they were getting trounced (the graph was showing they were winning, but upon further consideration, they were playing a handicap game and it was clear that plot was consistently going down, and it was pretty early in the game), and the remaining two also turned out to be what I believe rage quits because while KataGo thought they could recover, they left after an invasion and I think they just didn’t have the super human capabilities to see that not all hope was lost.

So. Out of 65 instances of people clicking that button in under 45 seconds, I found no abuse and I’m really glad they had that button to click so we didn’t waste the time of those fine players.

OGS players are awesome :slight_smile: And I think 30s is a good time for the anti-escaping button to be available.


total or from each player?

Also, unless something was left out - for example,
does a resumption reset the passes-in-a-row counter? -
this handles the main version of score cheating:
Resume enough so that you can pass 3 times in a row.


It does not reset the counter, so yes if you have someone who is resuming from the stone removal phase repeatedly this can help with that.


It would also help if they’re wrongly marking stones,
rather than also or instead resuming from scoring.


Is OGS considering any software solutions to score cheating? - #8 by IDe

reminded me of a potential issue here:
What about when KataGo’s confidence depends on whose turn it is?

In such cases, does ​ whether-or-not the other player is also passing ​ matter?


I figure if you can pass three times while letting your opponent place stones, and KataGo still gives you >99% chance of winning, it’s probably pretty safe.


So, we get a pass-fight if it’s swinging 99.9% for whoever’s turn it is?

(trying to be the first to have passed 3 times in a row, when
there’s an anti-seki where neither player can find the right move)

However, I now realize that is better than, giving the game to whoever is willing to wait there longer.

Why would anyone purposefully pass if it will tank their winrate? And what are the chances two people would do such a thing?

(At any rate, two passes puts the game in the scoring phase :wink:)


They would purposefully pass if it will tank their winrate when, most sequences are
bad for whoever starts, and each player would be more likely to go wrong than right.

(Resuming from the scoring phase puts the game back into a playing phase. ​ :slight_smile: )

I have a feeling in such a case the game probably gets reported anyway, players can’t agree on a score, resume pause resume pause etc.


I don’t understand what the escaper is supposed to see. If I close the tab during the game, the tab is just closed and I don’t see any notifications.


That brings me a question. Does it work only if you stay connected to OGS? So,excluding leaving OGS, closing all tabs, closing the browser, disconnecting the internet, and other cases maybe?


too long text in result in history
horizontal scroll bar always appears even with max width window


and this B+10.4 makes no sense