Board Game Arena

Hi,

I'd like to ask whether the current handling of abandoned games is working as intended, and suggest a possible improvement.

Table #908197138 (Catan, 4 players). Very late in the game, one player abandoned / timed out. The result was cancelled and the three remaining players were all ranked tied in 1st place, each receiving 75% of the normal ELO points.

At the moment of the abandon I was clearly leading with 8 VP (the others had 6, 4 and 4) and I would have won on my following turn. Because a tie is scored relative to rating, and my rating was higher than everyone else's, I received +0, while the two opponents behind me received +5 and +25.

I understand the maths behind this, and I'm not claiming a bug. But the practical outcome is that the player closest to winning was the only one who gained nothing, while the players furthest from winning were rewarded. That feels like the opposite of what the rule is meant to achieve, and it also creates a small incentive problem: leaving a game late can transfer points to trailing players.

Two possible improvements:

When a game is cancelled by an abandon, rank the remaining players by their actual game state (victory points, or the game's own scoring) instead of declaring a flat tie.
Alternatively, apply no ELO change at all to the remaining players, so nobody is rewarded or penalised for something outside their control.

Is the flat tie a deliberate design choice, or is it simply the simplest solution available? Thanks for any clarification.

lenadro wrote: ↑31 August 2026, 16:37 Is the flat tie a deliberate design choice, or is it simply the simplest solution available?

Both. It's the solution that works for every game, at every stage of completion. Some games have no running score that you can use to determine who's "closest to win", and in many games even the player with highest score at some point is not necessarily the most likely to win, maybe that player has already exhausted all their point sources, while another player has still some big scoring actions available. In short, there's no general way to evaluate who's winning, or closest to win, or most likely to win, so it's a flat tie.

That makes sense, thanks for explaining it! I hadn't considered how many games simply have no meaningful running score to work from.

If the conclusion is that we can't evaluate the game state, though, wouldn't the cleanest option be to apply no rating change at all to the remaining players? A flat tie is still a rated result, and its effect depends entirely on the rating gap: the higher-rated players in the table get nothing or lose, the lower-rated ones gain, purely because someone else disconnected. Nobody's actual play is being measured either way.

Leaving the ratings untouched would be just as game-agnostic and just as simple to apply, but it wouldn't hand out or take away points for something none of the remaining players caused or could control. As it stands, players at the top of the rating range effectively have their progress slowed by other people's disconnections, which happen fairly often at higher tables.

Has the "no rating change" option been considered and rejected for a specific reason? I'm curious whether there's a downside I'm not seeing.