Page 1 of 2
Has anyone ever "broken" a game by making an excessively long "turn"?
Posted: 15 April 2026, 16:50
by FrankJones
Hey all,
So, I had an unusual occurrence. I was playing a solo game in which it is possible to do dozens, perhaps hundreds of "moves" on a single "turn". Completing and ending the "turn" requires clicking "confirm end of turn".
But, because my "turn": had 100 or more "moves", the BGA interface seems unable to submit the move in the time required. Thus, every time I click "confirm end turn", it tries for exactly 15 seconds, after which I receive the same error: "unknown network error (504)."
Apparently, this error means the server did not respond before the (presumably) 15-second time limit, causing the attempt to "time out" and fail.
Progressing in the game is not possible. Fortunately, it is a solo game and won't incur a penalty, btu the same sequence is possible (though unlikely) in a person-versus person game.
Re: Has anyone ever "broken" a game by making an excessively long "turn"?
Posted: 18 April 2026, 15:14
by Exotic Dreams
I have never had this come up before. One would hope switching to turn based or abandon game would become an option.
Re: Has anyone ever "broken" a game by making an excessively long "turn"?
Posted: 18 April 2026, 15:32
by FrankJones
Exotic Dreams wrote: ↑18 April 2026, 15:14
I have never had this come up before. One would hope switching to turn based or abandon game would become an option.
Switching to turn based wouldn't help - the game is irretrievably "stuck". If it were not a solo game, the only option would be "mutually abandon game", and the opponent might not agree. There have been examples of other games in which the player had to forfeit or run out of time because of a "stuck" situation resulting from a rules bug or because of having no valid move.
The situation I encountered was odd because the rules work fine and I completed my turn, but it was too much information to be submitted and confirm the end of turn.
A very rare occurrence unlikely to occur in a competitive game.
Re: Has anyone ever "broken" a game by making an excessively long "turn"?
Posted: 18 April 2026, 17:04
by Jellby
Travel to France and submit your turn there? Perhaps being closer to the BGA servers/headquarters will allow sending all required information faster

Re: Has anyone ever "broken" a game by making an excessively long "turn"?
Posted: 18 April 2026, 17:24
by FrankJones
Wait. All joking aside, is this possibly a problem that would not be occurring if I tried submitting this move from a different device? I assumed the convoluted 200-move turn was the problem, but maybe that's not the only issue?
[I just tried using my iPhone, same error.]
Re: Has anyone ever "broken" a game by making an excessively long "turn"?
Posted: 18 April 2026, 17:24
by RicardoRix
Quite often, the norm is to break down turns into actions and each action is sent/received. DB's updated, ect. So that's why an F5 will not normally lose you your progress on the turn.
So not only that it is unusual for a game to have this 100 moves action, but also how it has been programmed is very unusual too.
What game is the one you are referring to?
It might not just be a server load, but more how the data is being handled and that might make it fail. You might be breaking some array buffer or other. And/or might just be some unwitting game developer bug.
Re: Has anyone ever "broken" a game by making an excessively long "turn"?
Posted: 18 April 2026, 17:34
by FrankJones
RicardoRix wrote: ↑18 April 2026, 17:24
Quite often, the norm is to break down turns into actions and each action is sent/received. DB's updated, ect. So that's why an F5 will not normally lose you your progress on the turn.
So not only it this unusual for a game to have this 100 moves action, but also how it has been programmed is very usual.
What game is the one you are referring to?
It might not just be a server load, but more how the data is being handled and that might make it fail. You might be breaking some array buffer or other.
The game is "Earth".
This game is unusual in that players can take "anytime" actions, even when not the active player, and while the clock is not running but the opponent's clock is running.
Depending on which type of "anytime" action is taken, and whether these actions are taken during a player's turn or at the end of a turn, these actions may be initially logged as "unconfirmed actions", which can be undone sequentially back to the start of a long series of such actions, and these actions are not confirmed and complete until confirmed at the end of the turn. They might be "hidden" from the opponent and the game log until the end of turn is confirmed.
These "confirmed actions" account as a move number. So, during this series of actions, the move number was incremented by 200-300, which I'm guessing is highly unusual. For comparison, a highly complex sequence in Terraforming Mars might increment the move counter by 30, because the turn consists of two actions, each of which might have 10-15 effects, each of which gets a separate move number.
When viewing a game replay of "Earth" and clicking the arrow that advances one move at a time, there are sometimes a lot of consecutive clicks that seemingly do nothing except increment the move number. I eventually realized each of these such clicks were advancing what was logged as an "unconfirmed action."
I also tried "undo all", and received a similar but not identical error message.
Re: Has anyone ever "broken" a game by making an excessively long "turn"?
Posted: 18 April 2026, 17:43
by RicardoRix
yeah that move number is essentially a view into server/client interaction(s). So it would seem that it's not the large data. Each move has already been recorded. It's more that it's just too much the game server to process the confirming/undo for some reason. Maybe the sheer number of DB interactions perhaps when confirming the move. This also ties in with the fact it is a 500 error, and that the device you use does not change the error.
I am of course purely speculating, but it would be my best guess. I don't think it's a client-server load. Maybe raise a bug with the game dev. They would be the best to confirm or deny. BGA wouldn't like to get involved till they know it's definitely not the game code.
I think but don't quote me, that the move number is directly related the server 'notifications' that are sent. Simply because a 'game replay' is a play back of the notifications one at a time.
Re: Has anyone ever "broken" a game by making an excessively long "turn"?
Posted: 18 April 2026, 17:56
by FrankJones
My coding knowledge and experience is far from advanced, but based on what I know and understand, everything you said makes sense. I'll post in the bug/suggestion for the game; perhaps the developer can shed some light on the situation.
Re: Has anyone ever "broken" a game by making an excessively long "turn"?
Posted: 24 April 2026, 01:19
by FrankJones
RicardoRix wrote: ↑18 April 2026, 17:43
yeah that move number is essentially a view into server/client interaction(s). So it would seem that it's not the large data. Each move has already been recorded. It's more that it's just too much the game server to process the confirming/undo for some reason. Maybe the sheer number of DB interactions perhaps when confirming the move. This also ties in with the fact it is a 500 error, and that the device you use does not change the error.
I am of course purely speculating, but it would be my best guess. I don't think it's a client-server load. Maybe raise a bug with the game dev. They would be the best to confirm or deny. BGA wouldn't like to get involved till they know it's definitely not the game code.
I think but don't quote me, that the move number is directly related the server 'notifications' that are sent. Simply because a 'game replay' is a play back of the notifications one at a time.
Trying things differently, I got this error instead:
Unexpected error: Error: generated notifications are larger than 128k (137234) (reference: GS1 24/04 02:10:06)
Any idea what that means?
I googled it. Apparently, the move size is too large. The google search even referenced BGA.