Loading...
  OR  Zero-K Name:    Password:   

cancel auto split public game lobby

5 posts, 274 views
Post comment
Filter:    Player:  
sort
8 months ago
the reason why i say cancel is it because i wait for a game to finished, it was a full lobby with more than 35+ and it split me and some guy out to the other lobby , we all left and rejoin the old room because people looking for big fight, now i have to wait again and there is a chance i will get split to new room again?

cancel the auto split public game lobby, people are willing to wait for a game. if they decided to make a room they will make one.

+2 / -0


8 months ago
Thanks for the feedback. I saw two splits in the logs and thought the first worked well, while the second at least failed gracefully. It wasn't perfect but might still have a positive outcome. I'd like to figure out how to solve failed splits.



There were two splits last night, two hours apart.

The first split was after a battle on DeltaSeigeDryDeluxe. The split put 44 people into two games of 23 players and 15 players. The drop from 44 to 38 seems natural to me, as some proportion of people will stop playing after any given day. Perhaps it indicate that !split was called almost immediately after the game ended.

After the split, Teams All Welcome #2 persisted in parallel for two more games. Games happened in the original Teams All Welcome at around the same time, so it makes sense to add them for a total player count.
  • The first round had 38 players (23 + 15)
  • The second round had 39 player (26 + 11)
  • The third round had 39 players (29 + 10)
Overall, 20 extra people got to play. Teams All Welcome #2 shrank over time though, which seems fairly natural and hard to fight. Perhaps if there are two teams autohosts active, of a certain size, then the larger one could be hidden from people logging on to look for a game? That seems pretty questionable though.

It looks like the rooms merged for a game on Taldarim_V3, possibly because the third game in Teams All Welcome #2 was quite short. After this game, there was another split.

The second split 41 players into rooms that initially had 24 and 17 players. But the 17 player room must have died because it doesn't show up in the replays. The 24 player side of the split also ended up being a 30 player battle on Otago. My theory is that the split happened almost immediately after the game ended, but then many people called it a night because they had just played a 59 minute game. At least, this was my theory when I first looked at the game, since two splits two hours apart is suspicious.

Each of the 30 players in the Otago game was part of the 41 player split. Two of the missing players were split into Teams All Welcome #2, while the remaining nine left Teams All Welcome. BErankSaber was the other one and said he was done for the night. So it is certainly bad that 15 players had to fiddle around to get back into the original room, and that MYrankHougo missed out, but I would rather work on the remaining issues than remove the system. It seems to be doing well overall, and even the failed split wasn't catastrophic.
+3 / -0
8 months ago
Idea: can split decision be also made based on the time of day? There will be a natural "peak time" (in which splits should happen) followed by "low time" (in which splits should happen less). Or something similar (don't do split if number of players wanting to play games decreased over last 30 minutes),
+0 / -0


8 months ago
In theory it could change depending on time. It seems easy to screw up though, and the active times can change from day to day. Making split dynamic seems quite tricky at the moment.
+0 / -0
split 2 games => more people get to play , 20% chance people fed up and leave.
no split => all wait and wait for their turn everyone stays

compare to the old times without splits , i think no split makes more player stays?
+0 / -0