thanks for the statistical effort     GoogleFrog ! quote:
I like how, so far, the teams hosts have been running for a larger proportion of each day. This seems important for attracting players. Although I would not rule out running 32 player games on particular days. |
i agree. i thought i am a definte vote for the 32p roomsize, but actually, the experiment told me the 22 player rooms are fun enough and 2 pots running worked. but some days switching it up to 32p rooms will be neat.
+3 / -0
|
I understand you all, but that was the big games that always pleased me. It's sad one of the last (the last) game that allowed that prohibit it now. Well, there is other things in life than videogame. I wish you great games.
+6 / -0
|
|
|
FWIW, based on the data and summary my view is changed from neutral to positive on this change, with the caveat that I think there should be an occasional day or event that allows for 32+ rooms for fun. Thanks Googlefrog for the work you've done on this.
+4 / -0
|
Looking at this plot, yes it is indeed more colorful, but that's not necessarily the effect we want. We want there to be multiple colors at once (i.e. multiple rooms), but this plot just shows players switching from one room to another while the old room dies. The rare times that the colors overlap over occurs for just a few dozen minutes, and they existed before the change as well.  I would argue the important thing is keeping the lobby alive, and that's a bigger issue than having multiple rooms. When a lobby is dying it doesn't matter if it have a 32 or 22-player limit, but having the 22-player limit brings frustration and can cause players to quit early--especially for the players who were playing but got kicked into spectating, they can be very frustrated and quit the game. (I know I would.) I stand by the opinion that 32-player is what makes Zero-K so unique, and I hope maybe the 32-player limit can be reverted (perhaps temporarily at first and see how the data fluctuates. This would also eliminate doubts about the past week data being special because the past week's been a lucky week)
+1 / -0
|
Lobbies dying when they get small is unsurprising, and there are split lobbies from before the experiment. However, the ones in the graph are Palladium, and most from earlier this year are either Palladium or the !split command. We are seeing more non-Palladium overlap with 22 players. Each grid line is six hours apart, so a relatively small overlap can still be two or three hours of games. That is quite a few parallel games. The proportion of lobby-death splits is lower than before, however it could be higher in absolute terms since there are more splits in general. I would argue that the important thing is increasing the times at which players can log in and play a game that they would like to play. This encompasses three things:
-
Whether there are any games running at the time a player logs in.
-
Whether they can find a spot in a game when it ends.
-
How closely the game they find matches their preference.
Preferences will vary, but it seems reasonable to weight these priorities from top to bottom. Having any game seems more important than having the exact game you'd like to play. I'm aware of players who are in the habit of logging on regularly to check for their favourite game type. This just seems like a hardcore behaviour, so it would be one that can sustain a community, but not really grow it, because new players won't have the habit. The most consistent feature of the data so far is that active lobbies extend into more hours of the day. There are more times when you can find a 4v4 or larger. On the more active days (ones that some might argue are the days relevant to the experiment), it is easier to find an 8v8 or larger. At this point it is even unclear whether there will be more 11v11s, which is a pretty large game. More consistent 11v11 at the cost of 16v16 seems like a trade many people would take. Getting kicked into spectating is annoying, but it is in service to two goals. The first is that spectators of a game are no longer punished (as hard) for trying to seed another game, since they can hop back to the main room and play. The second is that anyone who just logged in has a spot in the active game. The cost is that people can be pushed from playing after playing a few games, or even just one game if the room is building up. For everyone who plays a game and ends up in the waitlist, there is a player who waited for the game to end and then would have had to wait for a whole game to play. It's a tradeoff, and perhaps it will encourage the more dedicated players to go seed new rooms.
+7 / -0
|
expected queue times are lower for people that join rooms, but the rewards of seeding a new room are also lower because one can get moved out of being an active player by people joining in. consider this example for a new 4v4 room : - someone creates and seeds, people join and eventually adds up to 8 then they play - while they're playing, 8 new people join and start spectating - after the game ends, the 8 new people that joined after it was full displace the 8 that just played even the creator and the people that seeded it when it was almost empty get cycled out
+0 / -0
|
I think the simplest thing is to automatically split the game into 8v8. So when there are 32 players, you can create two full games. It doesn't make sense for a player to choose the other game and move, because one game is the same as the other and can't enter it. It's just a matter of getting used to it at first.
+0 / -0
|
quote:
More consistent 11v11 at the cost of 16v16 seems like a trade many people would take. |
What makes you think so? The polls show large differences between 11v11-13v13 and 14v14-16v16.   11v11-13v13 is actually the least voted option. If games will be more "consistent" is uncertain, so I expect little trust by players. On the social part, I think from forum posts and experience that players are actually not very eager to play. If the match does not fit their preferences (size) then they seem to prefer doing something else or wait instead of playing in a smaller match.
+4 / -0
|
Given that the clearly stated goal is to increase the number of people playing, objectively, the only way to measure whether 22 player limit works is for look at when the match totally fills to 22 and then other rooms seed. If more players are in a game playing after TAW 22 maxes, then it works. Otherwise, it does not work. Any other analysis purely is subjective. GF already responded to me that under the above analysis, the 22 limit failed. However, he wanted to add subjective analysis to support a contrary finding (adding when the 22 room is not full), which simply is not how experiments work. Experiments work by changing 1 variable and controlling for all others. Speculating that more people play just because TAW is set to 22 - even when it is not full, is pure speculation. The only possible exception would be the game immediately after TAW falls below 22.
+1 / -0
|
quote:
The only possible exception would be the game after TAW falls below 22. |
Not sure I understand what you mean. Hypothetically a host of 20 player being active for 10 hours versus a host of 32 player being active for 1 hour, can mean that "more people play" or "that people play more", both without having the host reach its limit. Besides, not sure if already decided, but a good experiment should also define a period BEFORE starting the experiment AND compare with a similar period in the previous years. There is quite a monthly variation between player number. The thesis of the experiment as I understood it is "smaller host might determine more parallel hosts, which might attract more people because it is less time waiting/and play is more engaging, which means people might stay longer and play more". Not sure if anybody thought that next day by putting the limit we will be "flooded" by users. How many new users are per week anyhow? Would we even notice if "more" would stick around or play couple of games more?...
+0 / -0
|
I mean that unless TAW is maxed at 22, nothing that happens has any relevance to the max being decreased to 22. Put another way, it is objectively impossible to determine whether anything that happens, when TAW is not maxed at 22, is related to the room max being set to 22.
+1 / -0
|
that does sound like a good argument to me a 8 player game should not be counted in the stats =P there was 4 22 player games played in last 24 hours
+0 / -0
|
My group of friends (cabrion, aioeieoiao, tratata, BETEP, Kabanos, TPABOkyp, AliyKorol and few more guys) completely left Zero-K precisely because of Lobby 32 cancellation, so I'm playing alone now. Excellent "improvement".
+5 / -0
|
I think the best way to sort this out is a zero-k battle with the winning team getting the team size limit they want. Only thing is one side can have 16 players, the other can have 11. On a more serious note, I like the idea of experimenting, and I personally hope the old limit returns and not because I love 32 player matches. I would tire of 32 player matches as I would of 22 or 14 or whatever. I really enjoy the variety and that different tactics are needed or available in different size matches just like with maps. Based on when I play, I have never seen it maxed at 32 regularly for extended periods and while it could become a problem (this would be a great problem to have to fix imo), it does strike me as solving a problem I don't see. It would be great if solving for not always having 32 player matches didn't result in never having 32 player matches.
+4 / -0
|
From what I've sampled, it's exactly the same type of games at 22 except that pala rushing is just not really feasable anymore. Arguably good.
+1 / -0
|
"Getting kicked into spectating is annoying, but it is in service to two goals..." I like the intention, but giving players the stick and forcing them to behave, to start a new room instead of playing, is just crazy. A game doesn't succeed by forcing players against their will.
+2 / -0
|
quote:
... The first is that spectators of a game are no longer punished (as hard) for trying to seed another game, since they can hop back to the main room and play. The second is that anyone who just logged in has a spot in the active game. |
Neither of the goals are sticks.
+2 / -0
|
i think time limt on games will have a better effect then max 22, as 32 games are ussaly longer, meaning if your not in the game you ahve to wait more. so having a 22 max isnt effective in my view. just do sudden dath ( units lose health at minute 20 or something...)
+1 / -2
|
|
|
