32-player TAW is not good for Zero-K

I left the experiment running for a few weeks, and then did not want to post stats that could bias the poll. But now we've got a good amount of data so I'm ready to post some results and make some changes.

Almost everything of interest is inconclusive, and most of what might be true is confusing. Here is the solid ground:

• !split was not used and the prevalence of multi-rooming was low beyond the first week.
• Activity on weekdays seems to have increased while weekend activity seems to have decreased.
• There were fewer large games.

As a result I will raise the room size to 32 player on Saturday and Sunday.

The least conclusive finding is the overall player count. Here is Steam average players:



Here are the site-generated unique daily multiplayer player charts:




It is all too noisy to say much, as activity is far too noisy. All I would say is from these stats is that the activity on weekends might be lower. We would need to run this experiment for months, maybe years, to see a change. So in lieu of clear player data, I have processed the battle size and duration data. The goal here is to find some changes that might feed into a mechanism that would say something about long term playerbase growth.



Here is the simplest stat: games per day. Overall there were more games. I only have the time at the moment to look at graphs and write, so I've done a count of the below- and above-median daily games.

	Days with above-median count	Days with below-median count
Games of at least 8 players	25	13
Games of at least 16 players	21	19

There were far more "good days" if we just want games with at least 8 players, and about as many good days as bad for games of at least 16 players. But games per day is probably not the most important stat for playerbase growth.



Here is the number of hours covered by games of various sizes. I think this stat is indicative of future growth since people are more likely to stick around if they can consistently find a game. It is especially important for growing into weak timezones. Here is the same breakdown of under- and over-performing days.

	Days with above-median coverage	Days with below-median coverage
Games of at least 8 players	26	13
Games of at least 16 players	16	22

Coverage is much better for medium games (8+) and worse for large (16+). But here is where it gets weird: the improvement is driven by weekday behaviour. There are 20 above- and 8 below-median days during the week, compared to 6 above and 5 below on the weekend. The weekend has better baseline coverage, but for some reason the change has brought the weekdays closer to the weekend baseline.



Finally, the playerminutes. See https://zero-k.info/Forum/Post/271928#271928 for an explanation of this plot. What we can see now, with more data, is that the weekends were probably hurt by the change, while the weekday results are inconclusive. Saturday is was below the median six weeks in a row, while Sunday had two low outliers. This supports the depressed weekend indication from the unique player charts.

The unsurprising part of these results is that the weekends were negatively affected, at least from the perspective of playerminutes and running 16+ player games. It makes sense that, if there was little multi-room behaviour, then you'd see fewer games. The main point was to test whether we would see more multi-room behaviour. The surprising result is that the timezone coverage of the weekdays improved. For most of the week the 22 player limit is not even relevant. My best just-so story is that long huge games (22+) put a strain on lobbies via player exhaustion, and on weekdays there are too few players coming in to keep the lobby alive.

This gets us to the result stated at the top: raising the room size limit to 32 on the weekend. The weekend saw no significant benefit in terms of timezone coverage, and paid for it with reduced large game coverage and playerminutes. In terms of player preferences, 30% of the poll said they wanted to play larger games, so all else being equal, it seems fair to raise the limit at least some of the time. If we were seeing significantly more 16+ player games, then I would probably keep the limit at 22, since 16+ is already quite large, but that is not what we saw. Friday could go either way in terms of defining the weekend, but it saw a significant increase in 16+ games and coverage, so I will leave it at 22 and see if the trend continues.

Switching the rooms around twice a week is an annoying manual task, so I don't know how long it can go for without automation. But I would like to see if the weird result during the week persists.

Seems to me like the experiment is running for the sake of running the experiment, and needs to be stopped. Here's why:

• It didn't work. "The weekends were probably hurt by the change, while the weekday results are inconclusive." Split command was not used and "multi-rooming was low beyond the first week." Multi-rooming is the desired result of the experiment, and if full time limit doesn't work, part time has even lower chance of forcing player behavior. The best the experiment would do is to make both weekdays and weekends inconclusive.

• It's strongly disliked. The data is noisy, what isn't noisy is player opinion. As of now polls say 27% strongly dislike, 12% dislike, while only 4% of people strongly prefer it. It's obvious people don't like the player limit, and I dare say nobody likes the fact that they can be randomly kicked to spec. Giving players the stick is not a way to make the game better.

I appreciate the experiment but perhaps it should end sooner than later.

i think most days have a certian time that the lobby is full its like 3 hours were the TAW is 22+, we all got used to dose hours and know that if we want a full game we should come at does hours.

I think it is frustrating for me, when i cant join an already full 22player Match in Workdays. I want to play full games with lots of people, I like 32player very much in the last years.

Do keep in mind though that 16v16 not being your favorite game mode to play doesn't mean you don't like it at all or don't want it to be an option. I, for one, generally prefer to play smaller team games over very large team games, but I still like those 16v16 games, and sometimes prefer them, and the fact that they exist as an option at all is incredible.

For the 32 player saturday change, can we do that starting from Saturday in the first timezone? or at least not the last. At least set it to Saturday for Asia or Europe. Now we have to wait for east coast US? Which is literally the last timezone?

"

Unfortunately, this is happening more and more often.

"

yes a room colapses when it looses too many players.. with huge rooms they had a natural defence.. now its like they have lower hitpoints

Zini & Smokedragon nailed it: With a smaller TAW room size, it collapses faster.

A thing that might improve the experience for both new/casual and the more competitive players is a kind of culture shift where they play in different rooms.

With the current setup, high ranks could join palladium, low ranks could join niobium.

I've seen and participated in populated palladium evenings.

Niobium may not see much play because the people that'd join it probably just prefer going to the most populated room they can join.

I generally avoid the TAW room when it has a high % below yellow rank. Blue or even yellow is ok (maybe palladium's min rank is too high atm).

Regarding minimum palladium entry, I view silver stars as try hard journeymen. They don't tend to have huge gaps in their play and will fill most roles competently. I can't say the same for mature gold ranks, who I see as sometimes having pronounced strengths in particular niches, but either being too inflexible or too unaware to meet the needs of their team.

Ofc, seeing this in terms of colours is very low resolution, but if I were to draw a line in the sand to filter for players capable of adapting to cutthroat palladium environment it would be around there.