| 1 |
Here is some early data collection.
|
1 |
Here is some early data collection.
|
| 2 |
\n
|
2 |
\n
|
| 3 |
https://i.imgur.com/wPis6hq.png
|
3 |
https://i.imgur.com/wPis6hq.png
|
| 4 |
\n
|
4 |
\n
|
| 5 |
This is a stacked plot of the players in teams games over the past 13 days. The stacks reveal the room type breakdown as well as peak players across all rooms. The values are ingame players, not people waiting in the lobby. The bars are the maximum for each lobby over a 10 minute period, to bridge the gap between games. This makes the plot more readable. The dotted black line is the average spectators across all games, heavily smoothed for readability.
|
5 |
This is a stacked plot of the players in teams games over the past 13 days. The stacks reveal the room type breakdown as well as peak players across all rooms. The values are ingame players, not people waiting in the lobby. The bars are the maximum for each lobby over a 10 minute period, to bridge the gap between games. This makes the plot more readable. The dotted black line is the average spectators across all games, heavily smoothed for readability.
|
| 6 |
\n
|
6 |
\n
|
| 7 |
The top contains information about each day. You'll probably have to open the full size image to read it. The main metric is playerminutes per day, which is the sum of the time ingame spent by all players. This uses game durations, so is unaffected by the 10-minute bucketing of the stacked plot. The top also shows how many games of at least four players were at least five minutes long. I also listed the average duration of these games, and a breakdown of game size.
|
7 |
The top contains information about each day. You'll probably have to open the full size image to read it. The main metric is playerminutes per day, which is the sum of the time ingame spent by all players. This uses game durations, so is unaffected by the 10-minute bucketing of the stacked plot. The top also shows how many games of at least four players were at least five minutes long. I also listed the average duration of these games, and a breakdown of game size.
|
| 8 |
\n
|
8 |
\n
|
| 9 |
The room size and waiting list changes happened at the start of Friday the 11th. The waiting list only took on its final form late Saturday. Prior to that, the information was not sent to clients properly, which made it a bit confusing. No split has been used so far, but multiple rooms were still able to run. This is early data so we should be careful not to read too much into it. The strong Wednesday is particularly suspicious, since the new game size cap barely comes into play.
|
9 |
The room size and waiting list changes happened at the start of Friday the 11th. The waiting list only took on its final form late Saturday. Prior to that, the information was not sent to clients properly, which made it a bit confusing. No split has been used so far, but multiple rooms were still able to run. This is early data so we should be careful not to read too much into it. The strong Wednesday is particularly suspicious, since the new game size cap barely comes into play.
|
| 10 |
\n
|
10 |
\n
|
| 11 |
https://i.imgur.com/YAuBoRc.png
|
11 |
https://i.imgur.com/YAuBoRc.png
|
| 12 |
\n
|
12 |
\n
|
| 13 |
Here is every day of 2025, grouped by day of the week. The green dots are baseline, the red dots are the days since the new room size limit. The Wednesday stands out as the 3rd best Wednesday of the year, while the other days are fine. This is enough to say that the experiment is worth continuing, and it has at least not reduced the available games. It is very promising if the results so far hold up, but my gut says that some of the effect size has to be due to external factors. The weekend will say how reliable multi-rooming is, and may even see a test of !split this time.
|
13 |
Here is every day of 2025, grouped by day of the week. The green dots are baseline, the red dots are the days since the new room size limit. The Wednesday stands out as the 3rd best Wednesday of the year, while the other days are fine. This is enough to say that the experiment is worth continuing, and it has at least not reduced the available games. It is very promising if the results so far hold up, but my gut says that some of the effect size has to be due to external factors. The weekend will say how reliable multi-rooming is, and may even see a test of !split this time.
|
| 14 |
\n
|
14 |
\n
|
| 15 |
Here are some fairly speculative thoughts that draw a bit too much from limited data:
|
15 |
Here are some fairly speculative thoughts that draw a bit too much from limited data:
|
| 16 |
* It is unclear whether the the 22 player limit will promote "small" (<= 10 players) games. The breakdown of small games does not show a trend. This is fine, because promoting these games is not one of the main goals.
|
16 |
* It is unclear whether the the 22 player limit will promote "small" (<= 10 players) games. The breakdown of small games does not show a trend. This is fine, because promoting these games is not one of the main goals.
|
| 17 |
* It is also unclear whether the number of large games (>= 16 players) is affected. The 22 player limit seems to be producing more games overall, many of which are large.
|
17 |
* It is also unclear whether the number of large games (>= 16 players) is affected. The 22 player limit seems to be producing more games overall, many of which are large.
|
| 18 |
* The games with the 22 player limit are shorter by 2-3 minutes on average. This is probably due to a lack of huge games.
|
18 |
* The games with the 22 player limit are shorter by 2-3 minutes on average. This is probably due to a lack of huge games.
|
| 19 |
\n
|
19 |
\n
|
| 20 |
https://i.imgur.com/MwOZxpe.png
|
20 |
https://i.imgur.com/MwOZxpe.png
|
| 21 |
\n
|
21 |
\n
|
| 22 |
I
can
also
scroll
around
the
data
from
earlier
in
the
year.
A
lot
of
it
looks
like
this,
with
the
above
being
an
extreme
example.
People
would
pile
into
one
host,
play
about
8
hours
of
games,
then
fall
off
a
cliff.
Is
this
due
to
a
long
match
that
finally
tired
everyone
out?
A
map
was
picked
that
people
would
rather
not
play?
This
is
probably
where
some
anecdotes
could
augment
the
data.
Sometimes
the
peak
bounces
back
and
there
are
another
few
hours
of
games,
sometimes
it
doesn't.
Spreading
this
peak
out
a
bit,
to
more
reliably
shift
into
another
six
hours
of
games,
seems
important
for
building
a
the
America
timezone
playerbase.
|
22 |
I
can
also
scroll
around
the
data
from
earlier
in
the
year.
A
lot
of
it
looks
like
this,
with
the
above
being
an
extreme
example.
People
would
pile
into
one
host,
play
about
8
hours
of
games,
then
fall
off
a
cliff.
Is
this
due
to
a
long
match
that
finally
tired
everyone
out?
A
map
was
picked
that
people
would
rather
not
play?
This
is
probably
where
some
anecdotes
could
augment
the
data.
Sometimes
the
peak
bounces
back
and
there
are
another
few
hours
of
games,
sometimes
it
doesn't.
This
seems
to
have
improved
a
bit
over
the
past
few
months.
Spreading
this
peak
out
a
bit,
to
more
reliably
shift
into
another
six
hours
of
games,
seems
important
for
building
a
the
America
timezone
playerbase.
|
32-player TAW is not good for Zero-K