Which spring engine game is the most popular in multiplayer

[Licho]

Run stats from same period (since start of december - last 2 months).

Battles from 1.12.2014 0:00:00
Total: 16186
Size: 0 Battles: 16 (bots vs bots)
Size: 1 Battles: 6215
Size: 2 Battles: 5289
Size: 3 Battles: 1274
Size: 4 Battles: 1023
Size: 5 Battles: 332
Size: 6 Battles: 445
Size: 7 Battles: 155
Size: 8 Battles: 245
Size: 9 Battles: 129
Size: 10 Battles: 172
Size: 11 Battles: 109
Size: 12 Battles: 99
Size: 13 Battles: 77
Size: 14 Battles: 97
Size: 15 Battles: 70
Size: 16 Battles: 90
Size: 17 Battles: 58
Size: 18 Battles: 72
Size: 19 Battles: 50
Size: 20 Battles: 44
Size: 21 Battles: 39
Size: 22 Battles: 23
Size: 23 Battles: 16
Size: 24 Battles: 13
Size: 25 Battles: 8
Size: 26 Battles: 8
Size: 27 Battles: 6
Size: 28 Battles: 8
Size: 29 Battles: 4

[Google_Frog]

Does that count games that did not end? Could you filter out games of less than 5 minutes?

[Licho]

It's only games that ended and reported to master server (MP hosted games, no crash, no cheats, no restart etc).

This is data for games longer than 5 minutes:

Battles from 1.12.2014 0:00:00
Total: 13789
Size: 0 Battles: 13
Size: 1 Battles: 5016
Size: 2 Battles: 4449
Size: 3 Battles: 1088
Size: 4 Battles: 948
Size: 5 Battles: 309
Size: 6 Battles: 425
Size: 7 Battles: 148
Size: 8 Battles: 234
Size: 9 Battles: 123
Size: 10 Battles: 164
Size: 11 Battles: 106
Size: 12 Battles: 97
Size: 13 Battles: 74
Size: 14 Battles: 94
Size: 15 Battles: 67
Size: 16 Battles: 88
Size: 17 Battles: 57
Size: 18 Battles: 72
Size: 19 Battles: 50
Size: 20 Battles: 44
Size: 21 Battles: 38
Size: 22 Battles: 23
Size: 23 Battles: 16
Size: 24 Battles: 13
Size: 25 Battles: 7
Size: 26 Battles: 8
Size: 27 Battles: 6
Size: 28 Battles: 8
Size: 29 Battles: 4

[1v0ry_k1ng]

That's 8,773 multiplayer games if the figure is correct.

In answer to the OP post then, the most popular game in multiplayer since 01-12-2014:

1) Zero K - 8773 games played.

2) Balanced Annihilation - 3072 games played.

3) Tech Annihilation - 481 games played.

4) XTA - 331 games played.

5) Evolution RTS - 241 games played.

6) NOTA - 119 games played.

That means there have been more multiplayer games of Zero K than twice that of all other spring engine games combined in this period.

[gajop]

Not all non-ZK games are recorded so you are comparing apples to oranges.
A far more interesting statistic is that there are so many 1v1 and 2v2 games going on - a scene that used to suffer in Spring.
What's up with the high amount of 3 player games? Chickens? FFA? That's a bit odd..

[Anarchid]

High prevalence of 1v1 in ZK would likely be somewhat affected by only 1v1 games being showcasted on YouTube apart from tournaments (when 2v2 gets casted every other month).

That likely attracts exactly the kind of crowd who'd play them.

Not all non-ZK games are recorded

ZK singleplayer games being saved as 1-player multiplayer games would affect this somewhat, but then Ivory's summation ignores those.

I presume non-ZK games played on non-automated hosts would be missed from the replay list, but i think the replays site actually 'watches' those as well. Do correct me if i'm being super wrong.

[bibim]

I'm not sure the number of multiplayer games is the best indicator of "multiplayer popularity". Time spent by players in multiplayer games seems better suited to me.


N.B: numbers are averaged of course, otherwise the graph would obviously be unreadable...

edit: actually this graph shows the time spent by players in all games played on the lobby server, not multiplayer games only. So I guess it's still not good to evaluate "multiplayer popularity".

[gajop]

Is that "Other mod" spike due to EvoRTS release on Steam?

[Anarchid]

The time fits: "Release Date: Apr 4, 2014".

[dansan]

I presume non-ZK games played on non-automated hosts would be missed from the replay list, but i think the replays site actually 'watches' those as well. Do correct me if i'm being super wrong.

No - it doesn't do anything that is not initiated from autohost or player side. Very seldom individuals upload their matches manually. The statistics it shows on the players pages are from SLDB (by Bibim) and afaik include also self-hosted matches.

[1v0ry_k1ng]

Not all non-ZK games are recorded so you are comparing apples to oranges.

Oh, in what circumstances is a game not recorded?

[PicassoCT]

Servername has to be "gagball"
Password "kinky"

[8611]

This is a little subjective and inexact: How I see/saw lobby when I log in at weekend EU evenings:
-Two "large rooms": 1x BA, 1x zK. Either large teams or large FFA.
-Two "smaller rooms" for each BA & zK: 2v2-3v3, maybe smaller FFA
-The TechA room, size depending on how stars align
-Smaller mods with few regular players: NOTA, XTA. Busy for a few hours, then nothing, repeat.
-BA Chickendefense room
-A few other mods being irregularly played or tested
-Some 1-3 1v1 games of BA or zK
(Purely based on observations and some screenshots)

The numbers/stats do not fully line up with that:
As said, inexact but for example all the time I see at least one BA chickendefense being played.
So 333 recorded BA-chicken games just seems low. (normal BA is ten times that)
But Chickendefense is often played in private rooms so it is not recorded, which might explain the difference between observation and numbers.
Again, inexact but one should compare stats a bit to what "one thinks makes sense."

Similiar zK having double the matches of BA does not seem realistic because in lobby it seemed about even. Of course that is not excact, but it either means:
a) zK regularly had double the games going than BA. Even if one wants to argue that zK has more rooms..*twice* as many seems unrealistic.
or:
b) zK games only take half the time to finish, so the same active room produces more matches per time. (especially 1v1 on small maps can be very quick?) (still means playerminutes are the same)
or:
c) replays.springrts records "too little" and zero-k.info/battles reports "too much"

bibim's graph seems more closely to reflect what is seen in lobby.

I'm not sure the number of multiplayer games is the best indicator of "multiplayer popularity". Time spent by players in multiplayer games seems better suited to me.

It was just to me the most easily available indicator To mind the different #players-per-match I listed how often which type happes.
If one assumes that games have similiar average duration then it seems okayish.

[img]download/file.php?id=9197[/*img]
N.B: numbers are averaged of course, otherwise the graph would obviously be unreadable...

edit: actually this graph shows the time spent by players in all games played on the lobby server, not multiplayer games only. So I guess it's still not good to evaluate "multiplayer popularity".

Can you elaborate a bit on that, like what is "time spent by players", does that also counts specs, or players idling in battleroom etc? Can you "zoom" into the april spike?

A far more interesting statistic is that there are so many 1v1 and 2v2 games going on - a scene that used to suffer in Spring.

BA "Duel": 921
zK "Size: 2": 4449
From looking at the lobby I can not explain where the 3500 extra games come from..? It never seemed that much more active to me.

It would not surprise me if a large part of the 1v1 games was played by a small number of people, devs or long-time players like Anarchid, GF, Godde, Drone playing each other:
One should keep in mind that the absolute numbers are not that high, a handful of dedicated players can easily drive up the percentages. (also see: TechA's blue_fields)

Oh, in what circumstances is a game not recorded?

The most relevant factors are imo:
-aborted games: replays.spring does generally not have them because autohosts do not upload them.
( On zero-K/battles one can filter them out by duration but not sure how trustworthy it is, there are many cases that seem strange.
http://zero-k.info/Battles/Detail/330044 - "Duration: 80 seconds" but players "died in 4 minutes" )

-private games: zK counts them, replays.springrts does not.
Any battle that in zK is named "XYZ's battle" ( http://zero-k.info/Battles/Detail/330035 )
would not show up on replays.springrts because hosting works different.

Would be interessting to see zK-numbers with only "official hosts", without the private ones.

[Licho]

As I said ZK does NOT list/count:

-aborted games (which didn't end with proper game over)
-too short games
-games with serious errors (desync) in logs
-games where someone enabled cheat

So it does not report "too much", though it's true it does report passworded games.

[8611]

So it does not report "too much",

I just used "too much/little" as in "too much/little to be easily compared" not as in "too much/little so it is broken."
Maybe better than "reports too much" / "reports too little" is just "reports more" / "reports less."

In your post viewtopic.php?f=1&t=33028&p=566112#p566018 ("(MP hosted games, no crash, no cheats, no restart etc).", did you filter these kind of games:
(must be based on info not on the page because on battles site they appear just normal)

http://zero-k.info/Battles/Detail/310292 - after few moments of doing "whatever" one players selfd's and I assume game is remade. Technically a "proper game" but very short and canceld..I think most autohosts that upload to replays.spring have a 5(?)min limit which filters most (not all) such games.
What about the "duration" sometimes not agreeing with the "player died in" info?

http://zero-k.info/Battles/Detail/83139 - one player is AFK. Not sure if zK still has the autostart-system but I also remember some of my own games where either I or the opponent was AFK. In other mods that is less likely because they have normal ready-up system.

http://zero-k.info/Battles/Detail/317122 Not sure what "desync in log" means but there are definately desynced games to be found on the site and beside player's comments they look normal on page.

[luckywaldo7]

I mean, I could play the game where one cherry-picks games out from the replay site, and I've only spent a few minutes there.

I think most autohosts that upload to replays.spring have a 5(?)min limit which filters most (not all) such games.

http://replays.springrts.com/replay/df1 ... 76904b301/
BlackHoleHost1 apparently does not have a <5 minute filter.

http://replays.springrts.com/replay/7b8 ... 51cdf4421/
An example of someone manually uploading a game with only 1 person in it.

I don't really see the point of all this, unless you have some method by which you think you can achieve "corrected" numbers for both zero-k and spring? (Besides apparently deciding what defines a "proper" game)

[8611]

Some hosts would sometimes say: "Game not uploaded, too short."
Maybe that changed and obviously I did not check that which is why I only I wrote "I think.."
Would be interessting to hear more on that.

Manual uploads are insignificant few,only about 150 total in all time:

Dansan 48
Cedna 24
root 13
[Room]Battle1 12
Jazcash 10
Danchan 8
Godde 8
[Dr]ShoX 7
thor[NOTA] 6
PepeAmpere 4
backpenther 4
burp 3
Rafal[ZK] 3
KoyoteKamper 3
zorro 2
[Fx]Drone 2
[semprini]Autohost 2
[Room]Battle2 2
[CoX]Malric 2
Friendly_1 2
DanNew 1
UploadTest 1
Aizen 1
[KH]massive 1
[teh]Teddy 1
djmad 1
law[NOTA] 1
[Evo]Forboding_Angel 1
[MARS]AlbA 1
[SmuG]flopflop 1
TheSeeker 1
[No]Culosburro 1
fishfinger 1
a1983[NOTA] 1
[teh]decay 1
ikinz 1
[wub]schlemmplatte 1
Anarchid 1
DeathStar 1
Arjack 1

I don't really see the point of all this, unless you have some method by which you think you can achieve "corrected" numbers for both zero-k and spring? (Besides apparently deciding what defines a "proper" game)

The point of this thread is to compare popularity of games in multiplayer.
That is not as easy as just taking the numbers from replay sites because they are recorded/counted in very different ways.

So when comparing numbers one has to keep that in mind and maybe come up with a method. Atm no good method has been found, yet.
For example replays.springrts.com does not record SP/private games so it would make sense to substract those kind of hosting from zK too, to get better comperasion.
If one wants to count games, then defining a "proper" game is the first step.
I do not see that as cherry-picking or botching if one explains why it is done.
I have posted my interpretation of the numbers, you can discuss it or post your own.

[bibim]

The statistics it shows on the players pages are from SLDB (by Bibim) and afaik include also self-hosted matches.

SLDB data are retrieved both from a lobby bot which monitors all that happens in the lobby, and SPADS autohosts which send GDR (Game Data Report).

The data retrieved by the lobby bot concern all matches, but lack info such as who exactly is playing and who is spectating, who won etc.
The data sent by SPADS autohosts contain much more detailed info, but obviously only concern matches played on SPADS autohosts.

So indeed, SLDB has data concerning self-hosted matches, but these data are incomplete and can't be used to compute win/loss statistics for example. That's why statistics shown in players pages don't include self-hosted matches.

[bibim]

this graph shows the time spent by players in all games played on the lobby server, not multiplayer games only. So I guess it's still not good to evaluate "multiplayer popularity".

Can you elaborate a bit on that, like what is "time spent by players", does that also counts specs, or players idling in battleroom etc? Can you "zoom" into the april spike?

These data are gathered through a simple lobby client (the lobby bot I speak about just above, which monitors all that happens in the lobby), it can't access info such as who is a spectator and who is a player in a battleroom. So here "time spent by players" means both players and spectators.
However it doesn't count players who are in the battleroom but not in game. Also, if a player exits after 5 minutes from a game which lasted 20 minutes, it only counts 5 minutes of player time for him of course, not 20 minutes.

[Jools]

The green one must be balanced annihilation?