Spring RTS Engine

I have updated my Spring lobby statistics, here are the changes:

• Bots stats are now separated from players stats => player stats more meaningful, games popularity more accurate
• Pending battles (i.e., non-running battles containing at least one non-spec player) are now separated from empty battles => avoid stats being polluted by all the empty battles
• Following games now also have their own specific stats: Tech Annihilation, NOTA and Evolution RTS
• Given how Zero-K popularity has increased lately, I separated games statistics in two categories for better graph readability: "popular" games (Zero-K and Balanced Annihilation) and "less popular" games (Tech Annihilation, NOTA, XTA and Evolution RTS)
• I removed maps stats (graphs weren't very useful, a different kind of statistic system should be used for this...)
• I updated some graphs for better readability

Graphs export location didn't change: http://planetspads.free.fr/spring/stats/

N.B: Due to the way RRD works, it may take some time before the new data has impacts on the existing graphs and is readable, especially those describing long periods of time.

N.B.2: The drop of players which appears in the graphs last night just after midnight is actually due to the separation of bots/players (yes, we have a lot of bots...)

this tool can't recalculate data for new ( less popular ) games ?

Looks good.

I smoothed almost all the graphs and added again some new ones, now I think I'm done with Spring statistics.

danil_kalina wrote:this tool can't recalculate data for new ( less popular ) games ?

I don't have data for games I didn't monitor at that time.

Here is a basic analysis of a few interesting graphs.

A decrease tendency is noticeable in the Spring playerbase over past 5 years, however it seems this tendency has reversed lately:


The popular battle sizes have clearly evolved, from 10-12 players before 2009, to 14-16 players after. Then the >16 players battles became quite popular in 2011 to the detriment of the 14-16 players battles, but this trend declined during 2012:


Games popularity also evolved, since Zero-K took the place of Balanced Annihilation during last year as the most played game:



The proportion of veteran players followed a steady increase during past 5 years. This might indicate that Spring is still popular among old players, but on the other hand that it has quite a low newbie retention rate (proportion of intermediate ranked players decreasing). However this graph is poluted by bots statistics, which weren't separated from players statistics until now...


Finally, an interesting way to visualize the battle sizes popularity during a day:

wow didn't even notice that big popularity shift from BA to ZK in 2011

The random blank graph bug is now fixed thanks to a patch from RRDTool author \o/

Could you make graphs with type of lobby utilizing ? and type of autohost

danil_kalina wrote:Could you make graphs with type of lobby utilizing ? and type of autohost

I don't have access to these data.

you have. you can analyse which lobby by CPU value

No, all lobbies don't use the CPU field wrongly.

ok, never mind

funny to see all those people playing the waiting game for the less popular mods

PicassoCT wrote:funny to see all those people playing the waiting game for the less popular mods

Less is more.

Its just that they could each have a game- if and only if, they could reach some sort of binding agreement on every guy playing one game of the other guys exoctic mod.

Its like watching game theorys prisoner dilemmas lazy cousin lurking.

Of course, instead you had a core of flexible players who would migrate across different games based on opportunity for an enjoyable experience; Flopflop, Godde, Saktoth, Google being surviving examples. I don't think anybody traded horses to get a match of their particular game.

bibim wrote:No, all lobbies don't use the CPU field wrongly.

We can only hope to see the day all clients use the cpu field wrongly, in perfect harmony with eachother, in the spirit of the brotherhood of mankind.