Spring RTS Engine

I collected the info of the last ~650 maps I played on in the past year and a half, various mods, but mostly *A mods.

Some very nice info can be gained from them:


Nothing special, as expected, 20 is dominant value, with the many 0s attributable to land maps.


This is area actually, not size, but the root of it shows a pretty even distribution from small to large. Area is map height*width in spring units.


Wow, there is hardly any variation of gravity in the mainstream, with mostly speed maps providing the low gravity environments.


A closely gaussian distribution! Can it be that spring mappers, as a collective, have stumbled on a great way to vary gameplay?

This graph gets even better with wind values culled to 25 (the max in *a mods):


If you want it, I can post the spreadsheet of my data.

label your axes!

Y axis is value for that attribute.
X is always number of map.

Those axes need to be flipped and labeled... it makes no sense to have ## of maps at that value on the x axis.

Wait, these look like PMFs or something... i donnot thinks these graphs are showing what you think they are. plz upload yer .xls spreadsheets and link me to them plz

no the axes are correct, though plz explain in more detail how and what a gaussian-looking function proves

Heh, made this spreadsheet cause I found a function in tasclient to get all the map data easily:
http://beherith.eat-peet.net/stuff/mapi ... ection.xls

Beherith is wrong when saying "x = # maps at value y"

Look at the x-axis; it is actually "number of maps at or below that y-value". So f.ex on first plot, there are 487 maps at or below 20 tidal. Longer "plateau" means "more maps at that level", so f.ex there are 487 - 271 = 216 maps at 20 tidal.

What these graphs show is that the analyzed properties of maps are roughly "normally distributed", which means it has lots in middle, and much fewer on edges; i bet it would look something like this if it was a histogram or something.

Dragon45 wrote:Beherith is wrong when saying "x = # maps at value y"

I never said that.
The way this this is plotted is that each map is a data point, from 1 to ~650. The data points are sorted then plotted where X is the map serial number.

Wind distribution is hardly gaussian:

It is, as far as I can tell. Min-wind just can't get below 0, that's why it doesn't produce sucha nice distribution.

min/max winds maybe not by themselves. But look at average wind.

Beherith wrote:Heh, made this spreadsheet cause I found a function in tasclient to get all the map data easily:

http://beherith.eat-peet.net/stuff/mapi ... ection.xls

Looking at tidal data it would make a lot more sense if there was some way to discard those maps that don't have water that is immediately accessible from the start of the game, or at least set them to 0; f.ex SmallDivide and DSD. Looks like there's lots of maps that have tidal defined but don't actually have accessible water; they're cluttering up the data...

Yeah I noticed that one too XD