[0.1x.x] [Harkonnen] Huge bias in biter north expansion

[MagmaMcFry]

Biters seem to build new nests (not the worldgen ones, Klonan!) almost exclusively in the northern hemisphere.

This bug may have been in the game for a very long time already, I think this has happened in all the worlds I've played, even in 0.12. After having this happen to ridiculous extents in my current world, I decided to experiment with it to see if this was actually a thing or whether I was just going insane (because I've found no other mention of this anywhere on the Internet).

Steps to reproduce and log:

- Get yourself the attached scenario
- Disable autosave, because we're gonna go fast
- Launch the scenario, using the default worldgen preset, with the following settings modified:
-- Set peaceful mode to yes (Edit: no), biter frequency to low and starting area to large (to give the biters room to expand)
-- In the advanced settings tab, set biter expansion groupsize and cooldown as low as possible
- Hit "Generate", then idle for a while as the scenario does all the work for you. if you like you can look at the map with the show_enemy_expansion_chunk_candidates debug setting enabled
- Look in the script-output folder for a log of the results

Example output:

Code: Select all

Spawner built in NE quadrant (553, -942)
Spawner built in NE quadrant (439, -878)
Spawner built in NW quadrant (-245, -853)
Spawner built in NE quadrant (40, -909)
Spawner built in NE quadrant (462, -971)
Spawner built in NE quadrant (578, -528)
Spawner built in NE quadrant (199, -781)
Spawner built in NW quadrant (-633, -187)
Spawner built in NW quadrant (-346, -552)
Spawner built in NW quadrant (-247, -582)
Spawner built in NE quadrant (973, -237)
Spawner built in NW quadrant (-557, -895)
Spawner built in NW quadrant (-430, -225)
Spawner built in NE quadrant (712, -742)
Spawner built in NE quadrant (158, -951)
Spawner built in NW quadrant (-490, -322)
Spawner built in NE quadrant (955, -804)
Spawner built in NE quadrant (737, -645)
Spawner built in NE quadrant (926, -795)
Spawner built in NW quadrant (-848, -591)
Spawner built in NW quadrant (-244, -859)
Spawner built in NW quadrant (-587, -938)
Spawner built in NW quadrant (-872, -640)
Spawner built in NE quadrant (745, -964)
Spawner built in NW quadrant (-470, -809)
Spawner built in NE quadrant (472, -758)
Spawner built in NW quadrant (-37, -593)
Spawner built in NW quadrant (-429, -571)
Spawner built in SW quadrant (-971, 365)
Spawner built in NW quadrant (-276, -878)
Spawner built in NW quadrant (-257, -748)
Spawner built in NW quadrant (-242, -487)
Spawner built in NW quadrant (-493, -632)
Spawner built in NE quadrant (731, -502)
Spawner built in NW quadrant (-161, -574)
Spawner built in NW quadrant (-144, -829)
Spawner built in NE quadrant (318, -643)
Spawner built in NW quadrant (-606, -854)
Spawner built in NW quadrant (-438, -622)
Spawner built in NW quadrant (-355, -470)
Spawner built in NW quadrant (-986, -386)
Spawner built in NW quadrant (-624, -328)
Spawner built in NW quadrant (-636, -81)
Spawner built in NW quadrant (-143, -824)
Spawner built in NE quadrant (651, -595)

[sparr]

Thanks for being so thorough!

[Klonan]

I did a small test, and the results are surprising:



Was my methodology wrong?

[orzelek]

Tbh Klonan result of your test must be wrong
Having biters expand that evenly is.. least likely scenario unless it's implemented like that.

In general I seen some strange behaviour when testing RSO and expansion on 0.14 were biters seemed to have preference in settling in certain areas. I did that after some bug reports from people that biters were not seen expanding with RSO. I didn't have big sample of tests so didn't try to report anything but north preference is indeed possible.
RSO makes bases much less frequent so preference of biter bases for expansion in certain areas would be more visible then.

[MagmaMcFry]

Klonan, I'm not talking about worldgen, I'm talking about biter expansion, the part where biters group together and build new nests while the game is running.

[I_IBlackI_I]

So would a better test method be this?

Remove all biters from 1000,1000 500,-1000 and -1000,1000 -500, -1000.
Give the game some time for them to expand.
Count the amount of spawners in the area defined above.

[MagmaMcFry]

It seems that none of you have actually taken a look at the scenario I provided. It automatically logs every single on_biter_base_built event with its coordinates and quadrant. As you can see from the scenario log I provided, and also as you can literally watch happening if you have show_enemy_expansion_candidate_chunks enabled while you're running the scenario, the biters greatly prefer to resettle in the north.

[Harkonnen]

I think I know the reason. Just one thing - you cannot replicate that in peaceful mode because biters do not expand in it.

EDIT: Fixed for coming release. Thanks for report Sanity has been restored.

[orzelek]

I think I know the reason. Just one thing - you cannot replicate that in peaceful mode because biters do not expand in it.

EDIT: Fixed for coming release. Thanks for report Sanity has been restored.

Out of curiosity - what was causing this?

[Harkonnen]

Once in a while code selects several random chunks, and of them all it picks one with highest expansion score. If there are several chunks with same highest score, it would always pick topmost one (and of topmost ones leftmost one). The fix was to select random chunk out of all chunks with highest score.

[StoneLegion]

Damn... My friend goes south east and I went north east Now I'm starting to see why