Experimental Biter expansion too strong

[worph]

From looking at the FFF post it makes sense that the best candidate chunk that gets chosen is the same every time, if nothing else changed.
I didn't test this myself.

Adding a bit of randomness certainly seems like a good idea if it holds true.

[Klonan]

Adding a bit of randomness certainly seems like a good idea if it holds true.

If there are many candidates, the game will pick randomly from the top 10%

[worph]

grafik.png (211.69 KiB) Viewed 1312 times

Even in the example given in the FFF, there is an area with a single candidate.
This is the edgecase I was talking about.
The person 3 replies up seemingly reported a similar behaviour close to his base.
All I'm saying is that this edgecase might not be too rare.

Maybe in those cases it could randomize between the candidate chunk and its adjacent 4 (or 8 with diagonals), so long as there is no pollution cloud nor an actual existing nest in any of them.

The reason being that when you destroy just that nest (for example because you have some drills or oilpumps nearby) it will come back in exactly the same location every time. It's not a huge deal, but I can see how it could break immersion to some degree, if biters kept building in exactly the same spot over and over again, no matter how often you remove them. A tiny bit of randomness could make that behaviour feel less algorithmic and more arbitrary / "natural".

Again, I didn't test this.

[Fletch]

... The reason being that when you destroy just that nest (for example because you have some drills or oilpumps nearby) it will come back in exactly the same location every time. It's not a huge deal, but I can see how it could break immersion to some degree, if biters kept building in exactly the same spot over and over again, no matter how often you remove them. A tiny bit of randomness could make that behaviour feel less algorithmic and more arbitrary / "natural".

Same phenomena seen in v2.0.x too. In my current playthrough (2.0.77), biters expanded to a particular spot within the pollution cloud -- I ride out there with a tank and destroy them. Sometime later my base is under attack again, from newly formed base in the exact same spot. Rinse/repeat this ~5 times (same spot every time). I even ran big power poles out there and set up a radar dish, and they expanded on top of that

I finally got around to building a bunker at a choke point to kill all the future expansion parties before they could reach their "favorite spot" again.

Perhaps an idea to add some randomness, is to make expansion coordinates have some time-limited anti-affinity so players don't get into this loop where expansions happen in the same spot repeatedly (game would need to track the last XX expansion coordinates and not pick the same place until that spot ages out of the history list). Its not a big deal though, and its taken a couple of years for me to even notice it.

[coffee-factorio]

I put in some more time. I was wondering if they weren't spawning out in the wilderness sometimes. Saw some larger nests from 2.1.7. They were not necessarily expansion directed at me.

It also felt like if I had them engaged, they'd choose to attack or at the point where pollution was intense. So I had one block off to the west in my check game where the biters impaled themselves, then I'll have to drive down south but the radar wasn't seeing much to be scared of from some open areas there and to the east.

It does feel like behavior of an animal rather than an aggressor. The biters never struck me as something with malice, so it plays towards that. Up to others whether that is as intended, but it's like how I'll occasionally see a skunk strolling around my neighborhood and we just go our separate ways.

[coffee-factorio]

But as a general "thing" and whether or not the biters normally move around randomly, there's an area to the north they like to settle on my map. And it is a thing where I can go out, work for 2 hours, and then I'll be really, really glad I drove up to that spot because they will make a run on my base from 2 directions if I don't manage that area. I'm going to gain +3 hours on this play through I think so skill progression/impact of strategy and motivation is a factor. You do have to be a lot more aggressive, but a strategy of running turret lines with good steel works. Spawning size at the level I'm at, pre-launch blue science, feels satisfying. I'm just getting big boys out at that point.

Marker points to give a sense of the radius of engagement.

Screenshot 2026-08-29 172721.png (301.34 KiB) Viewed 873 times

The .750 kilometer engagement zone feels about average for what I do.

[coffee-factorio]

Yeah, treating that game as a project and was both curious about behavior and wanted to work on some things. Some things that I can confirm in the wild: it's definitely random. But random is random, not easy: so in that north area I had two respawns. To the east where there's a lot of pollution, they weren't able to overwhelm a 2 gun cluster but the biters did run it dry of ammo. But they did try that area multiple (more than five) times in order to pull that off and I was doing mitigation by having bots porter piercing rounds in.

Got nuclear reactor up in a way I'm ultimately happy with but it's going to be several thousand steel before I have a depth of parts that I want on Nauvis. Not feeling like I need more than defenders and a tank. Judging by the size, if I missed my timing and let this go for 4-8 hours I would have bugs on my stuff. But old defense patterns holding up for now. Need to push it later to see if the biters can set up a base of fire on laser turrets by spawning out of range of them.

[coffee-factorio]

An interesting one from version 2.1.17:

I had a map that had previously held under a "stealth" build.
So, now I am cleaning the map of pentapod nests aggressively.

Pentapod spawn close up. They hit the edge of an invisible boundary. They happily consider a turret complex appropriate spawning. Feels correct.

Screenshot 2026-09-05 192529.png (203.47 KiB) Viewed 367 times

So this is a gun-turret line with a rocket turret. The expansion candidates are close...
But in this picture, there's no green circles in the jellynut swamp beyond the turret.

From a distance, we see that pentapods do not spawn beyond the turret cluster.

Screenshot 2026-09-05 192541.png (366.63 KiB) Viewed 367 times

On a whim I checked Nauvis.

Nauvis isolated rocket complex. Note the spawning locations bypass the turrets. This is desired.

Screenshot 2026-09-05 192602.png (61.52 KiB) Viewed 367 times

This gunline has candidates behind it and displays something of the "expected behavior". It is expected that biters will try and spawn near a line to melt it with acid, using worms. It's actually one of the things that drives progression in mods, in Space Exploration if you weren't on top of it and didn't get artillery up by going to space you were going to have a bad week. I cannot recall in 3 hours hearing explosion one.

Conventional laser line. Biters don't want to bypass the line. Haven't heard alarms indicating they are trying to break it.

Screenshot 2026-09-05 192629.png (79.29 KiB) Viewed 367 times

This is a more conventional line designed so that biters try and pass through, rather than spawn outside it. All spawn circles are in biter territory, nothing is attempting to bypass these lines.

From a distance it's clear the biters don't want to bypass the turrets.

Screenshot 2026-09-05 192639.png (311.88 KiB) Viewed 367 times

In terms of "good, bad, in between". It's in between verging on good. I built artillery ammo into a platform and got a number of upgrades for artillery over the day. I now have the "destroy a building with artillery achievement". Previously, the inclusion of tungsten beams into the ingredients list and the relative power of spidertrons made that something that wasn't worth pursuing. It did make a difference.

[worph]

Im experiencing biters on Gleba expanding into the exact same chunk over and over again now and it really feels as algorithmic and unnatural as I expected it to, a few replies ago. If not worse.
If there is only a single chunk available something really doesn't make sense. Why would some random invertebrate species go to the same exact spot over and over again, when there are hundrets of artillery craters and exploded munitions scattered around, when all they really try to do is find a suitable place to live.

Maybe track artillery strikes on nests and include this metric when considering viable chunks. So enemies kind of deprioritize places that have been hit by artillery very frequently.

[Alfonse215]

If there is only a single chunk available something really doesn't make sense. Why would some random invertebrate species go to the same exact spot over and over again, when there are hundrets of artillery craters and exploded munitions scattered around, when all they really try to do is find a suitable place to live.

Because they're not especially smart and they go where the terrain is suitable. They don't want to move near existing nests, and there's marshland over there, and that's the only spot within their walking range, so that's where they go.

Biters have more options for expansion due to having fewer terrain considerations. Pentapods have to be pickier.

[worph]

If there is only a single chunk available something really doesn't make sense. Why would some random invertebrate species go to the same exact spot over and over again, when there are hundrets of artillery craters and exploded munitions scattered around, when all they really try to do is find a suitable place to live.

Because they're not especially smart and they go where the terrain is suitable. They don't want to move near existing nests, and there's marshland over there, and that's the only spot within their walking range, so that's where they go.

Biters have more options for expansion due to having fewer terrain considerations. Pentapods have to be pickier.

As far as I understand, they use literally the same chunk evaluation that was updated a few weeks ago and had it's own FFF blog entry.

[Alfonse215]

If there is only a single chunk available something really doesn't make sense. Why would some random invertebrate species go to the same exact spot over and over again, when there are hundrets of artillery craters and exploded munitions scattered around, when all they really try to do is find a suitable place to live.

Because they're not especially smart and they go where the terrain is suitable. They don't want to move near existing nests, and there's marshland over there, and that's the only spot within their walking range, so that's where they go.

Biters have more options for expansion due to having fewer terrain considerations. Pentapods have to be pickier.

As far as I understand, they use literally the same chunk evaluation that was updated a few weeks ago and had it's own FFF blog entry.

Yes, but they have the limitation that the terrain has to be suitable. Biters can expand to any land tiles, but pentapods can only expect to certain kinds of land tiles. Specifically, marshy land tiles. This inherently limits the number of chunks that are even potentially viable for expansion targets. As such, they will inherently have fewer such targets available compared to biters.

That is, it's much easier for pentapods to be in a situation where there are only a couple of valid chunks that they even could expand to. And that's assuming the player doesn't abuse that and start landfilling marshes to block their expansion.

[worph]

Yes, but they have the limitation that the terrain has to be suitable. Biters can expand to any land tiles, but pentapods can only expect to certain kinds of land tiles. Specifically, marshy land tiles. This inherently limits the number of chunks that are even potentially viable for expansion targets. As such, they will inherently have fewer such targets available compared to biters.

That is, it's much easier for pentapods to be in a situation where there are only a couple of valid chunks that they even could expand to. And that's assuming the player doesn't abuse that and start landfilling marshes to block their expansion.

Please read the previous discussions in this thread about chunk evaluation. This has nothing to with what you are claiming. You are correct but none of the things you talk about apply in this situation.

The entire land around my farm is suitable. Yet they still keep building in exactly that marked spot over and over again, dozens and hundrets of times. But we know from other examples that the chunk evaluation algorithm can evaluate the chunks in such a way (also on Nauvis) where there is only a single viable chunk. And I am saying that this is inherently bad.

grafik.png (95.24 KiB) Viewed 59 times

Fix

If there is only one disconnected chunk some distance away from the next viable one do any of the following:

• ignore it entirely
• include some surrounding chunks and select one at random as long as it can be considered somewhat valid (i.e. accept shorter distance to other nests)

Or also evaluate automatic artillery strikes on past nests and discourage rebuilding there in the future.

[Alfonse215]

The entire land around my farm is suitable. Yet they still keep building in exactly that marked spot over and over again, dozens and hundrets of times. But we know from other examples that the chunk evaluation algorithm can evaluate the chunks in such a way (also on Nauvis) where there is only a single viable chunk. And I am saying that this is inherently bad.

As long as there are hard limitations on where they can expand, there will be times where they'll keep re-expanding to a specific place because that's the only place that's viable within those limitations.

And there should be hard limitations on where they can expand to. Because once you start softening those limitations, then they can just expand into the middle of your base, or towards each other, or in other unpleasant ways. And thus, you will no longer be able to predict what's going to happen.

Remember: the purpose of enemies is to act as a break on your expansion and something to spend resources on. You are supposed to be able to manage them, and managing them means that they have to have reliable behaviors.

If I set up artillery that can fire further than 7 chunks, I should no longer have to worry about expansions hitting my base from that direction. I just need to protect and feed the artillery. If I can do that consistently, the problem should be solved.

I do not see what's wrong with this.