Killing killboxes.

[RicRider]

The real solution is to make the AI actually have a goal other than "fuck over the player in any way possible."

If they lose all 25 of their guys and one of our guys loses a finger, they win. Any "fixing" of combat that doesn't address that fails to address combat.

By goal do you mean purpose or victory condition? I bet you half the players, including me, don't know what winning is in RimWorld either. Good luck trying to define it for raiders.

Joking aside, there's mods out there that fix raider AI slightly but you and everyone else out there including really smart people all understand that we're not talking about how people would react - but sadly - our limitations are how bits and bytes react. I think is where RimWorld breaks down as a colony simulator (because they all do, they can't fix the lack of AI having a singularity - not that it ever will but that's a tangent).

No guys, the only thing that will fix the stupid aspect of RimWorld is when the game has a modded or official working player versus player mode where the goal is to wipe out the other guy's colony.

Isn't this like the oldest debate in computer science too by the way? You guys looking for the holy grail of AI fighting people are looking for something that's not real and never will be real. The real action is when players fight.

[lugaruclone]

Also keep in mind huge raider counts are because of massive accumulated wealth... I just finished the game (year 6, hosted the emperor) and I was seeing raids of like 20 (from various directions) for my endgame fights.

The 4 imperial guardsmen and my various shooters where enough, with help from mortars and land mines placed 'along paths' that sometimes they hit, usually they do not. I also had turrets. They... distract enemies sometimes.

I feel like the killbox would only be needed had I mined out my entire maps resources or if they ever ad some more endgame foes like you see in mods (power armored pirates, massive mechanoids).

[Pangaea]

This is a game from 1.0, with a killbox that some detest. After dying many times without one, I ended up concluding it was a necessity. At least for how I prefer to play the games.

And honestly, taking on something like this in the open (and remember this is a fairly normal-sized raid/ship, it's not a one-off, and same with the 200+ tribals).

We still suffered heavy damages, one person died, and from what I recall 7-8 others were incapacitated and I had a couple of people trying to frantically get them into hospital and fix them up before they bled out. It was total chaos. But because we had turret support (until everything blew up from the ~15 inferno cannons), it was possible to survive.

But for all I know, Tynan also detests stuff like this, and wanted it gone. Who knows. But I liked to be able to play long games like this, and take on colossal raids 

[Karim666]

The real solution is to make the AI actually have a goal other than "fuck over the player in any way possible."

If they lose all 25 of their guys and one of our guys loses a finger, they win. Any "fixing" of combat that doesn't address that fails to address combat.

By goal do you mean purpose or victory condition? I bet you half the players, including me, don't know what winning is in RimWorld either. Good luck trying to define it for raiders.

Joking aside, there's mods out there that fix raider AI slightly but you and everyone else out there including really smart people all understand that we're not talking about how people would react - but sadly - our limitations are how bits and bytes react. I think is where RimWorld breaks down as a colony simulator (because they all do, they can't fix the lack of AI having a singularity - not that it ever will but that's a tangent).

No guys, the only thing that will fix the stupid aspect of RimWorld is when the game has a modded or official working player versus player mode where the goal is to wipe out the other guy's colony.

Isn't this like the oldest debate in computer science too by the way? You guys looking for the holy grail of AI fighting people are looking for something that's not real and never will be real. The real action is when players fight.

Hmmm. holy grail isn't, however the foundation of AI in this game for combat is good but rather weak. consider this: balancing is a purely statistical thing based experiments and results as in you would monitor the stats and periodically automatically adjusts the balance. the goal of a rimworld raid is easy to define because a raid is only one battle in a war...

i'm planning on working on a new AI for combat and would appreciate some help due to the fact i'm kinda rusty when it come to moding after 4 years of python only data science research.

tbh the AI should be able to build and maintain an attack base in every raid no only for sieges.

so no it's hard but not that hard.

[RicRider]

Hmmm. holy grail isn't, however the foundation of AI in this game for combat is good but rather weak. consider this: balancing is a purely statistical thing based experiments and results as in you would monitor the stats and periodically automatically adjusts the balance. the goal of a rimworld raid is easy to define because a raid is only one battle in a war...

I don't really understand your first paragraph. I think what you're saying is that game balance is achieved by looking at stats (from players I assume) and adjusting balance accordingly, but while that sounds good on paper, what does that actually mean? Can you explain in another way?

i'm planning on working on a new AI for combat and would appreciate some help due to the fact i'm kinda rusty when it come to moding after 4 years of python only data science research.

tbh the AI should be able to build and maintain an attack base in every raid no only for sieges.

so no it's hard but not that hard.

I definitely don't know anything about coding so can't help you there, but what I can tell you is that if your goal is to make an AI that can build and maintain sieges then that's a pretty good goal and I'd download that mod. What makes people like me download mods is partly how realistic they are and also how good they are at making me feel like I'm not interacting with a computer, but in a way interacting with my own intelligence... if that makes sense without sounding too pretentious? My point is playing RimWorld single player is a bit like playing chess against a computer. We kind of expect to lose but not because the machine is expected to be smart since it's just a dumb toaster but because we've artificially raised the difficulty through hit points, endurance, injuries, other mechanics, whatever. We're not even trying to make the AI smarter, just create the illusion that it is.

[fritzgryphon]

i'm planning on working on a new AI for combat and would appreciate some help due to the fact i'm kinda rusty when it come to moding after 4 years of python only data science research.

This.  You are doing God's work.

Killboxes work, and are necessary to use on high difficulty or ballooned wealth, because the AI takes no precautions against them.  Mechs always fall for it.  Sapper raids are a small improvement, but even then you can use troops to lure part or all of the raiders to the killbox. 

A human player, even if they couldn't visibly see deadfalls or IEDs, would know to avoid obvious restricted corridors, kill zones faced by sandbag bunkers and turret fields of fire.  They would see the wall and turret layout,  lol, then attack everywhere but the killbox.  Even with fog of war or restricted LOS (which the player doesn't have to deal with, so I don't see why the raiders would), a human player could still guess where the kill box is as they advance and explore.

And if the core of a colony is too well defended in general, a human raider would just burn undefended crops and structures, or plink at vulnerable exterior features from a distance.  No reason to go zerg rush right away. 

Then there's the option to simply withdraw if the colony is not vulnerable enough to attack.  Maybe come back when the colony catches plague.

If the AI could plan attacks like a player, killboxes would be dead.

[5thHorseman]

The real solution is to make the AI actually have a goal other than "fuck over the player in any way possible."

If they lose all 25 of their guys and one of our guys loses a finger, they win. Any "fixing" of combat that doesn't address that fails to address combat.

By goal do you mean purpose or victory condition?

Of the fight, yes.

If you drop podded in to an enemy base with the intent to ransack it, and saw that it had more mortars than you had people, and they could easily man those mortars and defend their walls, you'd probably run. At least, I would. Even if I saw that I could mine through a bedroom wall in order to set the bed on fire.

The AI? Nope that bed's a burnin' folks, I don't care that your sister just got shredded by a shell get in there you maggot.

THAT is what makes kill boxes not just useful, but necessary.

As to how to do it? I don't know. I don't care. It's not my game and I'm not even a programmer. But until the game continues to send wave after wave of tribals to die just so they can hopefully cause an infectious scratch on one of my colonists (which will then of course cause havoc when the doctor decides it's more important to play horeshoes instead of treat them, or go on an 18 day food binge when I try to force the treatment) I'll be trying to find ways to siphon them through a zone of assured death.

[FTR]

I never understood what's so bad in killboxes. To me they are necessary evil (though can you really call it evil if players like them?), especially for late game which doesn't scale very well with your wealth.

I mean, in late game an usual pawn will face impossible odds, so the only way is to use some form of prepared defenses.

On top of all that building and maintaining killbox takes resources, time and serious planning. Even in 1.0 my killboxes were often overwhelmed by enemies (perhaps I suck at building them) and all those nerfs just feel really unnecessary.

Unless tynan wants rimworld to be project zomboid type of game, in which you only play to eventually die and you only play to discover how will it happen.

[ShadowKatt]

I never understood what's so bad in killboxes. To me they are necessary evil (though can you really call it evil if players like them?), especially for late game which doesn't scale very well with your wealth.

I mean, in late game an usual pawn will face impossible odds, so the only way is to use some form of prepared defenses.

On top of all that building and maintaining killbox takes resources, time and serious planning. Even in 1.0 my killboxes were often overwhelmed by enemies (perhaps I suck at building them) and all those nerfs just feel really unnecessary.

Unless tynan wants rimworld to be project zomboid type of game, in which you only play to eventually die and you only play to discover how will it happen.

To be fair, it's been stated that it is a game that if you keep going you will lose. The entire game is built around land, build, research, build ship, and leave. ideally all in about three years or so. Going beyond that is a bit like going to the far lands in minecraft: It works, but the numbers get wonky. I usually don't even build a ship, I like to play colonys as long as I can, and it does become pretty obvious that beyond a certain wealth, beyond a certain age, beyond a certain population, the game has no idea what to do. it just continues scaling the only way it knows how which eventually results in the entire world landing in your base to wipe you out. There are mods you can use to help mitigate that, but that's the foundation of the game right there.

[FTR]

I usually don't even build a ship, I like to play colonys as long as I can, and it does become pretty obvious that beyond a certain wealth, beyond a certain age, beyond a certain population, the game has no idea what to do. it just continues scaling the only way it knows how which eventually results in the entire world landing in your base to wipe you out. There are mods you can use to help mitigate that, but that's the foundation of the game right there.

That's exactly what I am thinking, I couldn't describe it better myself. I am playing in the similar fashion. I DO eventually leave, but before I do I really want to enjoy my colony and beat every possible challenge and maybe later on, eventually leave. I hate rushing my games and I feel like unfortunately rimworld puts you on the timer to leave the planet or bad late game scaling will ruin you.

I wish there would be some endless mode or pseudo-endless mode in which main goal isn't to leave but to conquer the planet or something.

[Whifflepits]

Not even sure OP and I are playing the same game.

Every time I try to make killboxes the AI seems to pick up on it and send nothing but sappers and sieges my way. I eventually end up with so many holes hastily half-patched in my walls that I give up and just open everything up and start fighting alongside the turrets. Putting colonists behind solid cover in carefully picked places and having an open strategy

If anything I find it especially annoying that the AI doesn't have to deal with fog of war over any hidden areas. I can't recall how many times I thought I was being super-safe digging deep into a mountain only to have the AI drop in the valley next door and after digging 2 or 3 blocks open up a huge network of tunnels that leads them right next to my storage area.

It's always the damn storage area.....

Or mechanoids hot-dropping directly into my kitchen...