[Alpha 11] Sappers. Challenge accepted.

[Tynan]

ok good. that works. is there a hard limit on how much rock?

No. They don't really 'know' what they're pathing through. They just optimize their path by taking into account the 'cost' of each cell. Being in a turret field of fire is a cost, as is passing through a solid blocker. It really comes down to the ratio of these things in the specific circumstances.

I'll be working on balancing it.

[RickyMartini]

Im pretty sure the sapper is going to calculate to destroy the most optimal wall possible, as long as he can avoid turrets. Meaning he will destroy any single wall (or double wall) possible before actually going to dig way back into your colony.

In your example I'd say the zapper will just target a random single wall east/west and try to throw his grenade before dying, making the attack not any different from others.

Maybe concepts like this would be more likely to be "cracked" by zappers, since you can clearly see which areas are dangerous to step through and which walls offer the least amount of work to break through. I think that may also be the concept Tynan had in mind.



So basically, I guess as long as you have normal walls covering your base, they won't take the longest route possible just to break into your freezer on purpose.

(At least I hope.)

Edit: Ninja'd by dev

[Tynan]

Well in billycop's example, they'd probably approach the left or right edge (where turret coverage is lightest), approach the closest turret and start throwing grenades at it.

In Skissor's example, they'd probably go through the right wall. If they approached from the left they might go through the left parts of the bottom or top walls.

Also I'm amused your colonists are named Max and Min :p

[RickyMartini]

Now the next question making this either scary or terrifying is, if I barricade myself, will the sapper first calmly open up the right wall so the rest of the group can get in, or will the group attack together and the sapper will attack the wall at the same pace?

[hyperpeople]

From the changelog: "Created escort duty and attendant AI behaviors. Non-sappers can now escort the sappers."

I think they just go with them, avoiding turrets or shooting them.

[RemingtonRyder]

Sappers won't get anywhere near my walls. Unless Legolas runs out of arrows.

Hopefully he's learned a thing or two since the last time he tried to stop a sapper...

Indeed. I was also making a reference to Rimworld's infinite ammunition for most weapons. But one thing bothers me about that scene from the Two Towers. Like, why is that orc not a porcupine before he even gets near the wall. Plenty of other archers and he's carrying a big old sparkler which says 'Shoot Me!' 

[Adamiks]

"May 12
Many fixes and adjustments to sapper-related code.
Started adding deadfall trap."

Ok, so we have traps. Everyone happy?

[billycop32]

rimworld alpha 11: call to arms

oh this is gonna be a good one now that I know sappers are gonna be good and not op 
thick walls EVERYWHERE! XD

[UnknownSeeker]

ok good. that works. is there a hard limit on how much rock?

No. They don't really 'know' what they're pathing through. They just optimize their path by taking into account the 'cost' of each cell. Being in a turret field of fire is a cost, as is passing through a solid blocker. It really comes down to the ratio of these things in the specific circumstances.

I'll be working on balancing it.

This seems like playing against a cheating know it all AI. If I make a base that is not exposed to the sky then the AI should not be able to path find to any point that is not visible to the outside or that has not been previously scouted by the current raiding party (and depending how fancy you want to get, if any one escaped from the previous raid from that faction then they should also know the layout or if at one point they were friends and they explored your base then (but to make that last part fair you would have to be able to limit access on doors to not allow friends in). This would then lead to some interesting choices after battles, do you send your guys after the retreating enemy know they will take back information about your defenses or do you focus on bandaging your wounds and letting your guys get back to work).

Partly because this leads to a very easy counter to exploited this behavior. By making a dummy weak point room I can (say the rest of my base is covered by 5 block thick walls and this room has a specific spot with just a 1 block thick wall where i can pre-station cover for my guys to fire at it) make a predictable path for the sappers.

And partly because how interesting would the Trojan War be if they had already known that there were men inside of the horse. How can there ever be a surprise you and the enemy both know exactly what the other guys plan for victory is.

You may say that well then people are just going to make a roof over their kill boxes but that will significantly impact the effectiveness of big kill boxes. Because now they will have to add supporting columns every where limiting fields of fire of the turrets and causing a significant chance for the roof to collapse on top of the defenses. If they do build build enough columns that they don't have to worry about collapse then they will limit the fields of fire even further.

[billycop32]

unknownseeker, I like your point, but that's adding a very expensive point(in terms of time for Ty) so I think for the moment i'd rather let the raiders have omni view so ty can focus on better stuff like more endings.

[Tynan]

Actually I'm probably going to give them a faction-level knowledge model for traps.

But for turrets? It's a bit odd. The turrets are clearly visible. So they can just peek around the corner, see the turrets, tell all their buddies (by radio or otherwise), and everyone can re-adjust their approach. So I don't think turrets really lend themselves to a knowledge model; I'll probably ignore it for this.

Knowledge modeling enemies is very tricky in games; it's hard to communicate to the player what the enemy knows in a smooth, intuitive, non-disruptive, non-distracting way.

[Keychan]

I would like to know if sappers are going to be a almost separate unit with bodyguard pawns.  Like if there is a medium-large attack force and the sapper splits off with his/her own little team to break in while the main force knocks on the front door.

[RickyMartini]

From the changelog: "Created escort duty and attendant AI behaviors. Non-sappers can now escort the sappers."

I think they just go with them, avoiding turrets or shooting them.

So basically the terrifying one. Shit.

[Tynan]

I would like to know if sappers are going to be a almost separate unit with bodyguard pawns.  Like if there is a medium-large attack force and the sapper splits off with his/her own little team to break in while the main force knocks on the front door.

There's no plan for a two-pronged attack like that.

Though it's certainly an interesting idea...

[TLHeart]

ok good. that works. is there a hard limit on how much rock?

No. They don't really 'know' what they're pathing through. They just optimize their path by taking into account the 'cost' of each cell. Being in a turret field of fire is a cost, as is passing through a solid blocker. It really comes down to the ratio of these things in the specific circumstances.

I'll be working on balancing it.

This seems like playing against a cheating know it all AI. If I make a base that is not exposed to the sky then the AI should not be able to path find to any point that is not visible to the outside or that has not been previously scouted by the current raiding party (and depending how fancy you want to get, if any one escaped from the previous raid from that faction then they should also know the layout or if at one point they were friends and they explored your base then (but to make that last part fair you would have to be able to limit access on doors to not allow friends in). This would then lead to some interesting choices after battles, do you send your guys after the retreating enemy know they will take back information about your defenses or do you focus on bandaging your wounds and letting your guys get back to work).

Partly because this leads to a very easy counter to exploited this behavior. By making a dummy weak point room I can (say the rest of my base is covered by 5 block thick walls and this room has a specific spot with just a 1 block thick wall where i can pre-station cover for my guys to fire at it) make a predictable path for the sappers.

And partly because how interesting would the Trojan War be if they had already known that there were men inside of the horse. How can there ever be a surprise you and the enemy both know exactly what the other guys plan for victory is.

You may say that well then people are just going to make a roof over their kill boxes but that will significantly impact the effectiveness of big kill boxes. Because now they will have to add supporting columns every where limiting fields of fire of the turrets and causing a significant chance for the roof to collapse on top of the defenses. If they do build build enough columns that they don't have to worry about collapse then they will limit the fields of fire even further.

The AI does not need to cheat to implement an avoid death trap kill box.

See turrets, change path, to wall along side, destroy first block, re evaluate, did it open a path, or is there more blocks to be destroyed?  evaluate, again.

We have total vision of who is attacking, what they are carrying for weapons, and get to see their line of approach. Are we not the ones cheating?