AI Homeworld Defense

[chemical_art]

What if instead of global tachyon coverage, there was just a set of patrol routes and a constant spawn of AI decloakers? Avoid them, and you're fine and maybe the cloaking cheese won't be nearly as silly while you're rewarded for observation and clever movements by still being able to cloak.

If it was not random, a pattern could be found, and the defense is fairly useless.
If it was not random, but the area of decloaking is so large that you couldn't sneak by it by watching it, it is in effect global.
If it was random, one could save scum to get around it.
If save scumming doesn't work because the decloaked area is too large, it is in effect global.

[chemical_art]

I'm only joining the thread because somebody mentioned CSGs.


I'm watching you.
 

I'm still myself trying to digest the CSG reference. I suggested core worlds could be taken as a method of removing a brutal post in a siege like fashion. CSG's are not a method, they are THE method of allowing them be smashed like any other post.


As for taking core worlds being usual, I have never in my lifetime of experience found it useful to take core worlds. Neuter it, yes, but taking it and putting the AI HW on permanent alert and risk the consequences of CPA and Raid engines forever, while permanently removing the potential to bleed strategic reserve, vastly is worst then the pity reinforcements the core worlds get between attacks (and the strategic reserve you either fight there, or on the AI HW itself)

[Draco18s]

What about instead of making the guard posts all spawn on map-gen, make then sequential.

I think Keith/X want to stay away from stuff like this.

If it was not random, a pattern could be found, and the defense is fairly useless.
If it was not random, but the area of decloaking is so large that you couldn't sneak by it by watching it, it is in effect global.
If it was random, one could save scum to get around it.
If save scumming doesn't work because the decloaked area is too large, it is in effect global.

So, take the third option:
Randomized based on map seed.

You can learn it, but that only helps you for that game.

[LordSloth]

Huh, the CSG reference was perfectly clear to me.

To simplify, perhaps throwing out some key details while doing so:
The idea was that instead of spawning something like a Core Guard Post on just the homeworld, that -one- CGP spawns on multiple core worlds. If you manage to take out that CGP on a core world, it goes 'pop' on all the other worlds.

While the probability of having an overlapping field of fire death trap goes up, so to does the idea of having a place where they don't overlap. You could still go straight for the homeworld, but if one post proves impossible, then you could hit a core world and destroy it's counterpart and they all (for that AI at least) go up in smoke.

It's not my idea, so I'll let LT elaborate if necessary, but does that clear things up, chemical art?

It's not the concept of provided invulnerability the suggestion revolves around, but the 'take one down, they all go down' trait of the lower CSGS that it revolves around.

I've got a few thoughts on potential 'too easy' and potential counters to 'too easy', but I'm dead tired. If the idea gets discussed some more I'll elaborate.

I could imagine one brutal pick spawning only on the homeworld, a second brutal pick spawning on the HW and a third of the Core Worlds, a Core Guard Post (and clones) spawning only on Core Worlds, I could imagine the whole defense structure collapsing to easiness without planetary Tachyon, I could see a grab-bag approach actually allowing for an additonal brutal pick or two thanks to multiple avenues of taking the thing down. I could see Champions running away with things,  I could see it breaking the 'strength check' we so desperately introduced to counter all those 10/10ers, I could see it making a nightmare out of a snake map or even a less linear one that has odds of spawning something in the center somehow. I could even imagine a Fallen Spire Shard Retrieval, Dire Guardian Lair, Spire Archive, or awkwardly placed Co-Processor being made exceptionally difficult.

I like the idea for being interesting to think about... I'm too tired to actually draw conclusions however.

[Cyborg]

As interesting ideas go, I don't mind the discussion. I'm content to read along and crack a smile at a lot of this stuff. However, it's going to be a real tough pill to swallow if we are forced to incur AIP/take any worlds. I think that is probably the only settled discussion there is on this forum. We went to rails once, and it started raining dingle berries around here. I can't see that happening again.

But by all means, talk it out. I think the underlying complaint is real even if the CSG look-alike idea stinks.

[TechSY730]

Wait, how did this turn into ADDING an additional nasty thing to the Homeworlds?  Right now, the Superfortress isn't a pick.  Period.  Only the Trains or a specific AI type (Fortress Baron, IIRC) can cause it to appear, and there is wide agreement that Trains need fixing.  Why bring the SuperFortress into this at all?

It doesn't add them inherently, it puts them into the category of brutal post in terms of spawning. It means in non niche games you could see it more often, yes. However if it does it is in place and not in addition to a brutal post.  So it takes the place of for example a wrath post, rather then the potential of next to it. It also prevents the above situation of it appearing next to a poly-crystal crashing post.

So if anything, it brings super forts more streamlined: You see it in more situations, but prevent the situations where it wrecks everything.

Basically this. Basically, it makes the AI "pay" to use a superfortress on the homeworld (and stops them from showing up anywhere else short of Zenith traders); as it would be a brutal pick, the AI would have to give up something equally brutal in opportunity "cost".
Two, as a part of my (linked) suggestion, superforts would be removed from the list of things an astro train construction project could build.
And three, the AI as a whole (all difficulties) would be getting one brutal pick less; just a straight up, no strings attached nerf. (If this makes homeworlds too easy, then the non-brutal pick stuff could be buffed, but maybe not for the first "attempt" of this type of idea)

[Diazo]

Okay, trying to think on this from a higher level and limiting myself to making smaller changes to make it relatively easy to implement.

First: The AI HWs are essentially gear checks that you have to destroy. They are the only "must-take" systems in the game. That breaks a lot of AI War's charm in that every other system in the game is optional and you can go around them. However, that's the game we have and so let's try and make the options when you get there more varied.

I keep coming back to the strategic reserve as the way to do this.

The guard posts are immobile fortresses, they can't move and so there's no real strategic options in dealing for them, so let's make them steps in taking the system.

The way I am imagining that boils down to this:

Remove the brutal picks, changing the options to "support" and "combat" picks. The "combat" picks are all the direct damage guard posts (most of them) and they get rebalanced so that they have good health and decent firepower. (This would be a massive nerf to stuff like the Grav Lance.)

The "support" picks are then stuff like the Eyes, the CPA and Raid posts, etc.

An unsupported guard post should be easy meat against the players end-game fleet, but have enough health it stays alive for a minute or two. The DPS of a "combat" guard post should be enough that it takes a chunk out of a players fleet, but not be a serious threat. DPS of a "support" guard post does not really matter.

However, the health increase allows the strategic reserve to spawn and actually be a factor in the fight.

Do you ignore the strategic reserve and focus the guard post? Kill the strategic reserve and allow the guard post to stay alive longer? Kill the guard post over multiple raids, running away when the strategic reserve spawns?

Probably to make sure the AI does not run out of ships, you would have to do an extra reinforcement pulse on that AI HW only whenever a core guard post dies.

This keeps things pretty close to how they are now, but it removes the two biggest problems (in my opinion) with the current setup. Those being a) the DPS spread between the weakest and strongest "combat" guard posts is too big at the moment, on the order of magnitudes to big. The other being that while it is still tougher when guard posts spawn close enough together to overlap, they no longer have the potential to be insta-death to your entire fleet because even combined they can't solo your fleet.


Note that part of my reasoning for this is that I thing (have not checked the patch notes for the timing of when things go added) is that the problem of the AI HW's being too easy has been solved too much. We've got guard posts that can solo the players entire end-game fleet, we've got the strategic reserve to support the rest of the AI's forces in it's HW and we've got the changes to how the AIP costs for destroying the AI's HW are applied.

The guard posts soloing a players fleet and the strategic reserve have both gone in a the level of "pretty powerful" and could be argued they both went in powerful enough that they solve the problem on their own. Add in the other mechanic and you are now getting a difficulty beyond that which was expected.

I'll actually try to trawl the patch notes tomorrow to see the timeline on this, but I'm already late for bed tonight and it will have to wait.

D.

[LordSloth]

As interesting ideas go, I don't mind the discussion. I'm content to read along and crack a smile at a lot of this stuff. However, it's going to be a real tough pill to swallow if we are forced to incur AIP/take any worlds. I think that is probably the only settled discussion there is on this forum. We went to rails once, and it started raining dingle berries around here. I can't see that happening again.

But by all means, talk it out. I think the underlying complaint is real even if the CSG look-alike idea stinks.

Yeah, but it wasn't the invulnerability bit that was the focus, as far as I could tell, it didn't actually come up. At no point does LT mention needing to take core systems in order to pop the guard posts/picks. It was all about the 'pop one, pop them all (of that rank)' feature. You could approach the homeworld directly as always, or you could stage a raid against a core world then pull back without taking out a core command.

Core Guard Posts of one set would all go down if you popped just one of the clones.
Brutal Picks would go down if you took down either the Brutal installation on the HW itself or all of the Prime Posts in the core ring.  Unlike the Core GP clones, the Prime Posts would be  different structures (possibly weaker individually). I don't believe LT expected you to pop 3-8 more planets worth of AIP, from my co-op experience with LT.

On a side note... I want to be able to nuke the Zenith Traitor and take 'I'm out before he sells too many Super Forts.

[TechSY730]

On a side note... I want to be able to nuke the Zenith Traitor and take 'I'm out before he sells too many Super Forts.

Lol.

Though that does raise an interesting question?

Keith, what is the current probability that the AI will buy something when the trader leaves an AI system?
When the AI does buy something, what is the probability distribution of the various things the AI can buy?

[LaughingThesaurus]

Yeah, the idea of the tachyon guards is to make you at least keep on your toes and look for a way to get around the decloaking rather than just parading through the world unstopped with cloaked ships.

[Draco18s]

Keith, what is the current probability that the AI will buy something when the trader leaves an AI system?
When the AI does buy something, what is the probability distribution of the various things the AI can buy?

And what opportunity cost does it cost the AI to buy it?

[KDR_11k]

Instead of spawning them sequentially I'd say only a few of the HW's defenses should be active simultaneously, the rest is powered down and invincible. The number of simultaneously active defenses should depend on the difficulty or AIP. Kill one and another defense powers up. If the AI still manages an impossible combo perhaps detonating some warhead (EMP?) would re-roll the activation so you could try to change the combo at the cost of some AIP.

Perhaps this would make the AI's defenses too sparse, in that case the non-brutal CGPs should be converted to praetorian guards (exact same abilities) and be able to move.

[Draco18s]

Instead of spawning them sequentially I'd say only a few of the HW's defenses should be active simultaneously, the rest is powered down and invincible. The number of simultaneously active defenses should depend on the difficulty or AIP. Kill one and another defense powers up. If the AI still manages an impossible combo perhaps detonating some warhead (EMP?) would re-roll the activation so you could try to change the combo at the cost of some AIP.

Perhaps this would make the AI's defenses too sparse, in that case the non-brutal CGPs should be converted to praetorian guards (exact same abilities) and be able to move.

Now that's a thought.

[TechSY730]

Keith, what is the current probability that the AI will buy something when the trader leaves an AI system?
When the AI does buy something, what is the probability distribution of the various things the AI can buy?

And what opportunity cost does it cost the AI to buy it?

For buying something vs. not buying anything, there is no opportunity cost for the AI. The only thing that keeps this in check is the very low chance they will buy something for any one visit.

For buying item X instead of item Y, there most certainly is an opportunity cost. Unlike us, the AI can only buy at most one thing per trader visit. Also, they are not allowed to "stack" things (have multiple of thing N) except for a very limited set of exceptions (like core shield generators), and I think even in those cases there is a cap for how many the AI can have on a single planet.

[Draco18s]

And what opportunity cost does it cost the AI to buy it?

For buying something vs. not buying anything, there is no opportunity cost for the AI. The only thing that keeps this in check is the very low chance they will buy something for any one visit.

Question is:
Should there be?