Human Threat Meter?

[Drjones013]

I've been following the 'State of the Game' thread and was wondering if the 'fleet blobbing' problem could be fixed by the AI having a threat meter. If the AI detects high human FP in a system then it would respond by sending its SF in that direction to check the human advance without expending the resources of surrounding systems.

I could have SWORN that this was already implemented, and if so, the effect needs to be boosted.

[TechSY730]

Right now, special forces only reacts to significant human presence on "important" systems. Right now, planets that the AI consider important are those with "big" capturables (golems, advanced research stations, advanced factories, manufactories, etc), a world with Superterminal, worlds with AIP reducers (I think), core world, and homeworlds.

It would be nice if their list of "important" stuff could be expanded, like significantly large deep strike attempts or something. However, it shouldn't be too aggressive as then it would be too easy to "trick" special forces into traps or to a planet way far away from where you really want to go. (Being able to do this some is fine, it is a valid "mind game" after all, but the AI shouldn't make it stupidly easy to do so)

[LaughingThesaurus]

I kinda figured that maybe there would be some kind of a meter that goes up as a short term alertness thing. It increases when you have a military (and only military) presence on an AI planet. It goes down when you don't. Its rate of going up is indicated by the total FP of the ships you brought in, or maybe just how much your FP outnumbers the enemy's. As it fills to certain thresholds (every 25%?) the reinforcements for all alerted planets get a significant buff and the AI is much more likely to send waves. Maybe at one of the thresholds, the AI will alert special forces and send a wing of those in to push as well. If it reaches 100%, then very, very bad things happen. The AI amasses a huge counter-offensive.

[Draco18s]

There was a "too much firepower" trigger at one point.  Unfortunately it counted AI firepower in this calculation as well, so the AI would see too much (AI) firepower in a system and reinforce that system.

I'm sure you can see how this ended.

Ah, found it.  Patch 4.049:

The AI now has an anti-extreme-blobbing/bottlenecking mechanic built in very similar to the one for anti-deepstriking. For every 10,000 firepower or more on a planet, the AI home planets will spew out ships equivalent to that which is spewed out with the deep striking. If a player thus has 310,000 firepower on a single planet, that's as bad as doing 62 deepstrikes at once, for instance.

    * This applies to both AI planets and player planets. So if the players try to make an absolutely-impenetrable defense, the AI notices and gets pissed, and the same thing is true if the AI tries to make a crazy attack force.
    * This may make the fallen spire minor faction extremely difficult at the moment, thanks to the crazy-high firepower on the spire city buildings. We'll have to address that as we continue to evaluate the balance of this feature. However, for the moment the firepower of the fallen spire city buildings are all being undervalued by 2x, which should help a tremendous amount.

Patch 4.050:

The experimental anti-bottleneck/anti-giant-fleet from the last version was problematic and on shaky logical ground to begin with. It was an interesting brief experiment, and thanks to everyone for bearing with us and sharing your thoughts whether you loved it or hated it. But this is one mechanic where trying to fix it is likely to just spiral out of control, and we'd rather devote our time and attention to more promising avenues.

[LaughingThesaurus]

Any kind of threat-o-meter would be only for your own firepower. Depends on if that's feasible, or if it even really needs a solution. Like I've mentioned, blobbing is boring, so I don't do it. That's just how I play games. I like winning, I also like having fun, so I will win while having fun. But, if those experimental ways of enforcing the guerilla warfare don't work, I won't push it.

[Drjones013]

I'm actually talking about something more 'proactive': the AI has probes (or it just knows) to find out where human FP is concentrated and then attacks that area. It has to make decisions as to how to distribute its own FP to attack but the player might make that easier by having a fleet ball in that sector. Special forces shouldn't be 'special defenders' but rather the AI's tactical response to player concentrations of firepower.

We already have a system that says Warp Gate = attack and there's no way that the AI could win if it only used that feature. CPAs tend to attack random regions with some disruptive force (attacks on power and resources) but there should be a mechanic that attacks the player's offensive capabilities on the player's own planets.

Personally I fleet ball and think that it's fine-- as a history student I have to say that an AI programmed by humans would think either on a Clausewitz principle of 'destroy the opposition's ability to resist (ie army)' or OODA (observe, orient, direct, act). I would certainly want to destroy the human ability to resist in either case (maybe an AIP value for this behavior).

[KDR_11k]

Sleeping Giant AI. I don't think mere presence on an AI planet should be enough to raise the meter, I think only ships that are actually fighting should count so you don't need to flee through a wormhole, just disengage from the enemy and cloakers get some additional benefit. It certainly feels more guerilla to disengage and cloak regularly.

[Drjones013]

I just realized I may have to clarify a bit:

The AI should be sending SF against human concentrations on both human and AI worlds. This means that players who blob will have their blob systems attacked by available SF, both offensively and defensively. The AI would use its periodic reinforcements to assault the human worlds at a strategic level (meaning, the objective Destroy Human Command Center) but SF would be used tactically (disperse human resistance).

I see this as something the AI should be doing anyway, regardless of 'blobbing.'

[TechSY730]

I'm actually talking about something more 'proactive': the AI has probes (or it just knows) to find out where human FP is concentrated and then attacks that area. It has to make decisions as to how to distribute its own FP to attack but the player might make that easier by having a fleet ball in that sector. Special forces shouldn't be 'special defenders' but rather the AI's tactical response to player concentrations of firepower.

We already have a system that says Warp Gate = attack and there's no way that the AI could win if it only used that feature. CPAs tend to attack random regions with some disruptive force (attacks on power and resources) but there should be a mechanic that attacks the player's offensive capabilities on the player's own planets.

Personally I fleet ball and think that it's fine-- as a history student I have to say that an AI programmed by humans would think either on a Clausewitz principle of 'destroy the opposition's ability to resist (ie army)' or OODA (observe, orient, direct, act). I would certainly want to destroy the human ability to resist in either case (maybe an AIP value for this behavior).

There already is a feature where the AI will choose which planet to target based on firepower, threat fleet. However, threat fleet tends to choose planets with the least firepower, as it is the least likely to get curbstomped by a big "ball" of stuff on the other side and live long enough to do actual damage.

The ultimate problem is that if the AI is trying to take out human concentrations of firepower procactively, it would need absurd numbers of ships to be able to reliably take them out, and crazy numbers of ships to assure decent economic damage. I mean, if you can shrug off 2000 fleet ship waves (which is rather easy to set up a defensive system that can), would 3000 fleet ships from wherever really have much chance either?
I suppose that those extra ships could wait for the next wave, and then head on through when it comes in, if it determines that these together would be enough to have a decent shot at taking out the planet (or at least severely wounding it). (All the various "should I go in?" considerations already take this into account, so no new logic would be needed)

Now, I am all for expanding the special forces some:
Things like allowing the special forces to "split up" (so they don't all wait at the same place, but different groups can wait at different places).
Or treating a "planet under significant attack" (aka, a blob) as a planet worth defending (though not as high priority, and certainly not all the special forces as that would make it too easy to bait).
Maybe even have a mechanic that once special forces gets really large (near the cap), some of the special forces will "break off" and be put into threat fleet mode, joining up with the other ships "planning" a strategic offensive strike, slowly leading to a rather large amount of stuff seeking out your "weak points".

Whatever is done, care should be taken such that it wouldn't be trivial to bait the AI.

[Drjones013]

Attacking an enemy where they're weakest makes sense but unless the attack results in significant logistic damage to your target then the casualties (usually 80-100%) incurred by the AI aren't justified. The AI would be better served in pulling those forces out to fight again once it does its 'raid.' Totally down for this.

I also like the idea of SF reaching a cap and 'splitting off,' this could have some really interesting emergent behavior possibilities.

[KDR_11k]

Also there are no battle conditions in AI War that would make attacking a large force head on ever succeed without bringing an even bigger force. If the AI wants to strike your strong points it would need some sort of advantage to hit you with, neither superior numbers (should not happen unless you let AIP get out of hand) nor stuff like carpet bombing your planets with lightning and EMP warheads would be appropriate. The AI can wipe you out if it feels like it but it wouldn't be much of a game if the AI could just randomly feel like it.

[Drjones013]

There *should* be a happy medium here, like the AI gathering a mix of SF but then selectively sending out certain ships to counteract certain fleet balls to maximize damage. The AI is supposedly limited by its fight against the mysterious 'other' force; if I were in the AIs' position I'd be certain to stop (attempt to stop) the humans from being able to gather for a decisive strike.

I'm liking more and more the idea that the AI SF will target production facilities at weak points, or if the player has whipping boys established, coordinating threat fleets with SF, threat fleets to damage defenses while SF fleets attack logistics.

[TechSY730]

Attacking an enemy where they're weakest makes sense but unless the attack results in significant logistic damage to your target then the casualties (usually 80-100%) incurred by the AI aren't justified. The AI would be better served in pulling those forces out to fight again once it does its 'raid.' Totally down for this.

It sort of already does this. If after it take out a planet, it doesn't think it has enough that it can make headway against a next adjacent planet, it will either wait or pull-out (depending on where the AI decides the best "staging" location is) and start looking for the next vulnerable target. (If the ships are normal threat rather than threat-fleet, it may take a bit of time for those ships to "pull-out")

Also, to make the AI a bit more proactive, I suggested that AIs would be willing to go in if they think they have at least a good shot of massive damage, instead of always having to wait until they think they can "win", which would be especially nifty now that the AI is valuing turrets properly.

[KDR_11k]

I don't think the AI should attack without provocation. If it wanted to preempt the humans it would have sent an HK to the home command before the game even began.

[Drjones013]

The AI shouldn't attack without provocation but certainly a 200 AIP is sufficient provocation, no?

What I'd like to see is the AI saying 'oh look, the humans are massing THERE. I should do something about that.' Local defense forces are frequently not enough (which is why the AI reinforces the way it does) but it should have a mobile, unconstrained force to address the threat.

From a lore standpoint, the AI is hoping to deal with the human 'threat' after it deals with its big enemy. That means keeping human resistance down. It's not going to do that by waiting, and conversely, it's not going to do that by just dealing damage that the humans can replace. It needs to do significant resource damage by either

1) destroying the enemy fleet, which is expensive to build,
2) destroying the enemy resources, which are expensive to replace, or
3) disrupting the enemy ability to mass and maneuver, which is strategically delaying.

Also, to make the AI a bit more proactive, I suggested that AIs would be willing to go in if they think they have at least a good shot of massive damage, instead of always having to wait until they think they can "win", which would be especially nifty now that the AI is valuing turrets properly.

Great idea!