Guardians a little overzealous about counterattacking

[Giegue]

except the AI is alot smarter than mindless bees, and is described as not paying complete attention to the player.

plus, this sort of leaves an easy strategy to beating them: send in a couple ships to lure the AI away from its world, leave a mass of ships on the other side to destroy it, and raid whats left. It should work every time...


okay, look at it this way. two knights are assigned to guard a valuable peice of artwork that is treasured by their king. they spot someone walking near it, and relentlessly chase him out of the building, and keep chasing, leaving their post unprotected. another guy walks up, and takes this piece of artwork, while his buddies deal with the two knights who have just walked into a trap.

I'm all for the counter attack thing, but most of the guardians shouldn't be able to leave their planet. I understand that a select few should be able to, such as the raid guardian or something, but a gaurdian who is so persistant as to chase after an enemy and abandon the post they were assigned to isn't a very good gaurdian at all.

if not their behavior, I suggest a name change.

[HitmanN]

The issue isn't whether it's a good or a bad choice. It's a predictable choice, and as I recall it, 3.0 was advertised for having an unpredictable AI. Is this feature becoming extinct with 4.0?

You're right, the best thing for the AI would be to send everything it has at you the first second of a new game. But it doesn't. Why? Because the AI is supposed to simulate how much attention it's giving to your actions, aka AI Progress.

I'm just finding it hugely illogical that the AI would make a decision that sounds more like an AIP 1000 decision in a completely random occurrence like a player poking any planet at any time a little. Sending everything on a planet at the the player battles against the 'simulated concern' logic completely. It's like its concern is suddenly maxed out.

And yes, the Starfleet Commander is insane, that's definitely not something I'm arguing about, and it just reeks of a major imbalance when combining two seemingly ordinary features of a game together results in something so colossal in difference to anything else on the same AI difficulty/type level.

But, in the end all I'm concerned about is the way the AI acts predictably in this scenario, and in a manner not befitting of the assumed simulation of the AI's concern over player's actions. A full-on counter-attack is what I'd assume to possible starting at about AIP of 300 or so, when the AI really starts to worry about the player's accomplishments.

"z0mg! The player is attacking with a dozen Mk1 ships! Crush them with a hundred Guardians and Starships!" just doesn't compute. Actually... it's a pretty stupid decision, to be honest.

[Giegue]

I suppose its time to ressurect my "insanity" AI type suggestion? it seems to apply more now then it used to.

[Wingflier]

I also agree that in "standard" situations, it's a little ridiculous for the AI to send their whole planet worth of guardians and starships for every "poke".  However, as I mentioned in my first post, this isn't something that I have commonly experienced (not to say that it isn't happening to you guys).  From my experience, I have to do a whole lot more than a little poking to evoke that kind of response.  That kind of thing usually happens after I destroy a major structure like an AI Eye or Fortress.  From a logical and design standpoint, this makes sense to me because the AI sub-commander probably realizes that I've just made a major dent in their base, and that if they just sit and do nothing, I'll probably be back to do it again and again until they are all wiped out.

"z0mg! The player is attacking with a dozen Mk1 ships! Crush them with a hundred Guardians and Starships!" just doesn't compute. Actually... it's a pretty stupid decision, to be honest. Roll Eyes

I somewhat disagree with this.  The AIP (from my understanding) does not affect how the AI reacts to your decisions on each individual planet.  The AIP simply delegates how many reinforcements they get, how big the waves are, and over time upgrades the enemy ships to a stronger type (Mark II, III, etc.).  The AI sub-commanders on each planet are responsible for making their own decisions based in the current situation on how to react. 

On an individual planet level, before you attack, the sub-commanders probably don't consider you much of a threat.  They may send their excess units against you (border aggression) after their entire planet is swarming with units, but generally they aren't that concerned with you.  However, once you start sending attacking units into their base, suddenly they change their stance on things.  What does it matter that it's only MK1 units?  The AI are methodical, they don't take chances, they do not have pride or hubris like humans do.  They realize you have become a threat, and send what they have on that planet to crush you.  If the AI, after being provoked, sends its force to take out a vulnerable target when its the BEST DECISION to do so, I completely agree with that course of action.  However, when they are just doing it every time you poke them in every situation (which once again, is something I haven't experienced), then I agree it's very predictable and stupid.  But those are two very different things.

[HitmanN]

A few good points there.

From my experience, I have to do a whole lot more than a little poking to evoke that kind of response.

It seems to be a common tactic these days to 'lure' defenses away from a planet, into defenses, preserving offense to be used after that. Or that's the impression I've gotten from randomly reading the forums here and there. It's predictable and common enough to be reliable enough method. Dunno if it's tied to a difficulty level. I've used 7 and 7.3 myself.

That kind of thing usually happens after I destroy a major structure like an AI Eye or Fortress.  From a logical and design standpoint, this makes sense to me because the AI sub-commander probably realizes that I've just made a major dent in their base, and that if they just sit and do nothing, I'll probably be back to do it again and again until they are all wiped out.

A true point of retaliation makes sense. That's what I'd enjoy. A threshold, which varies by the severity of the situation and the AIP, at which the AI decides a full-on counter-attack is needed. Shouldn't still be predictable though. Sometimes killing an Eye of Fortress might not be enough, other times you may not even need to do that much, maybe at high AIP.

The AIP (from my understanding) does not affect how the AI reacts to your decisions on each individual planet.  The AIP simply delegates how many reinforcements they get, how big the waves are, and over time upgrades the enemy ships to a stronger type (Mark II, III, etc.).

Yes, and AIP was what defined the difficulty in 3.0, at least for me. All the games I lost in 3.0 were because the AIP got too high and I couldn't handle it. It was the AIP alone that made things harder. In 4.0 so far it's been almost completely the opposite. The only victorious game I played on 4.0 was chaos at the start, but almost a joke at the end, despite a relatively high AIP, and Avenger plot for both AI's (thankfully those have been buffed now, right?). I took several really unneeded planets too, but the consequence of that was more positive than negative in end-game. In the game I've been referring to in this thread, it's been totally hard, maybe impossibly so from the start.

Basically, what I'd like to see is the AIP returned to being the governing factor of how hard the AI is, and how badly it pushes on you, and regarding all of its actions, not just reinforcement and wave mark levels and numbers. Retaliation thresholds, CPA frequencies, Sudden raids, etc. Low AIP should make the AI not bother with minor losses, while high AIP makes every planet a literal beehive you really can't touch without consequences. 3.0 was slow and easy at the start, tough at the end. 4.0 has been mostly hard at the start, and mostly easier in end-game. That's my experience anyhow.

However, once you start sending attacking units into their base, suddenly they change their stance on things.  What does it matter that it's only MK1 units?

The Mk1 units are unable to cause damage against the defenses the AI has. The AI could just sit still and the Mk1 ships would probably blow up anyways. They should have enough ability to gauge the strength of the opposition, just like humans can. They're AI's. Calculating strengths, numbers and odds is what they should be good at. You don't hunt rats with cruise missiles.

The AI are methodical, they don't take chances, they do not have pride or hubris like humans do.  They realize you have become a threat, and send what they have on that planet to crush you.

Yup, and once I've fled through a wormhole, I'm no threat anymore. Defensive ships should return to guarding the AI's vital structures, and patrols should look for the escapees, if they have any reason to. Why follow and abandon guarding duties with full force for something that wasn't even able to cause anything.


BTW, nothing I say here is meant to be offensive. It's not like I hate AI War 4.0 or anything. Just sharing opinions and experiences.

[Wingflier]

It seems to be a common tactic these days to 'lure' defenses away from a planet, into defenses, preserving offense to be used after that. Or that's the impression I've gotten from randomly reading the forums here and there. It's predictable and common enough to be reliable enough method. Dunno if it's tied to a difficulty level. I've used 7 and 7.3 myself.

I won't deny that it could be a possibility.  I've been using difficulty 8 however, and it's not something I've experienced without destroying something "valuable" to them.  If it is happening on the lower difficulties, I'll assume it was an intentionally designed mechanic to make the game easier for the player; whether or not it makes the game too easy at that level is just a matter of opinion.

A true point of retaliation makes sense. That's what I'd enjoy. A threshold, which varies by the severity of the situation and the AIP, at which the AI decides a full-on counter-attack is needed. Shouldn't still be predictable though. Sometimes killing an Eye of Fortress might not be enough, other times you may not even need to do that much, maybe at high AIP.

I agree with this, the threshold even on "hard" difficulty is fairly predictable, but once again I'm only on difficulty 8.  On difficulty 10 I assume the AI wouldn't mindlessly send their forces at me unless they had a very good reason to do so. (This could be false, I'm just assuming the AI makes the best decisions on the highest difficulty).

Yes, and AIP was what defined the difficulty in 3.0, at least for me. All the games I lost in 3.0 were because the AIP got too high and I couldn't handle it. It was the AIP alone that made things harder. In 4.0 so far it's been almost completely the opposite. The only victorious game I played on 4.0 was chaos at the start, but almost a joke at the end, despite a relatively high AIP, and Avenger plot for both AI's (thankfully those have been buffed now, right?). I took several really unneeded planets too, but the consequence of that was more positive than negative in end-game. In the game I've been referring to in this thread, it's been totally hard, maybe impossibly so from the start.

Once again, I can't speak from playing any of the expansions (Zenith or Children), but on difficulty 7 the only real time I felt threatened was at the beginning as well.  After you get past that first mountain and reach critical mass, the AI doesn't seem to put up much of a fight and/or attack you enough to be a threat.  There is a BIG jump between 7 and 8 though, it is unbelievably different.  At 8 it's almost the exact opposite, if you are having trouble in the beginning, you aren't going to make it past the mid game; things just start getting harder and more insane the longer you take to win.  From what it sounds like to me, some of the AI types in the expansions are a bit overpowered at the start (especially that Fleet Command one holy crap), and so this may be leading to the impression that the game only starts out difficult, and gets easier as it goes on.  I could be completely wrong about this though, but the original game doesn't have a lot of the mechanics that could, in Starcraft terms "cheese" you at the beginning, like the Zenith Planet-Eaters, Dyson Spheres, and who knows what else.  The more content of that nature you add to the game, the higher chance you'll get of a bad start (in my opinion).

Basically, what I'd like to see is the AIP returned to being the governing factor of how hard the AI is, and how badly it pushes on you, and regarding all of its actions, not just reinforcement and wave mark levels and numbers. Retaliation thresholds, CPA frequencies, Sudden raids, etc. Low AIP should make the AI not bother with minor losses, while high AIP makes every planet a literal beehive you really can't touch without consequences. 3.0 was slow and easy at the start, tough at the end. 4.0 has been mostly hard at the start, and mostly easier in end-game. That's my experience anyhow.

I would agree that it might be neat if AIP affected how the AI reacted to you on a planetary level.  However, I would recommend playing difficulty 8 against one of the "more balanced" AIs in the expansion, and see if you still experience the game getting "easier" as you go on.  This could help in "narrowing down" the problem for us.

BTW, nothing I say here is meant to be offensive. It's not like I hate AI War 4.0 or anything. Just sharing opinions and experiences. Smiley

Me either!  I enjoy improving this game just as much as you do, I love good intellectual discussion.