the AI combat good and bad

from Sins Forums
the ai's offensive skills have greatly improved.

the ai's defensive skills do not exist.

as i stated in another post the last game i played was 1 star 1 AI. the pirates in that game were a bigger threat.


the AI needs to build more defensive structures, hangers and guns, and fewer of the other structures.

in the last game i played i only used 1 cap ship the rest of my fleet were cobalts. the game was set to normal, normal speed.

with the fighters off of the cap i was able to keep the Ia's fighters and bombers at bay long enough to knock out his carriers and hangers.

the AI should have been able to exploit my weakness or at least attempted to.

the vasari need to do hit and run raids. their caps need to be in system but avoid combat at lower levels.

7,077 views 21 replies
Reply #1 from Sins Forums Top
Daniel, whats the point? We all know the AI needs work -- the Devs have admited it, even. This is a beta. The AI is incomplete because the game is incomplete. While "rough" AI work can and should continue throughout the development process, the fine AI work that actually makes the AI work has to occur at the end of the development process.
Reply #2 from Sins Forums Top
it is called feed back. i believe that is what beta testers are supposed to do. not sit around and attack people who do make an effort to give feed back. if the devs don't want my feed back all they have to do is ignore it.
ObjectDock - Access shortcuts and frequently used applications from an animated dock.
Reply #3 from Sins Forums Top
not sit around and attack people who do make an effort to give feed back.
End of quote


If your calling my post an attack... trust me, it wasn't.

And feedback should be telling the Devs something they don't already know. Kinda like we don't go running around posting about how culture is still causing stringID errors, because its been reported to death?
Reply #4 from Sins Forums Top
It's quite alright to post. There is no requirement to read the whole forum before adding your feedback, I'm sure any duplicate reports (like indeed the missing string thing) will be quickly sifted out by the devs as such, so there's no reason to hold back.
Reply #5 from Sins Forums Top
And feedback should be telling the Devs something they don't already know.
End of quote



yes your right everyone has commented on the fact that the AI needs work and a few have made comments on what the AI needs to do to increase its offensive abilities. but if you will note i made comments on what it needed to do with defense.
Reply #6 from Sins Forums Top
which is a first as far as I can tell.

And yeah Ron asking someone what the point of their post is, and telling them they are being repetitive is essentially asking them to have not posted at all. Which by any definition Must be an attack on their post (not neccessarily on them as a poster though)

I think the AI overall is doing fine for defense, I often don't want to attack them until I'm well equipped because they do such a good job, their fleet command is poor, and their overall expansion needs work... but thats dynamic defense... their choice and quantity of buildings is overall just fine IMO.

I also think it has come allllloooong way too, my current SP game was a pretty good fight with the AIs in my Home Star... my ally and one enemy even went to another star... not soon enough though.
Reply #7 from Sins Forums Top
Daniel is entitled to his opinion, I happen to agree with him.

Yarlen and Kryo seem to do fine at filtering the posts for the Devs so why waste time argueing about it.
Reply #8 from Sins Forums Top
If your calling my post an attack... trust me, it wasn't.
End of quote



Ron i have noticed when someone posts something to improve the game. you usually respond in the negative.
Reply #9 from Sins Forums Top
True But Unfair Daniel, that doesn't mean he is wrong or not entitled to do so.

Don't fight fire with fire. Or I'll start throwing neutron bombs around.   
Reply #10 from Sins Forums Top
It is true. The AI does need work. It is also true that this issue has been reported to death. It is also true that the OP is the 1st person that i have seen to at least to give some sort of input that i am aware of on how to improve the AI. At least in his opinion. I have seen post on what the AI "shouldnt" do, but very little on what the AI "should" do.

We all have bitched about the AI, and the dev's are very well aware of the issue. However, I don't think any one of us (myself included) has made any kind of suggestion on how to "improve" the AI in detail. Other than fix the mass retreats the AI has a bad habit of doing as soon as you jump your fleet into a grav well.

There is no need to start flaming here. The OP is in his right. A simple "this is a known issue" would have sufficed. With the possibility of adding on to to the suggestions. So i will add these suggestions along with what the OP suggested.

1) Fix the mass retreats: Make it to where 75% of an AI fleet is destroyed before the mass routes begin.

2) Make the AI concentrate fire on key targets. Like that Marza that is razing the AI's planet. Right now the AI seems to match units 1 for 1

3) Stop the AI fleets from scattering all over the grav well. This is IMO what makes the AI too easy to beat. They spread out all over the place so you can pick them off 1 by 1 at your leisure.

4) The AI should NEVER retreat from its capitol world.









Reply #11 from Sins Forums Top
To be fair, that's almost entirely kryo doing the filtering.
Reply #12 from Sins Forums Top

1) Fix the mass retreats: Make it to where 75% of an AI fleet is destroyed before the mass routes begin.

2) Make the AI concentrate fire on key targets. Like that Marza that is razing the AI's planet. Right now the AI seems to match units 1 for 1

3) Stop the AI fleets from scattering all over the grav well. This is IMO what makes the AI too easy to beat. They spread out all over the place so you can pick them off 1 by 1 at your leisure.

4) The AI should NEVER retreat from its capitol world.
End of quote

We LOVE lists like this!!!

Reply #13 from Sins Forums Top
A simple "this is a known issue" would have sufficed.
End of quote


A) Daniel's been around long enough that I figured he would have known the standard spliel that followed my version of "this is a known issue" -- its a beta, game still in development so they can't finish the AI yet. Hence my "whats the point" comment

B) Looking back, my tone may have been (more) than a little off, and I apologize for that. Its hard to get your emotive content across through pure text.


Ron i have noticed when someone posts something to improve the game. you usually respond in the negative.
End of quote


Oh really? Take a look over here:

https://forums.sinsofasolarempire.com/?forumid=440&aid=167608#1363112

I hardly responded "in the negative" -- I didn't chime in and say good idea because people had beaten me to it, or point out that the empire "screen" wasn't going to happen because someone beat me to that, but I didn't say anything negative.

In general, I reply "in a negative" under two circumstances -- something that can't happen due to the production pipeline, or something that isn't an improvement. There are plenty of silly, ridiculous WIBNs out there, and I will admit I'm one of the first to say not a chance, but thats hardly the same as being a nay-sayer.
Reply #14 from Sins Forums Top
I disagree with Major Stress that 75% of the fleet should be destroyed before considering retreat. That will mean near 100% loss by the time the retreat is done. The decision should be made way earlier, as detailed below.

Okay, here's some other concrete suggestions.

* The AI moves to attack in a ragtag non-formation, getting slaughtered in ones or small groups

Make the AI collect its forces at a safe distance in the grav well before engaging, if at all possible.

---

* The AI sends special forces like the anti-fighter frigates through the enemy fleet to chase their particular prey.

Have support craft like the anti-fighters stand by the main fleet to defend against any such coming in. Conversely, fighters/bombers should be really conservative about entering the range of craft that will snap them like twigs, including the Kol when it has the flak ready.

---

* All forces will chase special targets like colony ships and siege frigates, forgetting about shooting at the fleet that is smashing them in the meantime.

Have the AI calculate whether the path to these special craft is safe, and either attack the intervening fleet or go around it if it's really set on taking those out.

---

* The AI will happily counterattack any enemy of any strength with whatever it has at hand, before retreating its lone capital ships.

Calculate odds of victory before engaging. Recalculate every 10 seconds or even more often.
- If favourable, attack (in concert!).
- If the odds are even, stand off from fighting until any reasonably nearby reinforcements arrive (for instance one hop away), even if this means losing a structure. Stay in the vicinity of any defences/repair platforms if present, unless the enemy is using mainly long-distance attacks.
- If the odds are infavourable (counting any inbound reinforcements on either side), retreat immediately to the edge of the gravity well or jump out and collect reinforcements before returning to the fray. If there are no reinforcements and the fleet support is capped, engage anyway of course.

---

* The AI will act completely independently when teamed.

Let the AI consider its allies' positions and whether it should join their forces for an attack/defence

---

* When AI capitals have about 2000 hitpoints they retreat, often too late to actually escape.

Calculate the time it will take to reach the edge of the grav well and escape, compare with the time necessary for the enemy to destroy it with the current forces (and take any abilities into consideration too, like ion bolt and phase out). Let the ship stay in combat if this calculation turns out with less time until destruction, as retreating is useless. This rule is only valid for smaller battles, if there's a huge battle the enemy won't be able to follow it without taking significant losses. Also the loss of the capital may actually be gainful if it means you'll destroy the enemy fleet.
Reply #15 from Sins Forums Top
When considering retreating or not, the AI shouldn't care so much about % loses, as much as its strength rating versus that of the enemy fleet.

What the AI needs to learn is that often times it is better to stand and do some damage, rather then try to run, lose all its ships and do absolutely no damage at all.

Concentrating its forces wihin gravity wells will go a long ways to increasing the combat capability of the AI.

Reply #16 from Sins Forums Top
Withdrawing from a losing battle should only be done when the losing battle is counter-productive by comparison to the expected cost of running, and cost is more than just the fleet, but your economic situation, the value of the target, your expected survival percentage if you run versus your expected kill percentage if you stay, and the chance of aid.

If an ai has two or three shipyards holed away and enough resources to replace a 100 fleet point loss and can kill 30% of a somewhat larger fleet in return, that's a pretty good idea. It can then quickly replenish those casualties while coalescing the rest of it's forces for a counterstrike on the now depleted fleet kicking it's ass. If it's broke or can't replace them fast enough, it needs to take the hit and get out of dodge till reinforcements come in. Tactical decisions like staying to kill the siege frigates to slow planet bombardment till reinforcements come should also be considered. If the ai can't afford to lose a planet, can't afford to lose the fleet, and can kill the siege frigates, it should probably kill the siege frigates and only then consider whether running is feasible. A heavily defended planet with a bunch of gauss platforms is vulnerable to even a small fleet of carriers and missile frigates. Considering whether it could successfully take those ships out before losing the fleet would also be a very good thing to do.

If there isn't any winning situation, as in no reinforcements, no hope of a matching or superior fleet, there's no point in just giving away a planet. Above all else, it most definitely needs to understand the alamo concept. You're going to lose anyway so you might as well make them bleed for the place. Knowing it has a choke point and heavily fortifying it for that stand would be nice, but right now it's like a chicken with it's head cut off, a steamroller is a guaranteed win because the ai will never have it's fleet in one place and always try to run either immediately or very shortly after the casualties start to pile up. Then when there is nowhere to run and it's stuck on the starting asteroid with loads of resources and no shipyard to produce and no fleet points to use, the leavings try to fight back.

Reply #17 from Sins Forums Top
as i said


if you want experience for your caps.


they shouldn't even get involved in the battle. they jump in and stay ready to run. when they start taking damage. or in my case the whole fleet targets them they run and the frigates do as much damage as possible.
Reply #18 from Sins Forums Top
This is what i am talking about These are all good suggestions.

I said a 75% loss before retreat as a rough guess. This was mainly to force the AI to stay, and fight vs having to chase them all over the map. Whats even more frustrating is when you know the AI has a force that will match, or better you. Yet they retreat anyway leaving its worlds behind for you to take. That idea was just to get the AI to put up a fight. Especially at its capitol. The AI should take into account that a key strategic world is in danger of being overrun. That may mean having to fight a losing battle if it will buy time for re-enforcements to arrive. AI allies should help an AI that is under attack by sending a fleet to assist. I rarely if ever see this happen.

The cap ships will start retreating as soon as their shields are down, and if there is only 1 cap ship in an AI group the frigates will retreat right along with him. Instead of engaging the attackers, and buying time for the cap ship to escape. This is not always the case, but i've seen this happen many times. If a cap ship is in danger the AI should switch target priority to protect the cap ship. The scattering about the grav wells doesn't help that situation at all, because it leaves key ships wide open, and vulnerable.

The only times i saw the AI stand up, and fight was if they heavily outnumber you, and again if the cap ships get destroyed a mass route will happen. Even if they still have the advantage.

Bovi hit on another annoying thing the AI does which falls in line with the AI "Scattering". That is the AI picking an unimportant target that is on the other side of the grav well, and sending ships through a gauntlet to get to it. Ignoring what is in the way.

Perhaps we can do this for the AI. When an AI enters a grav well they start out at "hold position", then all move in unison to a key posistion. While fighters, and bombers slug it out. If an enemy fleet is there then both will move toward each other (if other fleet is AI) once in range AI will chose key targets and engage using best ships and weapons. If no fleet is present switch to "local area", and concentrate on defenses. At that point siege frigs will come in, and planet bombing starts. All the time recalculating for situation changes like if a new enemy fleet jumps in. The main thing is the AI fleet must stay together which is why i wish there were formations in sins.

The only time the AI should ever have aggression on "gravity well" is if there are no enemy ships, or defenses present.






Reply #19 from Sins Forums Top
including the Kol when it has the flak ready.
End of quote


... Since when does the Kol's flak ability do anything but look pretty?
Reply #20 from Sins Forums Top
The Kols flak does take out bombers... Sometimes.

The ability just isn't near as effective as having a group of flak frigs escorting her.
Reply #21 from Sins Forums Top
Level 3 is pretty effective really, if you had two together and good timing, you could kill just about everything all at once with them. Since it hits everything for a little though, the first level is worse than not having anything, at least if it were in shield mitigation you'd slow down the damage, but killing them 20% at a time is slow...