AI Defense System: Moving Forward From Core Shield Generators

[TechSY730]

If I had realized that the AIP increase during deep-raids was an option for other, I would of voted for that instead of deep-raid triggered waves.

[Mithror]

Wow, a lot of discussion seems to have been going on while I was playing my game. (22 hours in and still haven't finished. I'm so slow!).

I'd like to chip in my opinion on this matter as well. Hope you don't mind. I am fairly new to the game.

I am going to draw examples from the game I'm currently playing, because well... that's all the experience I personally have!

First, I guess, we should kind of define deep striking. What criteria are to used to label an assault a deep strike. Is it when you have to go past 1 AI planet? 2? 3? In my game I have on average been sort of forced to skip 1, often 2 planets. I now need to take one extra planet, because it was the only one with an ARS left so it got the only CSG of type A. I'm not really bothered by it, because I intended to take that planet anyway as it's the planet I want to attack the second AI home world from. However, I do want to skip the 3 planets in between because they are of little value to me. They don't have high resources, they don't have advanced factories and they don't have fabricators. If we were to implement a system where because I'm attacking said planet, I would accumulate 1 AIP per minute, I could have just as easily taken another planet in between, because I'm sure it's gonna take me more than 20 minutes to take the planet. Of course the numbers can be tweaked to still make deep striking worthwhile, but I'm not very much in favor of this system.

Now, in my game, as I said, I did often skip 1 or 2 planets because they didn't offer me that much and the tutorial did teach me about the tactic as well as the many comments saying you need to carefully consider the planets you take as to not get the AIP too high. Having done so I now find myself in a difficult position to defend. I couldn't gather my troops anywhere for a big attack, because everytime I left a planet, the AI would come and attack it. This made me belief that there is a disadvantage to deep striking and that is that it spreads you out too thin. It creates too many attack angels for the AI to annoy you, so in my game I have started taking over the planets I skipped in order to reduce the number of planets the AI can attack me from. This causes my AIP to increase dramatically, but I think I can hold it. It wasn't easy to capture some planets fyi, because of the AI attacking me when I was attacking a planet, but doing this I feel I'll be safer, which brings me to my second point.

Why are you so against turtling in this game? From what I'm experiencing in the game, it seems like it's the only valid tactic, especially with high AIP. In RTS's there are typically 3 types of play styles: rushing, booming and turtling, where, in general, rushing beats booming, booming beats turtling and turtling beats rushing. The caveat with this system is that it is always best to rush, because rushing gives scouting information on your opponent. If he is booming, you win. If he is turtling, you switch to booming and simply outmacro him. It's this adaptability that most AI's in RTS games lack, hence why turtling is a popular method of play against the AI. I do believe AI's have evolved somewhat to punish turtling in RTS games. It used to be far easier to do so in past games (C&C) than in current games.
In AI War, the AI does not boom or rush, it mostly turtles. If you increase the AIP by taking planets (which could be considered booming), the turtle style gets accompanied with rushes. Since the AI does not boom and it will rush you eventually, you need to turtle. Sure, initially you are both turtling, but this is fine. You can use this to boom a bit, but eventually you'll need to start turtling to hold of the AI rushes. AI War is, I feel, a war of attrition. You may not like it, but its premise makes it so. You are playing against an opponent who vastly outnumbers you. The only valid way to win such a war is by attrition (guerilla tactics for example). Though I can always be conviced otherwise! To me it just seems like it's the most logical thing to do.

I think it would help if you could explain what type of gameplay style you want the players have to adopt so that we can make changes accordingly. If you don't want to see a turtle style play, what would you prefer us to do/try? Making deep striking more difficult will, as I see it, only increase the number of people who turtle.

From everything I have read, it seems there really is but 1 core issues you want to solve: players should not be able to win with a too low AIP (i.e. too early). Personally, I don't really mind that this is possible. Such is the case with most other games against an AI. Take RTS games as an example again. If you 10-pool an AI opponent in Starcraft, you'll more than likely win the game fast and easy. Is it fun? Definitely not and people will get bored of it pretty quickly (though doing it against a human does seem to make it more fun...). It would be fun if the AI could actually stop the rush like a good player would be able to and turn the game into a more intense battle. So I think the solution to this problem is not specifically to solve this issue by introducing new mechanics (though I actually like the CSG system, but it does need a few tweaks), but by improving the AI behavior. Here's what I propose.

My Proposal: Approriate AI response to player threats.

If the player is rushing for the AI, he should be able to detect this. After all, there are transport ships moving towards his home world. If you were doing stuff and suddenly you spot the player moving closer and closer to you, how would you respond? You would be more wary of this player. In AI War terms this would mean you'd increase the AIP. If a player comes within then less then or equal to one max transport trip of the homeworld, the AIP should immediately jumps up to (not with!) a certain number (say the 500 as was mentioned in the game before). If the AI was already at 500 or up, it would not change (the AI is already wary of you). Furthermore. If the AI makes this sudden change (so it was not already at 500 or up), it would get (one-time only) extra reinforcements on the home world. The number of reinforcements it could do would scale according to how much the AIP jumped. So if you were ate 499, you wouldn't have to deal with as many reinforcements as if you were at 200.

What this does is simple. If you play the game as you should, you'll slowly raise awareness (the AIP) until you reach a point where the AI is not liking you at all (500 AIP?). Moving in on the homeworld would be no different as we're playing now. However, if you want to rush the AI and think about taking him early, the AI will get pissed. How dare those puny humans attack him! He'll raise the AIP to 500 regardless and reinforce his homeworld dramatically, making it much much harder for the player to actually take it over.

This could be changed to tweak it more of course. Like having multiple reinforcements instead of one. Or reinforcing on both the homeworlds and the core worlds. I just feel this solution is more in line with the AI behaviour, rather then just an added mechanic, which most of the solutions seem to be.

(This does not prevent turtling, but none of the other solution do this either. With the CSG what most people seem to be wanting to do is take the planet, kill the CSG and then leave the planet to not have to worry about its defense.)

Hmmm, long post. Sorry 'bout that. I'll proofread it when I got some more time as I'm sure there are some mistakes in there (non-native English and such.). I wanted to post this in the other thread, but it was closed, so I hope you don't mind it sitting here. It does kinda belong here.

[Lancefighter]

So basically, attacking a homeworld incurs and automatic total of 500 aip, regardless of current aip?

Maybe also put the second hw on permanent alert for the reinforcements :p

[Mithror]

Threatening to attack the homeworld would increase the AIP to a certain number (I used 500, could be anything), but on top of that it would make the homeworld harder to take because the AI would get extra reinforcements on its homeworld. (Also I assume increase in AIP would be done through the AIP floor?)

I thought about making the second homeworld on permanent alert. It makes sense from the AI point of view, but it might make the game a lot harder. It would certainly encourage the player to attack both at once. I'm still on the fence about that one

[Ixiohm]

I have followed this and other threads on the AI core shield generators and their alternatives with great interest. I like several of the ideas already presented – however I voted for the core shield generators (although I agree with others that a single network would be a far more elegant solution).  Even so, I thought it time to contribute an idea of my own, or more correctly an alternative implementation of the hunter/killers which is quite different from Hyphas idea of them

The idea is inspired by the siege of Gondor from lord of the rings, where the city has a multi layered defense with several city walls arranged in concentric circles, when one wall falls the troops retreat inward and the fight intensifies until the epic stand of at the city core.
 
My proposed mechanic is similar, but kind of reversed. When the humans first penetrate the AIs defensive outer perimeter a hunter/killer will spawn at the AI home world - think a high mark mobile fortress able to go thru wormholes and totting a high mark ion or beam cannon. Preferably with a taunt from the AI - "Puny humans your advance stops here! Tremble at the might of the hunter/killers". The hunter/killer will roam inside the defensive perimeter and targeting human fleets based on Mobile Military (Firepower) and it won't pursue fleets outside its defensive perimeter. This could be happening early game or it could happen later depending on in which direction the player is expanding, as it is not directly linked to the number of planets the player holds. This should make it safer to expand away from the AI home world than towards it and deter from deep striking, as the strike forces will attract the attention of the hunter killers when they enter the AIs defensive perimeter.

The defensive perimeter is defined as any system x hops out from the AI home world. For example the first defensive perimeter is 3 hops out and the second perimeter is 5 hops out. This will likely have to scale with the shortest distance between an IA home world and the human home world, but the defensive layers should probably be 2 or 3. The hunter/killers should have about 3 marks levels corresponding to the number of defensive layers the AI has, and possibly some different configurations (ion cannon or heavy beam cannon). If a hunter killer gets destroyed it will re-spawn at the AI home world after say 20 minutes.

[NickAragua]

What would make me happy is an option to turn this "AI Defense Systems" stuff off.

I don't really understand all this effort to prevent an "early rush" against AI homeworlds, to be honest. If people want to skip, hop and jump to the AI homeworld with transports and finish a game in half an hour, that's their problem. I also don't particularly like being required to take any particular system or set of systems - a large chunk of the fun of this game is that you pretty much get to pick your objectives.

[Red Cossack]

I think a lot of the resistence to the shields in in how they are described/perceived.  If instead it was said that they are invulnerable shields and you must plunder certain AI systems to procure the codes to infect them with a worm and shut them down, people might be easier to accept.  I'm kinda detecting a fiction/lore/RP bias in some of the folks most opposed. 

In the end however, I think that having a selection of victory conditions is the best way, and considering the Spire storyline addition, it makes perfect sense.  Maybe even add a new one each expansion.

One idea I had, is to allow people to rush the Homeworlds, but have it take a certain amount of time to crack the shields and you have to hold off AI waves/units until you succeed.  I though the mechanic that makes the AI send out waves when it's neighbor worlds are not held by the AI was a step in this direction, and a similar mechanic could be a choose able option where each core item that is destroyed adds another wave of ships in that same system.

It's about choices, and the devs seem to be giving us choices and asking our input, so i think it will end up better in the end.

[x4000]

Okay -- time to drop back and punt.  Here's what we're going to do, for now:

1. Core Shield Generators will be default-on, but will be something that can be disabled in the fashion that Astro Trains always were.  By default for existing saves, they will be disabled.  There will be a command that people can use to re-enable them if they like.

Why?  Because enough people love the CSGs for reasons unrelated to the prevent-rushing discussion that I can't take them out: those people would be unhappy.  And I'm one of them.  On the flip side, enough people hate them, or even just slightly dislike them, that the option to turn these off should be there (as had been the case with Astro Trains before they were delegated to AI Plot).  But for newbies, I think these CSGs are actually helpful rather than harmful, as they provide some intermediate goals and good direction.  If you don't like them, you can turn them off.  Everyone should be happy with this, hopefully.

2. I'm not -- yet -- going to implement any other forms of AI Defense Systems (and yes, AIDS is a terrible acronym for that).

Why?  Because those other options require time and effort, and we have things that are clearly more important to everyone to be working on in the meantime -- like bonus ship balance.  I'd rather have that fixed up then a wide array of random AI Defense Systems.  I don't really agree with the statement that having multiple AI DSes would fragment the player base, as the playerbase is already pretty fragmentary in a lot of senses, but I do agree that having one more prime-level lobby option isn't a grand thing.  If we ever do any more of these, it would probably be a long while in the future.

3. The AI Progress for Deep Striking thing is going to be implemented, or something along those lines -- that was always a separate discussion from here, though, so I'm locking this thread but leaving the other one open.

Why? Because having the AI respond to deep striking is a completely different discussion from the Core Shield Generators at this stage.  At this point, my reason for keeping Core Shield Generators is related to the stuff in #1, not related to the deep striking stuff.

4. We're also making other semi-related changes to both bonus ship types (making ARSes more of a bonus), and to AI Progress floor.

Why?  These other semi-related changes will better address the problems with both the ARS and the super-low-planet-count games.  Again, moving away from CSGs or the AI DS concept as a solution for these things, and just keeping CSGs because of their other unrelated-to-this benefits for new players and general structure overall.

Thanks to all for writing in, and to those who PM'd me in support of CSGs, as well.  Time to nail this one to the wall and move in more important directions; hopefully the above is something that everyone can live with, since it allows you to play with CSGs if you like them, or ignore them if you hate them.  And it won't mess up your existing saves, AFTER the next patch.  If you don't want CSGs in your existing save, make sure not to update until the next patch.

Thanks!