Hows the new AI? » Forum Post by Horatio Hornblower

Edit: Just for the record, I'm not arguing that they shouldn't make the AI better. In my opinion, the better the AI is, the more I like it. Good AI = Good Thing. What I'm arguing is that it is NOT as easy as "just make the AI do X"

Never have I said it would be easy. Thats just your interpretation of my post. Im saying that the principle is easy, not the implementation.

And as for humans being able to outsmart AI, or exploit as you call it, that will always happen with the current tech we got. But that shouldnt stop the devs from improving it and having the AI be watchful of players ships close to their planets will, in my opinion, do that.

You dont, then thats fine and there is nothing else to discuss.

Successfully updated karma reason!

And I agree with Legacy, Yamota. You're suggestions simply aren't at all practical. There is no AI that would be able to out tactic a human player without some sort of rule change.

Well they are just ideas to improve the AI. But the main principle should be that if the AI see you lining up warships and transports around all his planets then it should react better than it currently does.

Also why are you two putting words in my mouth? Where have I ever said or hinted that the AI would be able to out tactic a human player?? Im talking about the very basic defensive tactics here and if thats not practical then the AI will always be easy to conquer and this whole thread is pointless.

Right now the AIs only defensive tactic is to put a bunch of weak ships in orbit without fleet management. And that's weak!

Successfully updated karma reason!

Wrong. They *sound* very basic. They are in point not that basic when it comes to coding. They are principles, not a coherent strategic system. And the thread is not pointless if you follow a 'rules change' strategy as oppossed to a strictly 'ai change' one.

Successfully updated karma reason!

Be that as it may, any reasonably democratic society usually frowns on 'unjustified' wars against allies. If you really care about realism (as oppossed to a realistic justification), make this one of the downsides to non-imperial governments. Some sort of penalty should be attached, though to balance out the huge, obvious advantage it gives to a player while still retaining the mechanical relevance of an alliance.

I would agree if this were Civilization, but Galciv is its own game. And we're talking about aliens here. Like the Borg, they LOVE allies. Oh yeah. Assimilate them all!

I do think it's perfectly reasonable never to get another ally again, though. If you backstab one of them, you just get redded out on the diplomacy screen with some message like, "you don't have a good track record with your allies, I don't think I'd want to be your ally".

This is silly. So you know an alien would not have a concept of a just war versus other aliens; despite the fact that they are presumably subject to cultural influence effects, trade technology, etc.

Besides which, this contradicts your earlier criticism that a morale penalty was ahistorical; now you are saying this isn't history, its fantasy.

So what criteria are you using to say that a moral penalty for backstabbing wouldn't be appropriate?

Successfully updated karma reason!

Actually, the more I think about, it might be more straightforward to simply give a large upfront diplomatic penalty for the use of the quick strike tactic, as oppossed to changing the war declaration rules, which could be cumbersome.

Successfully updated karma reason!

Hi!

A perfect exploit, then! Build up an alliance, then sneak-attack when they least expect it!

IMO there should be a "brake" to such kind of attacks. Like:
- a player can not attack a race he's not at war with;
- war MUST be declared in diplomacy window, but attacks can start only in the next week;
- the Non-Aggression Pact can be broken only diplomatically. The declaration of war can not happen in the same week as the NAP was cancelled;
- the Alliance can be broken only diplomatically. Relations between both races are changed to NAP. The declaration of war can not happen in the same week as the Alliance was cancelled.

Just those few simple rules could make sneak attacks impossible. To attack an ally would require 4 weeks passed before combat would start, and would give AIs quite more "room" when assessing threats.

BR, Iztok

Successfully updated karma reason!

Actually, the more I think about, it might be more straightforward to simply give a large upfront diplomatic penalty for the use of the quick strike tactic, as oppossed to changing the war declaration rules, which could be cumbersome.

Yup, suprise attacks should be regarded by other civilizations in a very bad light.

Successfully updated karma reason!

Probably the best counter to a sneak attack is to make sure you have fleets of your own watching whatever foreign ships are coming near your planets and starbases, so that you're ready to beat them down immediately if hostilities start. Whether you should execute a preemptive strike should be based on an analysis of how much damage they're prepared to do in the first turn. In other words, for each ship/transport/spore ship they have at a distance close enough to hit a target in a single turn, they should be regarded with more and more suspicion, especially if the target is vital and defenseless.

This'd only work, though, if the AI can react on a turn per turn basis, without the one turn delay it has now. Ideally, the declaration of war would not be a diplomatic message, but a bullet, laser, or missile through your transports.

Successfully updated karma reason!

Probably the best counter to a sneak attack is to make sure you have fleets of your own watching whatever foreign ships are coming near your planets and starbases, so that you're ready to beat them down immediately if hostilities start. Whether you should execute a preemptive strike should be based on an analysis of how much damage they're prepared to do in the first turn. In other words, for each ship/transport/spore ship they have at a distance close enough to hit a target in a single turn, they should be regarded with more and more suspicion, especially if the target is vital and defenseless.

This'd only work, though, if the AI can react on a turn per turn basis, without the one turn delay it has now. Ideally, the declaration of war would not be a diplomatic message, but a bullet, laser, or missile through your transports.

This is easier said than done. I have a feeling attempting any type of fancy tricks with the AI in this regard will simply lead to more holes.

The best, easiest solution, IMO, is to implement severe diplomatic penalties for the actual use of these types of attacks. This will mean that other AIs would more likely be at war with the player once his first target goes down and thus prepared to take down incoming transports.

Successfully updated karma reason!

I would argue that it is a relatively simple solution, albiet one that would require tweaking to have optimal values for the variables in question. I'm aware I don't have access to the AI code, of course, so it could be more complicated than I'm guessing, but it seems like a fairly trivial kind of check.

Loop through all foreign vessels in your space, look to see what planets and resources they are within a single turn of, and base some kind of suspicion modifier off that. When the AI sees enough camped out ships and transports, there is a rapid decline in relations and possibly a diplomatic message. If the suspicion of an attack is high enough, they could even declare war outright without having to go through the decline in relations.

As for the AI moving ships on a per turn basis instead of setting waypoints... yeah, that might be more involved, but that should probably be fixed anyways, if the AIs are to become better able to tactically go toe to toe with the player.

Also, shadowing enemy fleets is tough to do, but I was under the impression that they already did this with transports, so it could be extended to include combat fleets as well.

Successfully updated karma reason!

I'm not sure what you are talking about. The AI ALREADY does this, in fact I count on it to get the AI to declare war on me. If relations are very good to start off with, sometimes it takes a little more for me to make them suspicious enough to declare war, but that is why the strategy works at all. If they never declared war, then I would have to do it myself, and suffer negative diplomacy modifiers. What I do is make it look like I'm always defending myself by manipulating the AI into declaring war on me. Then in "defending" myself from the AI aggressors war declaration, I wipe them out.

As for other solutions about deteriorating modifiers for a surprise attack, I'm also not sure how you would calculate that. I am specifically not suprise attacking. I just move into attack position before war is declared (kinda like the allies did in the gulf wars). What I am doing is swallowing an enemy empire in one turn, so that I don't have to worry about a long war, and because most AI ships are launched and in fleets, and so not effective at defending my focus on attacking planetary defenders.

Successfully updated karma reason!

I'm not sure what you are talking about. The AI ALREADY does this, in fact I count on it to get the AI to declare war on me. If relations are very good to start off with, sometimes it takes a little more for me to make them suspicious enough to declare war, but that is why the strategy works at all. If they never declared war, then I would have to do it myself, and suffer negative diplomacy modifiers. What I do is make it look like I'm always defending myself by manipulating the AI into declaring war on me. Then in "defending" myself from the AI aggressors war declaration, I wipe them out.

As for other solutions about deteriorating modifiers for a surprise attack, I'm also not sure how you would calculate that. I am specifically not suprise attacking. I just move into attack position before war is declared (kinda like the allies did in the gulf wars). What I am doing is swallowing an enemy empire in one turn, so that I don't have to worry about a long war, and because most AI ships are launched and in fleets, and so not effective at defending my focus on attacking planetary defenders.

You answered your own question. If you were able to take a planet or more within one turn of a war declaration, the computer could view that as a surprise attack. The negative modifier would apply to ALL AIs, not just the one you are attacking. I.e. suspicion is so strong once you have initiated one attack, that it prejudices all AIs towards you. Simple, no?

Successfully updated karma reason!

I would argue that it is a relatively simple solution, albiet one that would require tweaking to have optimal values for the variables in question. I'm aware I don't have access to the AI code, of course, so it could be more complicated than I'm guessing, but it seems like a fairly trivial kind of check.

Loop through all foreign vessels in your space, look to see what planets and resources they are within a single turn of, and base some kind of suspicion modifier off that. When the AI sees enough camped out ships and transports, there is a rapid decline in relations and possibly a diplomatic message. If the suspicion of an attack is high enough, they could even declare war outright without having to go through the decline in relations.

As for the AI moving ships on a per turn basis instead of setting waypoints... yeah, that might be more involved, but that should probably be fixed anyways, if the AIs are to become better able to tactically go toe to toe with the player.

Also, shadowing enemy fleets is tough to do, but I was under the impression that they already did this with transports, so it could be extended to include combat fleets as well.

As already noted, the AI already does this, and as already noted, having the AI shadow the player's fleets, while sounding nice in theory, is complicated in practice. A simple rules change, with a negative diplomatic modifier from the actual use of a surprise attack, is much more simple and doesn't open the way for further exploits.

Successfully updated karma reason!

I'm not sure what you are talking about. The AI ALREADY does this, in fact I count on it to get the AI to declare war on me. If relations are very good to start off with, sometimes it takes a little more for me to make them suspicious enough to declare war, but that is why the strategy works at all. If they never declared war, then I would have to do it myself, and suffer negative diplomacy modifiers. What I do is make it look like I'm always defending myself by manipulating the AI into declaring war on me. Then in "defending" myself from the AI aggressors war declaration, I wipe them out.

As for other solutions about deteriorating modifiers for a surprise attack, I'm also not sure how you would calculate that. I am specifically not suprise attacking. I just move into attack position before war is declared (kinda like the allies did in the gulf wars). What I am doing is swallowing an enemy empire in one turn, so that I don't have to worry about a long war, and because most AI ships are launched and in fleets, and so not effective at defending my focus on attacking planetary defenders.

To further clarify...the negative modifier wouldn't just apply to a declaration of war by you, but regardless. This modifier would result from your actual tactical performance, which could also be seen as the result of the aggressive placement of your ships, which would rightfully raise suspicions among other players. This would almost always only affect the human player, but then the human already gets substantial benefit from the diplomacy system, so no harm no foul.

Successfully updated karma reason!

Wow, talk about a fast-moving thread

Also why are you two putting words in my mouth? Where have I ever said or hinted that the AI would be able to out tactic a human player?? Im talking about the very basic defensive tactics here and if thats not practical then the AI will always be easy to conquer and this whole thread is pointless.

Right now the AIs only defensive tactic is to put a bunch of weak ships in orbit without fleet management. And that's weak!

Now who's putting words in other peoples' mouths? I'm not trying to tell you that the AI will ever out-tactic the player. Just the opposite, in fact. I'm telling you that your ideas are unfeasable because of the AI's inherent tactical weakness. Your ideas rely on giving the AI some measure of humanlike intuition - the ability to "feel" a situation, if you will - which simply Will. Not. Happen. So, I'm saying that instead of trying to figure out how to do the impossible, we should try a different tack.

Jasamcarl's idea is one such "other tack". It doesn't rely on trying to program the AI to be a humanlike strategist/tactician, but on making an exploitive player face the prospect of a war on up to nine fronts...which, for all its weakness, the GC2 AI is more than capable of winning.

Successfully updated karma reason!

I think the AI has the same sensor limits as humans. So, a transport with 8 moves, which he can't "see," wouldn't register as a threat because it wouldn't register at all. We wouldn't want the AI to have more sensory ability than the player, would we?

Perhaps the AI should put more stock in sensors, because that would help it immensely. But maybe Frogboy has played this all out and it doesn't work so well if the AI puts too much stock in sensors.

Anyway, there needs to be a fix to this. Not that I'm implying it would be easy or even feasible at this point, but we must admit that this is a problem.

Successfully updated karma reason!

I would argue that it is a relatively simple solution, albiet one that would require tweaking to have optimal values for the variables in question. I'm aware I don't have access to the AI code, of course, so it could be more complicated than I'm guessing, but it seems like a fairly trivial kind of check.

Loop through all foreign vessels in your space, look to see what planets and resources they are within a single turn of, and base some kind of suspicion modifier off that. When the AI sees enough camped out ships and transports, there is a rapid decline in relations and possibly a diplomatic message. If the suspicion of an attack is high enough, they could even declare war outright without having to go through the decline in relations.

As for the AI moving ships on a per turn basis instead of setting waypoints... yeah, that might be more involved, but that should probably be fixed anyways, if the AIs are to become better able to tactically go toe to toe with the player.

Also, shadowing enemy fleets is tough to do, but I was under the impression that they already did this with transports, so it could be extended to include combat fleets as well.

As already noted, the AI already does this, and as already noted, having the AI shadow the player's fleets, while sounding nice in theory, is complicated in practice. A simple rules change, with a negative diplomatic modifier from the actual use of a surprise attack, is much more simple and doesn't open the way for further exploits.

It would be a simple solution, but it wouldn't cover enough cases or be a strong enough deterent. For example, the situation where the AI DOES attack in self defense (and they will do it, if you hang out transports in orbit long enough), having a diplomatic penalty is not ideal. Also, I'm not convinced it'd be enough... after all, the player is usually much, much more powerful after a successful war of conquest. Besides, thats not terribly believable, either... the most I'd think that a civilisation should do is react on more of a hair trigger to civs that employ surprise attacks.

And, like I said, I was under the impression that the AI would already shadow transports to some extent. If that's the case, extending it to attack fleets as well shouldn't be a huge difficulty.

The AIs desperately need to handle themselves better in the act of declaring and preparing for war. They are neglecting to prepare adequately before making a war declaration, denying themselves the advantages of a sneak attack, and often failing to make a crucial preemptive strike. When they do decide to shoot before they get shot, this can be exploited ass described by Wyndstar. The reason it's so easily exploited is because they recogninze the threat and declare war to counter... but without first moving ships into position or even attack on the turn they declare war, giving the player free reign to wipe them out before they will respond. Moving fleets into one turn striking distance of aggressors before declaring war and allowing attacks on the first turn would themselves go a long way towards remedying this, and a slightly more complex (I'm guessing, since the AI only figures transports into it) model for determining the risk of a catastrophic surprise attack would improve things as well.

Regardless, don't talk about an AI implementation being "too hard" to be done. Without access to the code, you can't make that judgement any more than someone can say that it is easy. Have you ever programmed an AI? If you have, you'll know that even a trivial improvement can take a lot of effort. Challenge of implementation is basically a non issue... its ASSUMED in AI development, and I can't imagine it getting simpler in a complex game like this.

Fact of the matter is, these are gaping holes in the AI's game, and they need to be fixed. While you can change the game rules to accomodate the AI, doing that too often and too dramatically waters down the game. I'd quite honestly prefer the AI to be improving its ability to prepare and initiate a war... areas where it's quite weak, which is why it is easy to exploit them there.

Successfully updated karma reason!

It would be a simple solution, but it wouldn't cover enough cases or be a strong enough deterent. For example, the situation where the AI DOES attack in self defense (and they will do it, if you hang out transports in orbit long enough), having a diplomatic penalty is not ideal. Also, I'm not convinced it'd be enough... after all, the player is usually much, much more powerful after a successful war of conquest. Besides, thats not terribly believable, either... the most I'd think that a civilisation should do is react on more of a hair trigger to civs that employ surprise attacks.

I think the diplomatic penalty would apply even when the AI is the one to declare war. After all, he can send around the surveillance information documenting your repeated, extensive and flagrantly aggresive violations of his sovereignty. If the ballon goes up and you end up taking out several of his planets in the first couple of weeks, it should obvious to all who was the real aggressor.

The other races will have every reason to fear your lightning assasination skills and therefore every reason to consider squashing you before they are the next target. To me, that's a pretty reasonable consequence of the sneak-attack tactic. No one is going to trust the guy who makes a habit of walking up to you with a smile then slitting your throat.

Successfully updated karma reason!

I would argue that it is a relatively simple solution, albiet one that would require tweaking to have optimal values for the variables in question. I'm aware I don't have access to the AI code, of course, so it could be more complicated than I'm guessing, but it seems like a fairly trivial kind of check.

Loop through all foreign vessels in your space, look to see what planets and resources they are within a single turn of, and base some kind of suspicion modifier off that. When the AI sees enough camped out ships and transports, there is a rapid decline in relations and possibly a diplomatic message. If the suspicion of an attack is high enough, they could even declare war outright without having to go through the decline in relations.

As for the AI moving ships on a per turn basis instead of setting waypoints... yeah, that might be more involved, but that should probably be fixed anyways, if the AIs are to become better able to tactically go toe to toe with the player.

Also, shadowing enemy fleets is tough to do, but I was under the impression that they already did this with transports, so it could be extended to include combat fleets as well.

As already noted, the AI already does this, and as already noted, having the AI shadow the player's fleets, while sounding nice in theory, is complicated in practice. A simple rules change, with a negative diplomatic modifier from the actual use of a surprise attack, is much more simple and doesn't open the way for further exploits.

It would be a simple solution, but it wouldn't cover enough cases or be a strong enough deterent. For example, the situation where the AI DOES attack in self defense (and they will do it, if you hang out transports in orbit long enough), having a diplomatic penalty is not ideal. Also, I'm not convinced it'd be enough... after all, the player is usually much, much more powerful after a successful war of conquest. Besides, thats not terribly believable, either... the most I'd think that a civilisation should do is react on more of a hair trigger to civs that employ surprise attacks.

And, like I said, I was under the impression that the AI would already shadow transports to some extent. If that's the case, extending it to attack fleets as well shouldn't be a huge difficulty.

The AIs desperately need to handle themselves better in the act of declaring and preparing for war. They are neglecting to prepare adequately before making a war declaration, denying themselves the advantages of a sneak attack, and often failing to make a crucial preemptive strike. When they do decide to shoot before they get shot, this can be exploited ass described by Wyndstar. The reason it's so easily exploited is because they recogninze the threat and declare war to counter... but without first moving ships into position or even attack on the turn they declare war, giving the player free reign to wipe them out before they will respond. Moving fleets into one turn striking distance of aggressors before declaring war and allowing attacks on the first turn would themselves go a long way towards remedying this, and a slightly more complex (I'm guessing, since the AI only figures transports into it) model for determining the risk of a catastrophic surprise attack would improve things as well.

Regardless, don't talk about an AI implementation being "too hard" to be done. Without access to the code, you can't make that judgement any more than someone can say that it is easy. Have you ever programmed an AI? If you have, you'll know that even a trivial improvement can take a lot of effort. Challenge of implementation is basically a non issue... its ASSUMED in AI development, and I can't imagine it getting simpler in a complex game like this.

Fact of the matter is, these are gaping holes in the AI's game, and they need to be fixed. While you can change the game rules to accomodate the AI, doing that too often and too dramatically waters down the game. I'd quite honestly prefer the AI to be improving its ability to prepare and initiate a war... areas where it's quite weak, which is why it is easy to exploit them there.

This is silly. Are you honestly saying that all areas within the AIs game of equally difficult to program? I don't need to be a programmer to recognize real complexity, and the tit for tat movements of ships sounds damn hard, or at the very least, ridiculousy resource intensive, versus something like maximizing planetary production. I believe Frogboy has made this difficulty known in the past, so stop passing off your wishful thinking as equally plausible. Care to name a game that does unit tactics at all well?

I would love for the AI to be more proficient in the tactical department, but there are such things as realistic expectations, and I would prefer to suggest specific, realistic solutions.

Successfully updated karma reason!

@ Entrophy Avatar

Fair point. In that sense, its not that bad an idea... but it's only part of the solution. If the AIs need one person to die off before they take the hint, it may already be too late!

@ Jasamcarl

No, I'm not saying that. All I was saying was that ALL AI is usually difficult, in my experience, to get it to do what you want, even on fairly trivial problem. All problems aren't equal, obviously.

And I'm not talking about a whole lot. What I'm suggesting is pretty simple, when you break it down, and each suggestion doesn't require the others to also be implemented.

1) System for guessing at risk, by figuring out how many of your planets are threatened at once. If there is a transport within striking distance of each planet, thats a big warning sign. If there are combat fleets also within striking distance, then you're probably going to be attacked.

For all I know, the AI may already be doing this.

2) Delay the declaration of war until you've moved fleets into striking distance of the threatening warships. This also would apply to AIs attacks. Don't declare war until you've got ships close enough to hit as many targets as possible in the first turn.

3) Allow the AI to attack on the first turn of war, making preemptive strikes actually effective. It has a one turn delay on its actions as of now, and that is a huge hole for players to exploit.

I'm not talking about an AI that is a tactical genius here. I've read the relevant posts, I'm aware of the complexity of the issue. What I'm talking about is better strategic planning, and removing what amounts to a tactical handicap on the AI.

Without access to the undoubtably complex AI code, I couldn't begin to guess at the complexity of the task, but it is much more trivial than implementing a chess player like AI for tactical combat.

Successfully updated karma reason!

Well I just wasted a couple hours playing a game on tough difficulty to realize that on turn 100ish, none of the 9 AIs have even built colony ships yet... They are just plying dumb or something... /sigh

Successfully updated karma reason!

I think the problem would be easier to remedy if planetary defenses are tougher than they are now. Offense/Defense relationships are heavily geared towards offense in the game. You CAN take out multiple planets in one turn all at once, with a couple fleets.

It would involve basic changes in game balancing, but I think that allowing orbital ships to act as a potential 8-ship fleet at all times, regardless of logistics, would go a long way towards strengthening the AIs position, without having to redo AI code. Right now it takes a Super Project to do that, and it's not really worth much as Supers go. Having it as a basic defensive mechanism would go a long way towards stabilizing galactic borders and wars.

Conceivably, this could be limited by logistics in the early game, so as to allow offensives, but the point at which attacking fleets rapidly overpower any defenses comes rather soon.

Strengthening planetary defenses to the point where they actually take time to overcome can, IMO, strengthen AI play significantly (and lessen the value of surprise attacks).

Alternatively, it's also possible to code other changes to the game. For example, you could modify the game such that military ships can only penetrate so far into enemy influenced-space. Limiting the range of military strikes makes battlefronts more influential and makes it a lot more difficult to take over multiple systems in one turn.

Successfully updated karma reason!

Alternatively, it's also possible to code other changes to the game. For example, you could modify the game such that military ships can only penetrate so far into enemy influenced-space. Limiting the range of military strikes makes battlefronts more influential and makes it a lot more difficult to take over multiple systems in one turn.

I don't know about using influence borders for that. However, life support *should* play a similar role. IMO, the 'free' life support range given for just having the tech should be very limited. If you don't put a life-support module on your ship, it should be pretty much limited to cruising from planet to planet in the same system.

Ships that are expected to attack deep into enemy territory will require a significant investment in life support. Between the requirements for life support and engines, a system defender ship could easily be twice as powerful as a ship designed to attack deep behind the lines.

Successfully updated karma reason!

I agree. Life support modules should be bulkier and more costly, and the passive bonus should go down. Otherwise, the components will remain as space fillers.

Thats a good point about planetary defense... it is a little easy to take planets from the AI. I think a bit of the problem might be that the AI doesn't get enough of the invasion techs at once, whereas the player is immediately going to back up his investment in planetary invasion with the planetary defense tech, orbital bombardment, space marines, and possibly tidal disruption. It's giving the player a huge advantage in most invasions early on.

Still, if your worlds are being invaded, you're already losing. Its always better to keep the fight away from your planets...

Successfully updated karma reason!

@ Entrophy Avatar

Fair point. In that sense, its not that bad an idea... but it's only part of the solution. If the AIs need one person to die off before they take the hint, it may already be too late!

@ Jasamcarl

No, I'm not saying that. All I was saying was that ALL AI is usually difficult, in my experience, to get it to do what you want, even on fairly trivial problem. All problems aren't equal, obviously.

And I'm not talking about a whole lot. What I'm suggesting is pretty simple, when you break it down, and each suggestion doesn't require the others to also be implemented.

1) System for guessing at risk, by figuring out how many of your planets are threatened at once. If there is a transport within striking distance of each planet, thats a big warning sign. If there are combat fleets also within striking distance, then you're probably going to be attacked.

For all I know, the AI may already be doing this.

2) Delay the declaration of war until you've moved fleets into striking distance of the threatening warships. This also would apply to AIs attacks. Don't declare war until you've got ships close enough to hit as many targets as possible in the first turn.

3) Allow the AI to attack on the first turn of war, making preemptive strikes actually effective. It has a one turn delay on its actions as of now, and that is a huge hole for players to exploit.

I'm not talking about an AI that is a tactical genius here. I've read the relevant posts, I'm aware of the complexity of the issue. What I'm talking about is better strategic planning, and removing what amounts to a tactical handicap on the AI.

Without access to the undoubtably complex AI code, I couldn't begin to guess at the complexity of the task, but it is much more trivial than implementing a chess player like AI for tactical combat.

Again, this is easier said than done, and, given that you are predicating some AI movement in predictable fashion on human movement, you will inevitably create tactical holes in the AI. The same is true for forcing the AI to distribute its ships heavily around one border, it would easily leave itself exposed to another assault, likely from the human player who is not affected by diplomatic wheelings and dealings to the same extent as the AI. Atleast as it currently stands, with the even distribution of AI ships, a player is likely to encounter atleast some resistance. Even shadow movement for defensive play will be open to exploits, or at the very least with require mad computations, making it technically less feasible.

I'm sorry, but I think there is still wishful thinking going on here.

Successfully updated karma reason!