Rome 2 AI fail again......

[19891225]

someone told me that Rome 2's general system is for limiting the AI's ability to stack small forces like previous total war game.......





I guess this doesn't work very well, huh?

It's 13 turn only, Rome only control Italy and they send 5 legions down to Sicily, I guess they are trying to take Syracuse which I just recently conquered. But by sending these armies to Sicily, what is left to defend Rome? Nothing...... I guess the Etruscan league can just walk into the Roman Senate with a wooden sword.......

Btw, Rome declare war on me in the very first turn, and we don't even have a sharing boundary at that time.........the AI's war target is so straight in Rome 2, I got declared war by the Spartan once and I am not even close to them nor a friend with their enemy.....

[Huberto]

Indeed . Bewildering CAI in TWR2, isn't it? Wait until next turn -- they will all move offshore.

[BigHairyMelon]

I've had the AI declare war on me and then park all of their armies in the middle of the sea so they just sit there are lose men every turn due to attrition while I'm free to sack all their cities. They stayed there for the entire duration of the war. Unbelievable.

[Huberto]

I've had the AI declare war on me and then park all of their armies in the middle of the sea so they just sit there are lose men every turn due to attrition while I'm free to sack all their cities. They stayed there for the entire duration of the war. Unbelievable.

All too true. This is a common occurance and once you take a few regions it's almost impossible to get the AI to declare war or attack you. Even more frustrating that they won't defend themselves.

[BigHairyMelon]

All too true. This is a common occurance and once you take a few regions it's almost impossible to get the AI to declare war or attack you. Even more frustrating that they won't defend themselves.

When I start getting big I usually get some vassals and military allies, I don't like painting the map my colour, I just take my choice provinces. The AI will attack my vassals/allies always which leads to some fun wars. The war in my original post was when Macedon declared war on Rome ( who I (The Arverni) liberated ) and thus me and all of my allies, pretty much everything west of Italy, fought the Greeks. Would have been interesting had Macedon not embarked on a trip to the sea.

[RedGuard]

well you know these bugs are really hard for a developer to find...

[Khan1281]

Are you still seeing the armies taking cruises all over? I used to see it but havent in a few patches. It is very stupid and something I was happy to see fixed, at least in my games.

[dardy]

Btw, Rome declare war on me in the very first turn, and we don't even have a sharing boundary at that time.........the AI's war target is so straight in Rome 2, I got declared war by the Spartan once and I am not even close to them nor a friend with their enemy.....

I'm calling bull the AI never declares war.

[RedGuard]

I'm calling bull the AI never declares war.

they do if they sense weakness, you have to be smaller in power than them or equal

[|Sith|Galvanized Iron]

They should just return to the M1TW 2D map already IMHO. At least the AI did massive invasions back then.

[Dago Red]

On the OP's post, you're playing as Carthage. Rome already hates you and vice versa so there's no mystery as to why you're at war. Declaring war has less to do with sharing borders than it does with all the other factors which are shown in the diplomacy window. If you want more reasonable diplomacy play the overhaul mod: Magnar.

As for the stacks, yeah. Broken system.

[AnonMilwaukean]

Rome 2 still has its bugs, but I primarily see full-stacks when the AI does decide to attack. The only time this doesn't hold up is when they assume they can overwhelm my garrison at a fortified city simply because they outnumber the garrison. In that case, they attack because auto-resolve calculates they should almost be guaranteed to win (auto-resolve seems to neglect factors like boiling oil, and the siege AI still performs somewhat badly now that units can't just climb walls) only to be defeated almost every time.

Now, seeing partial armies move around isn't entirely uncommon. I often use a partial army to bring reinforcements up to my primary combat armies. So, in some cases, seeing a couple small stacks doesn't mean the AI intends to never fill them, as they may be moving to reinforce another army.

All that said, I still occasionally do see a bunch of small stacks with no real transferring of troops. Even this is sometimes explainable, however: when I spot a faction that once had many provinces and has lost most of them, it seems entirely plausible that the AI has small stacks simply because it is trying to fill them and lacks the funds it once had to refill them rapidly.

Krisslanza is correct on this: there are many factors as to why the AI has small stacks rather than 20-stacks. There are economic issues (is the AI refilling those stacks, but just doing so very slowly because of lack of funds?); there are logistical issues (is the AI trying to just bring up fresh units to another army?); there are strategic issues (is the AI trying to hunt down lots of small stacks of rebels or slaves in its provinces?); and there are issues of competing priorities (the AI wants to protect or attack a number of settlements at once, and can't compute which should take priority). It's not like one bit of code is just poorly programmed, here: I assume there's a lot of calculations going on, calculations that we might not even consider the AI to be making.

[pajomife]

Rome 2 still has its bugs, but I primarily see full-stacks when the AI does decide to attack. The only time this doesn't hold up is when they assume they can overwhelm my garrison at a fortified city simply because they outnumber the garrison. In that case, they attack because auto-resolve calculates they should almost be guaranteed to win (auto-resolve seems to neglect factors like boiling oil, and the siege AI still performs somewhat badly now that units can't just climb walls) only to be defeated almost every time.

Now, seeing partial armies move around isn't entirely uncommon. I often use a partial army to bring reinforcements up to my primary combat armies. So, in some cases, seeing a couple small stacks doesn't mean the AI intends to never fill them, as they may be moving to reinforce another army.

All that said, I still occasionally do see a bunch of small stacks with no real transferring of troops. Even this is sometimes explainable, however: when I spot a faction that once had many provinces and has lost most of them, it seems entirely plausible that the AI has small stacks simply because it is trying to fill them and lacks the funds it once had to refill them rapidly.

Krisslanza is correct on this: there are many factors as to why the AI has small stacks rather than 20-stacks. There are economic issues (is the AI refilling those stacks, but just doing so very slowly because of lack of funds?); there are logistical issues (is the AI trying to just bring up fresh units to another army?); there are strategic issues (is the AI trying to hunt down lots of small stacks of rebels or slaves in its provinces?); and there are issues of competing priorities (the AI wants to protect or attack a number of settlements at once, and can't compute which should take priority). It's not like one bit of code is just poorly programmed, here: I assume there's a lot of calculations going on, calculations that we might not even consider the AI to be making.

Why you mention Krisslanza,both work in CA? This AI is miserable dont try to branch it with futile and empty argumentations,the AI guys not do a good job,and the lead designer,those guy who implement the transport ship system, is the major culprit for this game be a disgrace.Do not get tired in the forum,with pointless arguments, just do your work.

[Huberto]

All that said, I still occasionally do see a bunch of small stacks with no real transferring of troops. Even this is sometimes explainable, however: when I spot a faction that once had many provinces and has lost most of them, it seems entirely plausible that the AI has small stacks simply because it is trying to fill them and lacks the funds it once had to refill them rapidly.

Krisslanza is correct on this: there are many factors as to why the AI has small stacks rather than 20-stacks. There are economic issues (is the AI refilling those stacks, but just doing so very slowly because of lack of funds?); there are logistical issues (is the AI trying to just bring up fresh units to another army?); there are strategic issues (is the AI trying to hunt down lots of small stacks of rebels or slaves in its provinces?); and there are issues of competing priorities (the AI wants to protect or attack a number of settlements at once, and can't compute which should take priority). It's not like one bit of code is just poorly programmed, here: I assume there's a lot of calculations going on, calculations that we might not even consider the AI to be making.

I think you are giving the TWR2 programmers too much credit. Yes there are calculations that the CAI is making. But it isn't working. It's generating many small stacks, often floating at sea, stacks that do not attack and do not defend. It doesn't seem to be able to do many of the basic things that CAI's in previous Warscape games were capable of doing: fielding full armies and the offensive against the player.

I am not an AI expert by any means but it strikes me that many of the new features introduced in TWR2 such as stances, limited armies requiring generals, insta-transports and land-sea battles have completely messed up AI campaign map behavior. It's playable but uninteresting.

[AnonMilwaukean]

It's generating many small stacks, often floating at sea, stacks that do not attack and do not defend.

That is the one issue I have a hard time understanding, and that I tend to agree with other posts on: while there are plausible explanations for the small-stacks, I don't understand why I see them occasionally floating in the middle of the sea.

I mean, I see that behavior often when an AI faction no longer owns any settlements whatsoever, and in that case it makes some sense, as floating at sea at least keeps them in-game for another couple of turns (often long enough for their agents to still try and sabotage my own settlements). In this case, the AI is remarkably human-like in basically trying to play spoiler to me in some revenge-like manner, forcing me to hunt down their last two units to stop the agent spam.

But in the handful of times I see small-stacks floating someplace while I know that that faction has a settlement remaining, then yes, I agree strongly that there is no reason to have the small-stacks floating adrift. That's the one situation where I don't know what possible reason there is for the AI to prioritize not only having small stacks, but to prioritize putting them in a position wherein they can neither effectively defend nor attack.

[krisslanza]

That is the one issue I have a hard time understanding, and that I tend to agree with other posts on: while there are plausible explanations for the small-stacks, I don't understand why I see them occasionally floating in the middle of the sea.

I mean, I see that behavior often when an AI faction no longer owns any settlements whatsoever, and in that case it makes some sense, as floating at sea at least keeps them in-game for another couple of turns (often long enough for their agents to still try and sabotage my own settlements). In this case, the AI is remarkably human-like in basically trying to play spoiler to me in some revenge-like manner, forcing me to hunt down their last two units to stop the agent spam.

But in the handful of times I see small-stacks floating someplace while I know that that faction has a settlement remaining, then yes, I agree strongly that there is no reason to have the small-stacks floating adrift. That's the one situation where I don't know what possible reason there is for the AI to prioritize not only having small stacks, but to prioritize putting them in a position wherein they can neither effectively defend nor attack.

The fleets stuck at sea also baffle me. People talk about them a lot on this forum, like they're commonplace, but playing on patch 9 so far I've never actually seen it.

Carthage and Libya have a healthy number of fleets in the sea at the moment, yes, but they've been using them constantly keep pressure on my borders in Magna Gracea and Corscina. They've even managed to slip some of them close enough to actually attack, and take settlements before I can stop them. I did used to see the random fleets at sea prior patch 9, but so far I'm not seeing them just sitting there for no reason. I just can't imagine what is making the AI do it though, other then it perhaps cannot tell the "difference" between land and sea.

But even with that, it doesn't explain why it'd dump all their armies into the ocean either. And if we can't think of any reasons, I cannot imagine CA is going to have much more luck trying to figure out why the AI keeps doing that. Heck, it could be something entirely unrelated to CAI 'pathfinding' that is causing it, who knows? If code worked logically and as it should all the time, we wouldn't even have bugs in the first place.

[RedGuard]

It really is baffling that this is once again another feature that was supposedly supposed to revoultionize total war, but once again its just another fail on CA's part. Its a lot more fun going back and playing M2 and realizing you actually have to build up a fleet before you can go gallivanting across the ocean, that's probably also the reason the ai is half retarded and goes on cruises all the time because there is no thinking to attain it.

[Tullaris]

Rome 2 still has its bugs, but I primarily see full-stacks when the AI does decide to attack. The only time this doesn't hold up is when they assume they can overwhelm my garrison at a fortified city simply because they outnumber the garrison. In that case, they attack because auto-resolve calculates they should almost be guaranteed to win (auto-resolve seems to neglect factors like boiling oil, and the siege AI still performs somewhat badly now that units can't just climb walls) only to be defeated almost every time.

Now, seeing partial armies move around isn't entirely uncommon. I often use a partial army to bring reinforcements up to my primary combat armies. So, in some cases, seeing a couple small stacks doesn't mean the AI intends to never fill them, as they may be moving to reinforce another army.

All that said, I still occasionally do see a bunch of small stacks with no real transferring of troops. Even this is sometimes explainable, however: when I spot a faction that once had many provinces and has lost most of them, it seems entirely plausible that the AI has small stacks simply because it is trying to fill them and lacks the funds it once had to refill them rapidly.

Krisslanza is correct on this: there are many factors as to why the AI has small stacks rather than 20-stacks. There are economic issues (is the AI refilling those stacks, but just doing so very slowly because of lack of funds?); there are logistical issues (is the AI trying to just bring up fresh units to another army?); there are strategic issues (is the AI trying to hunt down lots of small stacks of rebels or slaves in its provinces?); and there are issues of competing priorities (the AI wants to protect or attack a number of settlements at once, and can't compute which should take priority). It's not like one bit of code is just poorly programmed, here: I assume there's a lot of calculations going on, calculations that we might not even consider the AI to be making.

Ah 'the AI is so smart, we just don't get it' argument. Campaign AI has always been a neglected part of TW games. This is because investing a lot in AI does little for marketing and reviews. Reviewers don't play the game long enough to notice a botched campaign AI and in terms of marketing you can say you have the most sophisticated AI but it 'doesn't work' because of bugs.

Also you can only start the work on campaign AI when most of the other stuff has been done.

However, for us TW gamers, campaign AI is the factor that adds the most replayability. A good campaign AI provides a challenge. For me the lack of a real challenge is the biggest problem Rome II has. Because of this lack of challenge the game is pointless and no fun for me.

If you think about what makes a good campaign ai? Campaign AI is merely a decision tree, a script so to say. If you want to provide the most challenging game you will need to balance the complexity with your scripting capacity. For example, because of the 2d maps Shogun 1 and Medieval 1 had relatively effective campaign AI's. They were simple and provided a challenge. The move to 3d maps made the campaign AI harder to script. It works with waypoints now and prioritizing targets according to aforementioned decision tree.

Now think about what we want from a Rome II campaign AI. I want it to seek peace with stronger nations that threaten its borders, unless that neighbour is already at war. I want it to declare war on weaker nations. I want it to be relentlessy launching full stack armies at the nations it is at war with, prioritizing those nearby. I want to keep armies for defense depending on how many enemies threathen it. These different considerations should decide where the AI moves its armies.

Looking at what we have at the moment in Rome II, you can confidently say that it is doing a poor job at balancing these considerations. If the decisiontree (script) is bad or not or whether there are bugs is not an issue here. What is an issue is that again CA sells the game without a campaign AI with a minimum of effectivity. This makes the game boring. Pretty, but boring, which is a shame.

[AnonMilwaukean]

Ah 'the AI is so smart, we just don't get it' argument.

What I've been saying is that fixing the AI is not as simple as just changing a few pieces of a line of code or two. My argument has been that the coding of the AI is more complex than many are assuming, which is not the same thing as saying that the AI itself is so smart that we don't get it. Again: my argument is that AI coding is more complex and multi-variable-based than is being claimed, which in fact causes the AI itself to fail in decision making and execution more often than most human players would.

[Huberto]

What I've been saying is that fixing the AI is not as simple as just changing a few pieces of a line of code or two. My argument has been that the coding of the AI is more complex than many are assuming, which is not the same thing as saying that the AI itself is so smart that we don't get it. Again: my argument is that AI coding is more complex and multi-variable-based than is being claimed, which in fact causes the AI itself to fail in decision making and execution more often than most human players would.

And what I am saying is that CA/SEGA should be blamed, and blamed harshly here. There is and was no reason to take such large steps backwards in CAI coding, backwards from NTW and S2 to the point where the game does not provide a challenge and the AI's campaign map behavior is often baffling.

From my perspective the CAI is the single biggest, game-breaking flaw in TWR2. That it still remains so six months after release speaks volumes to me about the company's commitment to making quality games.