Something's wrong: campaign AI and dodgy battles in Rise of the Samurai

[frogbeastegg]

EDIT: This issue is now fixed. There was a small file missing from my original install of RotS. After getting steam to verify the game cache and replace that file, the AI has transformed from senile ditherer to the ferocious daimyo we all know and love. Phew!



What are people thinking of the AI in the RotS campaign? I'm very sorry to say that I am finding it to be a massive step backwards. It's awful, so much of what made me love Shogun II's AI -and thus S2 -is now missing in action. This is ETW wearing a kimono. S2's easy mode is harder than this, and its very hard was more logical in clan behaviour!

I'm seeing loads of nonsensical declarations of war. Clans which are on the far end of the map declaring and then not doing anything at all. Friendly clans rated at honourable and trustworthy, which I have trade routes and good relations with, and which are 100% under my family's influence are also abruptly declaring war. My sister clan prefers to support tiny minors over me. Meanwhile the people who hate me, or sensibly placed to wage war on me, are peaceful and happily signing trade agreements. I'm not a large force, I've not been conquering large areas in a short amount of time, and I have been paying attention to keeping my friends. In S2's campaign I would not be seeing this.

The diplomatic AI behaves as if there is a pre-built toggle tied to turn limit. For the first 25 or so turns everywhere was peace-peace-peace. It felt like the AI had gone to sleep, as few provinces were changing hands and only a couple of minor clans were destroyed. I saved my game and quit for the night. On loading today it's non-stop war-war-war without reason, logic, or cause. On turn 37 I quit.

One of the clans which declared war on me immediately turned around and offered peace the next turn. Another clan did the same thing later, and re-declared the turn after that.

The campaign army AI is much worse as well. It's shuffling armies back and forth, back and forth in a never-ending conga. I've witnessed one army running back and forth burning two little locations of the map turn after turn, year after year. Meanwhile the massive garrison in the nearby castle watches while sipping tea. When the AI declares war on me it basically does nothing. It ignores my unguarded castles. It struggles to capture any enemy castle. It sails its ships past my unguarded trade fleets.

I am also finding the battles to be a significant step backwards. On very hard I am waltzing through them as if I am on easy with a set of stat and morale boosts for my men. I'm getting legendary victory after legendary victory, dicing my way through much larger forces, and absolutely destroying evenly matched ones. I can't credit it to my stunning tactics either as I hardly know what most of these units do and am basically just standing there in my deployment position, letting the AI come to me. I loose a few volleys with my archers, let them automatically skirmish out of the way, let the AI charge into my naginata levy line, and then flank with my pair of swordsmen. Every single time, easy victory with manageable losses. I'm not using superior units either. With the bonuses the AI's army should be better than mine. I should be taking a heavy mauling at best and being routed at worst.

I defended a level 2 castle (fort?) once against a vastly superior force. I completely destroyed the enemy with only a handful of men making it onto my ramparts. I issued 3 orders, one each to a melee unit requesting that it move up to stand closer to the ramparts in case the climbing enemy made it inside. I didn't move the rest of my melee force, and let my archers fire at will. Based on my siege experiences in S2, I was expecting a tough battle, perhaps even defeat, based on the armies and fortification level involved.

While it's able to advance in good formation, the battle AI has a single tactic: charge! It barely tries to flank. It doesn't use missile superiority. It doesn't try to lure me out. It doesn't hide. It won't make me come to it.

I actually went into the menu to check that I hadn't chosen easy instead of very hard by accident.



I am close to heartbroken. I loved Shogun II's AI because it was smart, dangerous, and mostly sane. It helped make Shogun II one of the best strategy games I have played in my ~15 years of gaming. Sure the AI made some bad moves or did silly things now and then, but far, far less often than most strategy games. Heh, it probably made fewer dumb mistakes than me! And I loved S2's battles - it was easy to lose with bad tactics or inattention. I had high hopes for the RotS campaign, I've been anticipating it since the day I first heard about it. It should have been a guaranteed hit. While some of the ideas and changes are interesting, without the capable AI and meaningful battles the game is empty to me. It's not worth playing because there is nothing to play against.

This is not the S2 I know and love. This is a shambling zombie. What's gone wrong?

[Hamata]

I think it's fine i like the campaign mechanics etc

[econ21]

This is not the S2 I know and love. This is a shambling zombie. What's gone wrong?

Oh crumbs. Now I hurriedly take back my desire to have the RotS changes (e.g. the campaign map) retrofitted to the original campaign. I hope it has survived in tact, as it was - as you say - a formiddable challenge on higher difficulties. One I confess I have hitherto largely ducked.

[Marshall Louis-Nicolas Davout]

I'm lucky I didn't buy it first. When it goes to March 2013,maybe I'll buy it then.By then it should be fixed.Of course,the campagin shoulnd't affect SH2TW.You can still play the orginal campagin as well.

[Monk]

I've noticed some really troubling things the same as you have froggy.

Enemy armies actively avoiding cities i've left unguarded, odd troop movement, passive AI... It's very strange and has made getting a handle on the new DLC decidly uninteresting. I can't deny the new map looks great, but everything else.. :no:

[Madae]

All the problems I have with the game aside, I've only noticed a few things that make me scratch my head. AI agents have a tendency to get stuck and stop moving/acting, and I've also noticed small armies doing that too (but maybe that's some kind of new tactic - stand in my territory an inch or two away from my castle for a few turns).

The only thing that has changed with the battles as far as I can tell is what units you field and, coincidentally, what kind of tactics you'll be using with said new units. A lot of my best troops are archers and archers/melee, so my tactic has always been pretty straightforward - shoot until they get close or die. AI still runs for the high ground, tries to engage in said archer-duels, and as far as I can tell, does it's best in sieges, though my overwhelming and experienced forces will have no problem in defeating a few levies. AI seems a lot more rough than in the GC, but not really that much worse off.

Diplomacy I actually don't have anything wrong with... Hell, it's about the only thing in ROTS that I actually really like. My Daimyo has 7 honor. My allies love me and my enemies respect me (I even got a global message that all of the clans had sincere respect for me, regardless of our disagreements). I have not once had someone declare war on me that seemed completely unwarranted or otherwise unintelligent. I'm about halfway through my game, and my allies have stuck with me every bit of the way, regardless of whether or not I've helped them (my honor seems to be making up for my lack of being a helpful ally). These guys have all given me units as a show of faith in my ability. I guess I could probably go on for awhile with the things I like about it, but I'll just say that I haven't researched any diplomacy stuff, and I've been doing great. Also, the mid game is a lot different than the early game. I was at war and taking a beating, and was getting attacked at inopportune moments by differently aligned, and angry, clans. I almost quit at one point, but I fell back, suffered my losses and then regained them when I could.

At the current stage; Ochi attacked my sister clan who has been taking a beating most of the game. I was just about finished with Fujiwara when this happened (had them both down to their last provinces), so I hurried up and took everything in the north and moved back down south, which was slow going for my main army. Meanwhile, I built up a new army and moved them in on Ochi. The first province I took (Uesugi's home province) had been getting converted by my agent for about the last 20-30 turns, so it was ripe. I immediately turned it with my agent and had a good defensible position between that province and the one below it that I had already owned (these were the only two ways in for Ochi). Ochi's big army stack in Uesugi's province that had been slowly trying to backdoor through the forest to the province next to it in the east (which is seriously the first time I've seen the AI try to do that, but that was pretty much the only way for him other than ship to get there - I owned the province next to it that had the road-way in) was now sandwiched between the big forest and my army, so I took it out and that was that. They immediately attempted for peace after that, but I can't really do that considering how they've been treating my sister ally. This is where I stopped playing last night.

In closing, I'm still torn on whether or not I like this as much as the GC, but it's Shogun, and Total War, so it's still fun in its own right. The things I've seen that are wrong with the game are just small bugs that can easily be fixed. To me, everything the AI does is relatively the same as the GC, with minor differences because of agents and unit types.

[frogbeastegg]

Which difficulty level are you playing on, Madae?

I have been doing a bit of reading over lunch today, and I've noticed the beginnings of a trend across the three TW sites. Most of the people talking about the AI are playing on normal or easy. The descriptions they give are pretty much the same as yours, in that it describes an AI similar to that seen in the original game but with a few minor hiccoughs. The few people who are reporting problems like the ones I am seeing are all playing on very hard. I begin to wonder if something is bugged on that level.

If I get chance to play this evening I'm going to purge all of my savegames, get steam to verify the game files are all undamaged, and start a new game on a different difficulty. See what happens.

[Madae]

I usually play on Hard, though I don't remember if I set it at that or not (I'm not at home at the moment). I can understand the AI being quicker to take advantage of a situation or attempt to do something that may seem like a dumb idea in higher difficulties - I have played on Very Hard, but I rarely play Legendary - but I don't consider that a problem with the AI. The AI will try to win, regardless if the situation is hopeless for one clan or another - you really never know what can happen and it may turn out good for them. Ultimately, it makes more sense for them to throw themselves at you then to just sit back because you're too strong. They simply won't win if they do that. Even then, it's more difficult when you start having wars pop up everywhere, because that subtracts from potential trading partners and makes it more difficult to defend your territory. I like to think of it as the AI using weak clans as fodder to take away your attention so a stronger clan can make their move(s). Logic doesn't necessarily factor in to that - you'd like to think they're separate entities, but they're really not.

I say give it a shot again. Maybe you'll see some different results. You'll never really know what goes on in the AI's "head". There might have been perfectly good reasons for one to declare war on you thinking it was a good idea, just to retract it because someone more threatening stepped up (this also subtracted from potential trading partners, and you never know when someone will land an invansion army).

The problems with AI charging in battles is also something I commented in another thread - you basically only have a few types of units in this game; you have infantry, you have archers, you have archer/infantry, and then bow cavalry. Variances in tactics, I think, will be because, without a variety of cavalry, the slow moving nature of infantry and the prevalence of archers, you won't see much flanking trying to take place, because most battles I'm in are decided before they even get near me, and I think this has a lot to do with some units just being simply overpowered (foot samurai).

I've also noticed that, typically, allies are harder to control in harder difficulties - having just a high positive relationship is rarely enough. Most of the time, wars will start and I usually think its because I'm not a very good ally (which is mostly the truth with some clans). Others could simply be that they're taking advantage of the situation. Besides, in the GC, everyone wants to be Shogun.

My problem with higher difficulties is that it's not harder because the AI is "smarter", it's just that they get more bonuses to their units in battles, which isn't really fair. Sure, it's hard, but it's hard because they, technically, cheat. In a perfect world, they would just be smarter, and difficulty would be rated not by a generic difficulty setting (easy/normal/hard/whatever), and more like in Galactic Civilizations, where it was based on intelligence of the AI (from dumb, to smart, to godlike). Stat bonuses factored in to it somewhat, but it was more or less how well they managed their economy, how aggressive they were in capturing territory or taking advantage of weak races, and other things. A fair bit of luck also factored into it to, however, in concerns with how they dealt with other races, and how lucky they were to get a few good core planets under their control early in the game (maps are randomized).

[frogbeastegg]

I spent around 80 hours testing the game out on the four main difficulties while researching the guide, with around 50 hours of that time on hard or above. Plus the many hours spent playing for enjoyment. I became very familiar with how the battle and campaign AI behaves on the various levels, and what kind of things provoke what result. There are lots of little rules which together add up into a semblance of logic, rules which are applied consistently across the board wherever an applicable situation arises. These little rules are a core part of what makes the AI work. Many of them were broken in my RotS experience. If I gave the sengoku AI an army with a similar infantry/archer composition to the ones I faced in RotS I know it would try to shoot me, that it would attempt to flank on at least one side, and that I may end up having to go after it instead of having it come to me. If I gave it a similar army which outnumbered me by more than 600 men I know I'd be swamped from all sides. If I added a few superior melee units in that mix I know I would see them thrown against the weakest part of my line. There would also be a decent chance it would have hidden units in the available forests in most of my RotS battles. If a distant clan declares war on me in the sengoku game I know to expect a naval invasion within 4 turns, probably less. If I leave a castle exposed and a hostile AI is nearby it cannot resist attacking it. And so on through scores of examples. It's an old, respected friend.

I haven't found any difficulty controlling my allies when playing on hard or very hard. I've had clans stick with me for an entire campaign, through realm divide and beyond. This is the first game in the series where diplomacy actually works.

GalCiv2, now there's a game! Not one of my top favourites any more, but definitely up there. That's primarily due to the AI. Yes, I too dislike cheating AI. Civilisation's higher difficulties disgust me because it's nothing but a pile of cheats piled on top of the normal level AI. The cheating in prior TW games used to annoy me; S2's toned it down quite a bit.

Steam just finished validating my install. It downloaded a 23MB file so fingers crossed I will now see an improvement. Time to try it out ...

[Madae]

I spent around 80 hours testing the game out on the four main difficulties while researching the guide, with around 50 hours of that time on hard or above. Plus the many hours spent playing for enjoyment. I became very familiar with how the battle and campaign AI behaves on the various levels, and what kind of things provoke what result. There are lots of little rules which together add up into a semblance of logic, rules which are applied consistently across the board wherever an applicable situation arises. These little rules are a core part of what makes the AI work. Many of them were broken in my RotS experience. If I gave the sengoku AI an army with a similar infantry/archer composition to the ones I faced in RotS I know it would try to shoot me, that it would attempt to flank on at least one side, and that I may end up having to go after it instead of having it come to me. If I gave it a similar army which outnumbered me by more than 600 men I know I'd be swamped from all sides. If I added a few superior melee units in that mix I know I would see them thrown against the weakest part of my line. There would also be a decent chance it would have hidden units in the available forests in most of my RotS battles. If a distant clan declares war on me in the sengoku game I know to expect a naval invasion within 4 turns, probably less. If I leave a castle exposed and a hostile AI is nearby it cannot resist attacking it. And so on through scores of examples. It's an old, respected friend.

I would argue this, but since I don't have a history here or a following, nor anything to my name that would make anyone believe me over you, I'm just going to have to settle with "I don't agree" - it'd mostly be an argument about who has played the game more. You can argue that it feels wrong to you all you want, but maybe you're expecting to much with all the changes the game has had. I went through this same scenario when the game first launched - everyone had problems with it... I didn't, so I can't relate at all.


The cheating in prior TW games used to annoy me; S2's toned it down quite a bit.

It's still cheating. Which, again, is fine for some people, but not for me. I'd prefer they actually be challenging, and not handicapped one way or another.

[frogbeastegg]

You are missing my point ~:)

Based on the descriptions you have given, your AI is still functioning in a respectable manner. I'm not suggesting that you are imagining that. I'm trying to pin down what is wrong with my game. Maybe there's a bug with a certain setting, maybe that 23MB replacement file will fix something, maybe it's something else entirely. If I don't try to find out then I have no chance of a solution. RotS is not attractive to play when I win battles by leaving my levy-grade (I haven't used anything above the occasional attendant grade unit) army to its own devices, and when the campaign AI is a barely-there bunch of random actions.

The only difficulty I had with S2 was that I expected it to be dumb and it wasn't. Three turns into my first game I learned that, started again, and never looked back. If I tried most of the stuff I've been able to get away with in RotS in the main game I'd be crushed in no time.

[Nowake]

My problem with higher difficulties is that it's not harder because the AI is "smarter", it's just that they get more bonuses to their units in battles, which isn't really fair. Sure, it's hard, but it's hard because they, technically, cheat.
Are you chaps sure about that? I've never had any experience to indicate it rides on a pillion in any of my games. I would rather suspect the AI is simply limited on lower difficulties. Now, I will mention as I always do that I never played on anything but legendary difficulty etc. etc., however, even on that, I've never had units with the same stats behave worse for me than for the AI. What I did experience was different AI behavior depending on factors like a general's rank.

To give you a basic, yet what I would call classic example: I have a "trick" to help my first Heroic Victory come faster in the beginning, as I need the bonus to Honor. Now, what I face in those first two-three years are armies of Yari Ashigaru and Bow Ashigaru, same as mine (well, I only use YA at that point actually). The AI, having generals no more evolved than mine (one-two stars), will almost always sort its defensive line by placing his Yari Ashigaru on Spear Wall. The problem for it arises from the fact that Spear Wall actually tires one's troops over time. That means I wait for half the time to expire (sixty minutes time limit) before proceeding with my charge -- its Very Tired Yari Ashigaru are never able to put up a decent fight anymore of course. However, that never happens in my battles against rank three or four generals. Same units at their disposal, same line of battle, same Spear Wall -- once a unit becomes only Winded under Spear Wall, it is immediately taken out of that formation and told to stand easy so to speak. And of course, it's alacrity improves significantly when maneuvering etc.

Now, this is just one instance where it is quite clear how the AI is simply limited to properly show differences in behavior. There are many, many more, even in the way high ranked generals behave on the campaign map, but generally it's more of your insight telling you the AI would have proceeded otherwise if its general would've been less apt, because you can't really see the exact same situation unfold twice in the exact same game in the exact same spot.

There are "bonuses" as you say to the AI, yet those are economic and diplomatic as far as I saw: clearly, the AI armies and fleets, while not spawning as some declare, are created and sustained by infusions of koku every turn (though even in that case, not as high as one would think, the AI uses the highest level of taxation aswell -- and pays the price in revolts of course -- which does increase its income significantly for enough time to allow it to create its huge army corps). Moreover, its recruitment times are lower also, being able to recruit Samurai in just one turn right from the beginning. Also, it has more recruitment slots. E.g. 1547, Hojo retreats with just a general in Izu's undefended city; two seasons later it had recruited -- while under the direct observation of my first ninja -- six units of Katana Samurai. Now, Izu, being its capital (Stronghold level), should only have allowed it two recruitment slots, and of course, by 1547, it was impossible for its Katana Samurai to benefit from the reduced recruitment times tech. So it had one recruitment slot and minus one season recruitment times on me. Plus, the funds to recruit six Katana Samurai units while having only one province after it had payed me off for peace with five thousand koku. As to diplomatic relations e.g. in my campaign as Date, Takeda ended up with a sprawling empire of twenty seven provinces before I began my campaign against it. And while R.D. hit for me on over eighteeen provinces (campaign set to Long) Takeda certainly had no R.D issue at all.

So, indeed, it has some ancillary perks which allow it to create and field armies faster and of better quality, yet I never found it "cheating", it still has to go through all the steps the player does, its logic is limpid and its units won't win in a perfect one versus one "just cause".

[Madae]

Are you chaps sure about that? I've never had any experience to indicate it rides on a pillion in any of my games. I would rather suspect the AI is simply limited on lower difficulties. Now, I will mention as I always do that I never played on anything but legendary difficulty etc. etc., however, even on that, I've never had units with the same stats behave worse for me than for the AI. What I did experience was different AI behavior depending on factors like a general's rank.

As far as I know, and which was confirmed on some level by CA themselves, was that the AI, on harder difficulties, get bonuses to morale and other unit properties (I can't find the original post on this). This came with that whole issue where people were thinking the AI were "spawning" armies at inopportune moments, which it does on some levels (such as when a clan loses all of its family members, it spawns a new family member, or pirates that randomly spawn around Japan), but not in the way most people were thinking (AI spawned a full stack of troops that were conveniently hidden in trees to make it look like they were hiding, which is not true).

At any rate, I can't remember all the specifics, but morale was definitely one of them, and they also get bonuses/penalties on recruitment depending on the difficulty (extra slots/reduced slots, increased time/reduced time). The AI is also more prone to "mistakes" and "miss opportunities" on easier difficulties, where on H/VH, they will take full advantage of any opening they see and act accordingly to their strength, but also depending on the difficulty setting.

Technically, there isn't a "even-ground" difficulty where both the player and the AI are the same. Easy AI receives a heavy penalty, and Normal receives only a slight penalty. From there, on Hard they receive a slight bonus, and Very Hard they receive a bigger bonus. Legendary takes the Very Hard bonuses and slaps on other restrictions unique to that difficulty (like no battle map, and limited vision).

This may, of course, have changed in ROTS, but I doubt it. On the whole, the AI is more passive on Easy/Normal, and more active on Hard/Very Hard, and diplomacy, alliances and vassals are a lot harder to sustain the higher you go - but also, like frog said, it is still possible, it just requires a lot more work and effort on your part to keep them from betraying you. It is still rather straight-forward; the more you interact with and help your ally through various means, the more likely they will stay with you. Ignore them for but a moment, and you and start building up that "useless ally" penalty, which can easily turn the tables the harder you go.

However you look at it; giving someone something that you don't get is cheating. Taking something away from them while leaving you untouched is handicapping. Either way, the AI is getting screwed or pumped, there is no in-between.

[frogbeastegg]

Are you chaps sure about that?
I asked CA for some info about this when I was writing the guide, and also gathered up everything I could find posted by them on the three forums. Chopping the guide's text into a breakdown minus the chatty stuff:


Easy.

The AI is quite generous towards the player at this level. It seems more likely to direct its aggression at other computer controlled clans, although that is not to say it ignores the player altogether. It will deliberately make mistakes, or fail to follow up on opportunities.

The AI has limitations at this level. Its production and building priorities are handicapped. On the battlefield the AI has limitations to its accuracy, morale and melee attack value, and the player has bonuses to them.

Normal

On normal, the campaign AI has a small penalty to its production and building priorities. It seems quite balanced in its attitude towards the player; it will punish foolish moves more often, and appears to have no preference between player or AI controlled clans when it comes to aggression. It will do as it sees fit without biasing towards or away from humans.

On the battlefield the AI still receives slight penalties to accuracy, morale and attack, and the player receives a slight boost to those areas.

Hard.

On this level, the AI is no longer handicapped, and instead it receives a small boost to recruitment. You should expect rival clans to be more aggressive towards you.

On the battlefield the tables have now turned: the AI begins to get bonuses to its accuracy, morale and melee attack values. The player receives nothing.

Very hard

The recruitment boost is a bit more significant at this level.

The battlefield AI bonuses of hard mode still apply, and now the AI gets a larger boost. The player still receives neither boost nor penalty.

Legendary is very hard with interface changes, as you will already know.

All of the AI bonuses are half the size of the ones awarded in previous games, according to Lusted.



That 28MB file appears to have made some difference. It's too early to say for certain as I'm only 17 turns into my new campaign. However, by 17 turns into my first game I was aware the AI was having trouble so I suppose that is a good sign. There's stuff actually happening in the early turns, where before Japan sat there staring blankly into space. Diplomacy is functioning. The AI is managing to expand, and use its armies. The real test will be the continuation of this game. I called it a night when I received a declaration of war from a clan which can't reach me. We'll see if they follow up or not. What happens with the other clans is also important. In my first game the true craziness all hit from turn 20 onwards.

For the first time out of ~11 RotS battles I saw the battle AI do something other than run straight at me. Admittedly the result was a shambles and still not a patch on the battles in the original, at least it tried. Heaven knows why it decided to split its numerically superior army up into two attack waves, allowing one to be routed while the other half of the army stood back and watched. Once the first wave were routed and my lines reorganised, the second wave were sent to share the same fate. If it had used its missile superiority it could have shot me to bits. If it had sent in the entire melee group in one go, flanking on both sides, my army would have shattered under sheer weight of numbers. But no, it didn't, so I won easily and without doing much. At this rate I shall never need to field anything other than my current mix of nag and bow levy plus pair of sword attendants.

[Kurisu]

I think it's tough to judge AI motivations or deem them flaky when you can't see the whole map and all possible variables. Although, I agree there are some suspect cases. I believe there is a script available to turn off FoW, but haven't tried it.

I just watched one of the Fujiwara clans march into my territory and take Hitachi with a nice sized stack while I was busy on the opposite front with the Taira. This made my blood boil, of course, but it's exactly what you want to see from an AI that is leaning towards hostile and you're showing weakness.. Then to my astonishment, the whole army marched off a few turns later after I made peace with Taira and started marching in that direction. I easily reacquired Hitachi and sent an agent to investigate the whereabouts of the Fujiwara army. Sure enough, there was a full stack from an allied clan threatening them in Fukushima (right next door). Subsequently, an ally of the Taira clan with whom I had just made peace saw my new weakness and took advantage of their open borders treaty with the Taira and once again the opposite front was under threat with my main army off in Hitachi. So that all went pretty respectably.

It's also the case in the S2 original campaign that I've seen AI armies do highly questionable things like approach a defenseless town and then turn around when there was no obvious threat elsewhere (me having good line of sight into their huge province and they with no other enemies). Turn after turn, like they're on patrol or something. That situation still bugs me, but I haven't seen it in my RotS campaign yet. It's certainly possible that in some of these cases, the AI is hovering around some numerical threshold for threat detection that it seems to get stuck on. If that's the case, it could definitely use an additional "do I look retarded?" check :stare:

So my experience after 70 turns has been pretty positive, based on the above and other observations. I have seen the AI do some screwy and seemingly indecisive things with agents: walking in and out of my territory, returning to the same spot for multiple turns without doing anything. Perhaps they want a peek, but don't want to risk losing the agent. I currently have a neutral clan's shirabyoshi in one of my provinces repeatedly distracting a single levy unit. Is the AI trained to rank agents in this manner? That's all I can think of. It's quite possible the agent routines need some work with their new priorities and purposes.

I also went through a situation similar to the one described in the OP where I was at peace for an eternity, then got hit with domino war declarations. Before that happened, I was questioning whether the AI was being too passive. However, there is some logic and realism to that behavior, including distant clans piling on just so they can exploit commerce raiding.

All in all, I think we need to play more campaigns to get a more complete picture. Frogbeastegg seems pretty down on things, so I hope it's just bad luck with some anomalies. Removing FoW would be helpful for for some testing. I can't comment on the battle AI just yet as I'm still getting used to not being able to see my damned troops . That's my only complaint so far, really. The unit spacing is too distant and perhaps too uniformly so. The small sizes don't help either. IMHO some clumping would be appropriate for both aesthetics and visibility, similar to the 'rabble' formation and spacing present in Empire for the armed peasantry which many mods took advantage of to represent less organized units. I wonder if that would work here? Further, does the AI perhaps see loose units as an ineffective target for prolonged archery exchange and thus more vulnerable to charges?

For the record, I'm playing VH campaign and N battle difficulties.


edit: I've posted a small mod to the mods section to disable FoW for RotS in the event anyone is interested in getting a better look at what may be influencing AI decision making (such as external threats) or watch it's army and agent movements.

It's here:
https://forums.totalwar.org/vb/showthread.php?138301-FoW-Removal-for-Rise-of-the-Samurai

[frogbeastegg]

Further, does the AI perhaps see loose units as an ineffective target for prolonged archery exchange and thus more vulnerable to charges?
Now that's a thought. The AI barely ever uses loose formation in the sengoku game. That's why my archers always have a slight advantage on a unit to unit basis. The sengoku AI is happy to target my archers in loose formation, but what if it is unsure of how to use its own units when they are in loose? Or perhaps it doesn't have values assigned to the RotS units so that it can properly understand what they do? Or perhaps something has changed and the battle AI overall has lost something which it needed to provide the previous level of tactics?

I'm going to have to try some custom battles with sengoku and RotS era units. 13 yari ashigaru, 1 general, 6 bow ashigaru should be functionally similar to a 13 naginata levy, 1 general, 6 bow levy army. Or I could go naginata attendants and naginata samurai, but the RotS AI isn't fielding many attendant grade units in my game so I'm not sure that would be as useful.


edit: I've posted a small mod to the mods section to disable FoW for RotS in the event anyone is interested in getting a better look at what may be influencing AI decision making (such as external threats) or watch it's army and agent movements.

It's here:
https://forums.totalwar.org/vb/showthread.php?138301-FoW-Removal-for-Rise-of-the-Samurai
If I see any more bizarre behaviour I shall try that. It's a better monitoring tool than my current setup of spy ships entirely innocent trade ships.

The [Date-replacing] Fujiwara are the distant clan which declared war on me as I called it a night. They have a larger army. They have a massive navy. They are classed as being rich. They are surrounded by friends and allies, and the homeland looks secure. There's no land border contact between us. Based on the sengoku game, that should add up to a naval invasion very soon. If so, I shall be a happy frog (possibly a dead one too ;p). If not, I'll be very disappointed.

[Templar Knight]

The AI is fielding nicely balanced armies in my games, mostly Levies and Attendants with Samurai support, which is nice and historical. The Campaign AI is really good in my game, I would say it slighty better than the origional, which was already great.

[Kurisu]

I see attendants mixed in as well, except when the AI is pressed, where it trains a whole lot of levy (which is appropriate). Building priorities and luck (what it has left in a war, building-wise) would be a large factor in army composition, so we'd do well to look at that in relation. It's a shame there are so many variables we can't see for testing purposes. AI income would be particularly helpful to know at times.

edit: removed that bit about the console since it does nothing at present

[Madae]

I see attendants mixed in as well, except when the AI is pressed, where it trains a whole lot of levy (which is appropriate). Building priorities and luck (what it has left in a war, building-wise) would be a large factor in army composition, so we'd do well to look at that in relation. It's a shame there are so many variables we can't see for testing purposes. AI income would be particularly helpful to know at times. It seems there's now an active console in game (shift ~). No clue yet as far as commands go. They've been making overtures to the modding community lately, perhaps this is another :surprised:

Couldn't you technically just open up the diplomacy window and speak to them? It will tell you their prosperity, which seems to be similar to a per-turn wealth stat, but probably includes other factors. All I know is that "rich" is one of the terms. However, this might be based on total income per-turn, and not considering modifiers like unit upkeep.

[Kurisu]

Couldn't you technically just open up the diplomacy window and speak to them? It will tell you their prosperity, which seems to be similar to a per-turn wealth stat, but probably includes other factors. All I know is that "rich" is one of the terms. However, this might be based on total income per-turn, and not considering modifiers like unit upkeep.

Yes, that's what I currently look at. The vagueness is perfect for gameplay, but not for analysis for the reasons you mentioned. It's also unclear if that indicator is relative to the player or global. Same goes for power. I believe I've noticed those change somewhat more drastically than you'd expect in a short period, but then there are only 5 or so ranks.

edit: I suppose those would have to be global

@ Frostbeastegg -- Somewhat OT related question: Is there any information in your guide or elsewhere as to how the AI personality (temperament and integrity) fields work in diplomacy? Specifically, I mean... the descriptions are pretty intuitive (e.g. ruthless and treacherous), but it's unclear what their weights are relative to one another and to which diplomatic events they might be tied.

[Madae]

I know you weren't asking me, but I'd just like to interject and say that the clan above Shimazu is classified as "Peaceful" and will almost always sit down and do nothing all game. About the only times they will ever go to war, at least in my games, is when their allies get attacked and they feel inclined through some means to join in, or if they get attacked and they're forced to defend themselves (and then obviously the Realm Divide if they happened to live that long). Other than that, they're about the only clan that I see that will never do anything unless provoked. Honma could be another.

On the other hand, Takeda is classified as "Aggressive", but I'm also not sure if they are set that way in every single game, or if it's random. At any rate, about 90% of the games I play, Takeda is the powerhouse.

Finally, I'd be interested in knowing if there is anything on this as well, as I typically won't ally or vassal anyone that has, at one point, been classified as "treacherous". Call me crazy...

[Kurisu]

@ Madae Yes, exactly. The extremes are pretty clear, but the middle classifications are vague as far as what to expect and how to deal with them wisely. Takeda has 'sufficient' integrity IIRC, but seems to break deals whenever the wind changes, so not so obvious. Proximity is also likely a strong factor. I assume that temperament is related to expansion behavior and integrity to the willingness to break deals at need, but it could use a little more transparency.

[Madae]

I'd be willing to bet that it has a little to do with how they act in the game as well. Such as; if they loot a town when they capture it, or if they break alliances with factions because of different circumstances, and maybe even the honor of their daimyo. Which would mean that, if they are "treacherous", it wouldn't be based entirely on a predetermined behavior, and more or less is determined by what happened in the game. I suppose some easy testing would be to just see if a clans' integrity changes based on events in the game - was it "honorable" and then "treacherous", or is it always that way regardless of what they do? Set up some alliances/trade and see if it changes when it breaks.

[Kurisu]

Temperament and integrity are fixed values, as far as I know. Those would be base values modified by other factors (random gameplay happenings, what the player does, etc), which then determine the course of AI behavior and decision making. I presume it works much like the agent system: just as a monk with a 97% chance to incite revolt and a high chance of escape may still fail and be executed (to your astonishment :inquisitive:), so too might a clan who is both peaceful and steadfast break an alliance and attack (given that you show some weakness). Not often, but when it happens the natural reaction is to assume something's wonky. In the case of agents though, we can see some numbers and understand that 3% is a real possibility. Not so with something subjective like 'sufficient'.

This relates to the OP where we're seeing these 'pile on' war declarations. Has anyone looked at the personality traits of the involved clans? I have a test game going on where chained declarations just occurred. I didn't think to look, but I think I'll back up and examine the clan traits and relationships. IMO what it comes down to is the need for more UI feedback so that when some extraordinary event occurs, players aren't confused or disappointed and can have more fun learning the mechanics.

edit: I should add that, though the numbers in the diplomatic tooltip for various factors (both player influenced and random) do give detailed numerical feedback, there's still no way to even estimate real odds if you don't know the formula. I'm at 105 relationship points (very friendly attitude) with my neighbor who is ambitious and sufficient -- so what's the risk of war for leaving a nearby, visible province undefended after I go to war with 2 other clans?

[frogbeastegg]

@ Frostbeastegg -- Somewhat OT related question: Is there any information in your guide or elsewhere as to how the AI personality (temperament and integrity) fields work in diplomacy? Specifically, I mean... the descriptions are pretty intuitive (e.g. ruthless and treacherous), but it's unclear what their weights are relative to one another and to which diplomatic events they might be tied.
Yes, there's some information in the diplomacy chapter. It's descriptive rather than hard numbers though, and based on my research rather than info from CA. Since I tried to keep to the impersonal, I didn't include a lot of observational material as it relies far too much on saying "In my games" and "I saw", and before you know it the guide becomes a badly organised story-time.

All of this is applicable for Shogun II's AI, not the RotS. I haven't managed to play that again yet, and certainly haven't put nearly as many hours into it.

I found that the text labels are generally very reliable. They aren't a cast-iron guarantee, particularly if you do something to 'provoke' the AI, if the AI clan grows very large and powerful, or if you do not pay attention to keeping the friendship rating high. An honourable, peaceful clan is still likely to break peace and attack you if you leave an unguarded castle on your border for many turns. If you have an alliance it's less likely. However a dishonourable or aggressive clan would attack sooner, and a clan with both of those negative tags instead of just one is most likely of all to attack. Having an alliance helps a lot with honourable, peaceful clans. For example, I found that as the Shimazu on hard, the Sagara would attack me in the opening parts of the game if I left my capital unguarded and moved my one army elsewhere. It typically took 3-7 turns of this before they declared on me. If I allied with them then they never attacked, not once, no matter how vulnerable I left myself. They also sent armies to fight at my side once they had finished building a full stack in addition to a decent defence force.

Sufficient is a low rating for honour. It's saying the clan will do the bare minimum to keep face. I don't trust them at all, and would never leave a tempting target where they could see it.

AFAIK Takeda are always set to be aggressive, it'd their historical personality for the time period. Often they expand or quickly get killed. Takeda will come for you as soon as they judge it expedient, typically when they decide you are vunerable due to weaker armies and/or unprotected borders and/or a series of setbacks like losses and other clans declaring on you. If they grow larger than you, watch out! It's possible to keep them as faithful allies for a long time, however this is the less likely of the two possibilities and it takes more work than with other clans.

Peaceful clans are very sedentary, as Madae said. If they expand it is usually because they have been attacked and struck back in self defence. They make excellent neighbours, and good trading partners provided their port is not at risk of being captured by another AI clan.

Treacherous clans: never give them a hostage or make any deal which you consider important! Don't pay them a lot for any deal unless you can afford to throw money away. Otherwise, go for it provided you are prepared for the inevitable betrayal. As soon as they scent a tiny hint of benefit in turning on you, they will. Sometimes buying a few turns is all that you need.

Ambitious is unreliable. The clan wants to expand, and sooner or later it will. If you can aim it at other targets you can usually keep them off your back for a time.


I'm at 105 relationship points (very friendly attitude) with my neighbor who is ambitious and sufficient -- so what's the risk of war for leaving a nearby, visible province undefended after I go to war with 2 other clans?
A lot does depend on the wider situation, however as a rule of thumb I would not recommend it. The AI will see you as over-stretched unless you have another army spare for that border. The clan may not come for you to begin with, but each turn the situation persists the greater the temptation. That particular temptation aside, a single costly battle could be enough to reduce your power status to the point where that clan decides to come after you.

Put a half stack of ashigaru in the castle, mainly archers with a few yari for backup. That should do enough to warn the AI off unless it really wants your blood, or something bad happens on the overall stage. It still may come for you however, there's never any guarantees. It's less likely, and the archers should allow you to do a lot of damage when you defend. Giving gifts to increase your friendship score should also delay them turning on you. Sometimes it buys a single turn, others it buys long enough that the situation changes and the AI no longer deems an attack beneficial. Losing a province is not a disaster provided you can take it back.

Consider making some diplomatic deals to secure your flanks, if you have not already done so. Fighting on all fronts is very dangerous. If there's no one else suitable, perhaps consider a military alliance with this neighbour, if it can be managed cheaply. Don't expect it to last forever, but if it buys you a couple of years to deal with your other enemies then it's a deal well made. Still put some troops in that border castle in order to reduce temptation, and check on the friendship status every couple of turns. Try to get it to 150 and keep it there if possible. 100 can be pretty dicey.

[Marshall Louis-Nicolas Davout]

So has there any been any patch to fix this? Surely CA should do that. I don't think they will leave it just like that.

Yes,the Takeda are aggressive! Look at my pic if you don't believe me!

http://cloud.steampowered.com/ugc/594692365021841677/73F32029B3EDFE4F99CE681E0F3FB87522056EFE/

[Nowake]

I found that the text labels are generally very reliable. They aren't a cast-iron guarantee, particularly if you do something to 'provoke' the AI, if the AI clan grows very large and powerful, or if you do not pay attention to keeping the friendship rating high. An honourable, peaceful clan is still likely to break peace and attack you if you leave an unguarded castle on your border for many turns. If you have an alliance it's less likely. (...) Consider making some diplomatic deals to secure your flanks, if you have not already done so. Fighting on all fronts is very dangerous. If there's no one else suitable, perhaps consider a military alliance with this neighbour, if it can be managed cheaply. Don't expect it to last forever, but if it buys you a couple of years to deal with your other enemies then it's a deal well made.
In my experience, the single most important criterion for how leal the AI chooses to be is the number of conflicts it is engaged in. A clan at peace with all its neighbours and capable of fielding even only one army corps will look for a target no matter its relationship with it or how powerful that target is in 95% of cases; large garrisons will deter the opening of hostilities even for prolonged periods but most of the time the way less expensive variant is to bait the attack and be done with it.

With that in mind, it must be said that a variant to weaving your own diplomatic web of alliances – and the optimal variant, in my opinion – is working on your target’s diplomatic situation rather than yours. Rather than become its ally to buy yourself time, buy off its partners into breaking alliances and trade agreements. It is often times a lot easier than you’d expect. You find its next target, rather than merely trying to make sure it won’t focus you. Always attempt to drive a wedge before making any move against a clan – especially useful in not getting yourself at war with three new clans at once when that galling yet “peaceful” distant fleet transporting a full army corps is sailing towards you and you would really rather declare war first and destroy it at sea than wait for it to land before opening hostilities.
If I have sufficient funds, I tend to break even the alliances of far-away clans, simply to keep their fiefs in a fragmented war-torn state.

Hard. On this level, the AI is no longer handicapped, and instead it receives a small boost to recruitment. You should expect rival clans to be more aggressive towards you. On the battlefield the tables have now turned: the AI begins to get bonuses to its accuracy, morale and melee attack values. The player receives nothing. Uff, I honestly read your findings both when perusing your guide and when you linked them in another thread, and each time the bit about increased melee attack values and accuracy never registered. Talk about selective reading... :shrug:



Oh well, even if applying myself to simple tasks like reading seems forlorn now, on a more positive note (read: yay!), I will get into RotS in the next few days :2thumbsup: as I've finally found the time to finish my Legendary Mori campaign. It was (surprisingly to me) my second fastest Long-set (40 provinces + Kyoto) campaign yet, having taken only eighteen years (1545 - 1563).

[frogbeastegg]

So has there any been any patch to fix this? Surely CA should do that. I don't think they will leave it just like that.
I think a part of the problem may have been that 28MB file which was missing from my original install. Since steam recovered that file the game has been better. I'm not ready to declare a total fix or a continuing issue yet; I haven't had time to play the game for several days so I have 6-7 hours experience without the 28MB file and only 2 with. Should finally have time to play more this afternoon.


With that in mind, it must be said that a variant to weaving your own diplomatic web of alliances – and the optimal variant, in my opinion – is working on your target’s diplomatic situation rather than yours. Rather than become its ally to buy yourself time, buy off its partners into breaking alliances and trade agreements. It is often times a lot easier than you’d expect. You find its next target, rather than merely trying to make sure it won’t focus you. Always attempt to drive a wedge before making any move against a clan – especially useful in not getting yourself at war with three new clans at once when that galling yet “peaceful” distant fleet transporting a full army corps is sailing towards you and you would really rather declare war first and destroy it at sea than wait for it to land before opening hostilities.
Gasp! All of those broken agreements and new wars have nothing to do with me or the 5,000 koku I happened to accidentally leave on the table while visiting! Clan frog is entirely honourable, honest and reliable. We would never consider meddling in others' diplomacy! I resent the accusation! :saint:

Yes, the odd little touch to third party diplomacy can yield nice results. It takes some delicacy though; it's easy to accidentally get a clan wiped out instead of weakened, and then there's a whole new problem to deal with. Fun!

[frogbeastegg]

I am happy to report that the 28MB file appears to have fixed my campaign AI problems. During this afternoon's game I did not witness anything I would describe as poor or dubious. The AI is doing naval invasions, making appropriate declarations of war, managing its armies correctly, capturing castles, using diplomacy, sending out agents, raiding trade routes, and all those other vital motions which result in a satisfying game. Once past the point where the three families declared open war, the game became the ruthless struggle I enjoy so much in the original Sengoku campaign. Thank heavens for that!

The battle AI is less impressive so far. I'm still winning easily despite being heavily outnumbered much of the time, and it's still not attempting to flank or do much out of the basic line of soldiering. It is making more of an effort than in those first games, in that it makes some simple manoeuvres and uses its archers to shoot at me once the melee is joined. Perhaps it will pick up if the campaign AI gives the battle AI some samurai and attendants to play with? I'm around 40 turns in and it's still mostly using levies ... as am I, for economy reasons. Hmm. The AI families are all larger than me, and the AI gets those 'very hard' level boosts, so I'd expect it to be able to field a good number of attendants at least.

No idea what the 28MB file was, but evidentially it was very important.

[Nowake]

That's quite a bit of good news!
I played RotS for about four years (Summer of 1549 now) today without noticing anything conspicuously egregious in the AI's actions (then again, I played far too little).

My own campaign is that of the Fukuhara Taira (holding Kyoto, Settsu and Kii and having the Sasaki of Omi and the Watanabe of Yamato as vassals). No particular inclination towards their faction, simply thought it would make a nice change to not play the scrappy upstart for once, and rather begin as the suppresive faction.
Plus, their original position is quite complex -- almost frustratingly so, especially when you don't really know what you're doing. I've decided to just sign trade agreements with everyone and develop economically (I've a plethora of neighbours).

It kind of works, now, after four years, I managed to raise a five thousand or so koku seasonal revenue by occupying two foreign trading posts (Chinese Texts and Incense), secure the allegiance of Shinmen of Tamba and of Ochi of Wakasa (why is the Clan destroyed? should it not be presented simply as Clan subdued? especially when the Clan was favoring you out of its own free will), though both times while their troops were away, so I only got the raw provinces (and as we well know, both are historically among the poorest) and I've only had Isonokami of Iga declare war on me just now -- wipped out his mainly Levy troops with my own single stack of almost only Naginata Attendants (I am trying to play to the clan's strengths which include a few bonuses for Naginata troops, plus, they seem good all-round troops, the equivalent of a terrestrial medium bune to my mind; hopefully I won't have a shock when meeting samurai). My plan was to gain the allegiance of the Kagawa in Awa and then subdue Tosa and Iyo in Shikoku, Sanuki being held by Yashima Taira already, yet I believe I wasted six turns and a few thousand koku on futile attempts already so now I am converting Iyo and hopping for better luck.

Anyway, my domain is a miserable gallimaufry right now, very difficult to defend should the diplomatic climate take a turn for the worse, yet I am very much impaired by my own vassals in developing a compact fief.

Gasp! All of those broken agreements and new wars have nothing to do with me or the 5,000 koku I happened to accidentally leave on the table while visiting!
Oh, by the by, do not think all those you shoulds were directed towards you in particular, I was writing for the generic topic reader.

The AI is doing naval invasions
You know, I am a bit annoyed by that. During SJ, I only saw the AI sending amphibious military expeditions towards me, so I thought it's just an extra "Die die die!" added by the level of difficulty, yet now I've witnessed a fleet packed with troops sent by Kitabatake in Ise against Ouchi in Nagato. And I just don't understand, is there any historical precedent for these type of maneuvers? I would understand if e.g. a fleet was sent to occupy Iwate by a clan based in Hitachi and fighting in Fukushima a daimyo holding the five northern provinces, but that never happens. Just out-of-the-blue amphibious invasions from clans half a world away, totally unsustainable should they succeed. It's very "hmmmm"...

[frogbeastegg]

I managed to raise a five thousand or so koku seasonal revenue by occupying two foreign trading posts (Chinese Texts and Incense), secure the allegiance of Shinmen of Tamba and of Ochi of Wakasa (why is the Clan destroyed? should it not be presented simply as Clan subdued? especially when the Clan was favoring you out of its own free will),
I had a mission to secure the allegiance of a clan which had 100%. my family's (Minamoto) influence. On successfully acquiring the province with my 'sushi I was informed that the mission had failed because I conquered the province by force. The 'sushi's mission is called 'request allegiance', so surely that's the right way to handle that mission? I can't find any other way to do it in the encyclopaedia. There may be a few minor bugs in the mission system, as I had another mission which gave me 2 years to kill a specific enemy general. It failed on the very next turn, claiming to be out of time. The target was still alive so it didn't fail for that reason.


Oh, by the by, do not think all those you shoulds were directed towards you in particular, I was writing for the generic topic reader.
I was attempting to be funny. If the AI were capable of asking I'd deny paying someone else to stab it in the back.


And I just don't understand, is there any historical precedent for these type of maneuvers?
In the Gempei war the Minamoto landed a force on one of the islands, I think it was Awaji, and fought a long, messy battle which lead to the Taira presence on the island being destroyed. Shortly after that, the Taira were destroyed completely during a naval battle which took place in the narrow inland sea between Shikoku and Honshu. I don't remember the specifics, it was a while ago that I read about it. There were a few others at the end of the Sengoku period, such as Hideyoshi's invasion of Kyushu. All the same, it looks like a rare strategy rather than a common one.

[Nowake]

I had a mission to secure the allegiance of a clan which had 100%. my family's (Minamoto) influence. On successfully acquiring the province with my 'sushi I was informed that the mission had failed because I conquered the province by force. The 'sushi's mission is called 'request allegiance', so surely that's the right way to handle that mission? I can't find any other way to do it in the encyclopaedia. There may be a few minor bugs in the mission system, as I had another mission which gave me 2 years to kill a specific enemy general. It failed on the very next turn, claiming to be out of time. The target was still alive so it didn't fail for that reason.
As I've not experienced it directly, my best bet would be a bug as well. The way gaining allegiance worked for me until now was through the peaceful transfer of the province while the entire army of the ruling clan was away (it dissappeared completely upon my taking over of their homeland, did not linger as rebels or anything of the sort). Now, while I was attempting to subdue Kagawa in Awa, their army being headquartered in their fortress, I had a few "succesful" attempts, i.e. attempts in which I was being told I may not secure the allegiance of the entire garisson, thus going ahead with the attempt would entail me declaring war (formally) first -- and the Declare war panel was appearing. This was happening while the clan was ruling only that one province, as when holding multiple regions it seems declaring war is a sine qua non for securing the allegiance of your target province anyway so that would not have surprised me. I've declined every time, as the clan was allied with three clans, the clan of my relatives among them, thus I could not risk it.


I was attempting to be funny.
I know. And you were. Though me writing it this dryly doesn't look like I appreciated the jest sigh. Anyway, I typed the caveat because I was unsure if your humour was not also attempting to transmit that I should not try to teach you to suck eggs -- I believe that's the british expression which may cover it? I get them mixed up sometimes /shrug. Plus I was told quite a few times now I can sound condescending in cases when it was so not my intention, thus currently I am probably overcompensating all over the place. But humour is not wasted on me!


In the Gempei war the Minamoto landed a force on one of the islands, I think it was Awaji, and fought a long, messy battle which lead to the Taira presence on the island being destroyed. Shortly after that, the Taira were destroyed completely during a naval battle which took place in the narrow inland sea between Shikoku and Honshu. I don't remember the specifics, it was a while ago that I read about it. There were a few others at the end of the Sengoku period, such as Hideyoshi's invasion of Kyushu. All the same, it looks like a rare strategy rather than a common one.
Oh, was not aware of the first trivia bit. But yeah, still, see, the maneuver was part of an on-going war with both sides having vast strategic interests across Japan. And Hideyoshi's invasion of Kyushu was again a campaign against a last remaining stronghold. Yet what the AI is doing seems to me more as if Uesugi Kenshin would send an expedition against Chosokabe in Shikoku sometime between the fourth and fifth battles of Kawanakajima -- we're in the presence of a total lack of arguments for it I think.

I'd rather have the AI genuinely fight me for those foreign trade posts. Least I believe the Kyushu-based clans should be scripted to declare war for their ownership. As the situation stands right now, it feels a tad like exploiting a mechanic. Yes, the AI will occupy them, but with only a token fleet of trade ships which more often than not gets chased away by wakos, opening the opportunity for you. The revenue gained is so huge it is really a game changer, it's inadmissible the AI lets you get away with it, don't you think?

[econ21]

I'd rather have the AI genuinely fight me for those foreign trade posts. Least I believe the Kyushu-based clans should be scripted to declare war for their ownership. As the situation stands right now, it feels a tad like exploiting a mechanic.

Yes, that's one of the few criticisms I have of the STW2 AI. I feel, after all the patches, the ETW AI did a better job of challenging you for trade posts. In STW2, it reminds me a little of MTW, where you could have a unmolested pan-European sea trade network that put your income far above that of the landlocked AI.

[andrewt]

I think a part of the problem may have been that 28MB file which was missing from my original install. Since steam recovered that file the game has been better. I'm not ready to declare a total fix or a continuing issue yet; I haven't had time to play the game for several days so I have 6-7 hours experience without the 28MB file and only 2 with. Should finally have time to play more this afternoon.


Gasp! All of those broken agreements and new wars have nothing to do with me or the 5,000 koku I happened to accidentally leave on the table while visiting! Clan frog is entirely honourable, honest and reliable. We would never consider meddling in others' diplomacy! I resent the accusation! :saint:

Yes, the odd little touch to third party diplomacy can yield nice results. It takes some delicacy though; it's easy to accidentally get a clan wiped out instead of weakened, and then there's a whole new problem to deal with. Fun!

I wonder if they made the game using Empire's stupid AI and then changed it midway through development. That file that was missing must have contained the AI's modifiers and it was corrupted somehow during your installation or patching.

[Monk]

I wonder if they made the game using Empire's stupid AI and then changed it midway through development. That file that was missing must have contained the AI's modifiers and it was corrupted somehow during your installation or patching.

It was corrupted for me also. Froggy and I both bought the DLC from gamersgate, and we seemed to be the only ones having this problem. In both our cases, verifying and downloading the 23mb file fixed the AI issues. :dizzy:

[Abokasee]

Early game the AI can be extremely scary, particularly the Taira with white (not gold... Fukuhara?) as they can expand west where there are virtually no opposing clans of any significance to stop them.

However later on in the game the AI does some riduculous things... like armies made up mainly of Bow Monks... why? It just seems the AI has been done very roughly as it doesnt seem to have the ability to build a balanced list.
Aside from the AI I really like Rise of the Samurai.

[frogbeastegg]

Phew, it's been a busy few days.

I'm going to edit a comment about the 28MB file into the top of the first post so that anyone new reading it will know it's fixed and not normal RotS behaviour.

Trade nodes: I find that the current setup works for me. If we are at war the AI is ruthless at hunting down ships I try to send to the nodes (or elsewhere on the map for that matter!), sometimes to the point where I can't manage to get any ships through even if I send them in a decently sized navy. The AI is happy to blockade my trade ports and pirate my trade lines. I have had trade nodes taken off me by clans and pirates, however this is the least frequent behaviour. If it were more active then, for me at least, it would begin to impinge on my enjoyment of the land game. I dislike the naval side, and I really do not enjoy building fleets and painstakingly shuffling them around the map.


I wonder if they made the game using Empire's stupid AI and then changed it midway through development. That file that was missing must have contained the AI's modifiers and it was corrupted somehow during your installation or patching.
If I were forced to guess at what the file does, I'd suggest it was some kind of priority control for buildings, diplomacy and army movements. That's a layman's guess though; I have no idea how to code an AI. I suppose it might have some relation to ETW, as that AI's major problem was its inability to prioritise successfully. It's the same game engine, so I wouldn't be surprised if many AI basics were carried over to S2. One old CA interview did say that the AI problems were related to the original coder leaving and no one else being able to pick the work up and finish it as originally envisioned. The description they gave of the planned ETW AI is remarkably similar to what we have in S2.

[Kurisu]

If you run Steam's integrity checker, it will always download a new copy of shogun2.dll. The file always has a different hash value, but it's functionally equivalent and the same size (a little under 24MB currently). This isn't due to corruption and may be a red herring as far as diagnosing a wonky installation with regard to the perceived AI issue.

Just sayin' :bow:

[frogbeastegg]

If you run Steam's integrity checker, it will always download a new copy of shogun2.dll. The file always has a different hash value, but it's functionally equivalent and the same size (a little under 24MB currently). This isn't due to corruption and may be a red herring as far as diagnosing a wonky installation with regard to the perceived AI issue.

Just sayin' :bow:
Curiouser and curiouser. I checked the size of my shogun2.dll to see if it's the same size as the one steam downloaded. It's 23.6MB. If I run the verification now I get a 23.6MB file downloaded. That's a clear match. But the file which fixed my problems was definitely reported as being 28MB. So either shogun2.dll has changed in size since, or it downloaded a different file, or it two files and lied about getting one, or it reported the size incorrectly. I'd love to know what happened; the difference to my game before and after that file is like night and day.

Shogun II: when you're not playing with its samurai, it's playing with your mind. :laugh4:

[Kurisu]

Curiouser and curiouser. I checked the size of my shogun2.dll to see if it's the same size as the one steam downloaded. It's 23.6MB. If I run the verification now I get a 23.6MB file downloaded. That's a clear match. But the file which fixed my problems was definitely reported as being 28MB. So either shogun2.dll has changed in size since, or it downloaded a different file, or it two files and lied about getting one, or it reported the size incorrectly. I'd love to know what happened; the difference to my game before and after that file is like night and day.

Shogun II: when you're not playing with its samurai, it's playing with your mind. :laugh4:

Heh, yes. Though that's Steam's doing here. Other games have the same issue when you verify them via that process. There's typically one file that it will replace every time with a new copy that has a different hash value (mathematical ID). I would speculate it's a DRM feature. In any case, the closing dialog box that says "All files verified" or whatever, doesn't take that particular mismatch into consideration, so it's inconsistent with the reported download size. Technically it's correct, since the file is legit, but it could be more verbose about what it's actually doing. You're forced to inspect the date stamps or checksums of the game files themselves to learn more.

I'm somewhat skeptical that the whole Steam verification thing has anything to do with suspect AI behavior. The reason for that is the way all Shogun's data is packed together in relatively large pack files. The contents are read and aggregated at game launch. The RotS content is mainly in patch9.pack (along with packs for movies, shaders etc). Then there's the main executable and the main dll file. If either of the latter were missing, corrupt or incorrect versions, I suspect you'd have much less subtle and specific issues (like instant crashing). Patch9.pack is nearly 1.4 GB, so if that was somehow compromised you'd be seeing a huge re-download.

I know you feel strongly about some sort of correlation, but perhaps it's just a coincidence. I still occasionally see much of the behavior you described in the OP both in RotS and in the S2 campaign. That includes armies running repeatedly back and forth between two points on the campaign map, seemingly bizarre decisions regarding attack opportunities, agents sitting around for years etc. I think that the game's AI is a complex beast (much more so than it's sometimes given credit) and it's simply prone to logical hiccups given rare circumstances. When it's good, it's great, but it certainly stands out when it isn't. Perhaps your first experience was simply a bad day for the AI. How many games did you play while it still seemed funky?

[frogbeastegg]

Perhaps your first experience was simply a bad day for the AI. How many games did you play while it still seemed funky?
With the bad AI I had 37 turns in one game, which at the speed I play is around 10 hours. I also restarted to make sure that I hadn't selected easy by accident, so that's another few hours. Both games used the same clan, the Kamakura Minamoto. I exited and reloaded the game several times. I was fortunate enough to have a lot of time to sink into the game those first two days.

With the good AI I have around 15 hours in a fresh Kamakura game, which is only about 45 turns as I've been spending a lot of time fighting battles. I also have maybe 4 hours in a second campaign with the Taira clan which gets the naginata bonuses. Most of that took place over a single week; I haven't had time to play in the last couple of weeks. The good AI is counted as beginning when I verified the cache and started another new game, as it was noticeably better within the first turns.

(I mention that I mostly played in long blocks as a matter of reference, because at one point RTW had a proven bug which caused the AI to break each time you reloaded. Anyone playing for several turns per session experienced an extremely passive AI.)

[Kurisu]

With the bad AI I had 37 turns in one game, which at the speed I play is around 10 hours. I also restarted to make sure that I hadn't selected easy by accident, so that's another few hours. Both games used the same clan, the Kamakura Minamoto. I exited and reloaded the game several times. I was fortunate enough to have a lot of time to sink into the game those first two days.



That would be odd then. Ah well, the important thing is you're enjoying it now. I had an OCD moment reading this thread and felt compelled to investigate :dizzy2: