What Are you Playing 7: Deadly Sims

[Resileaf]

And it's done, I have completed the main campaign of AC 3. Desmond is dead ("It will be over in an instant" my ass, that was a solid five seconds), Juno is unleashed, apocalypse is canceled, all that good stuff.

It was a good story overall. It would have been more enjoyable without the collectibles distraction though. The Homestead quests were also a highlight, and Achilles' passing was particularly impactful. I kind of think you could make a pretty good game of just the Homestead development. I think the game glitched on me or something though. I don't think I got all the epilogues. It told me that I had to go to Boston, but then the quest marker just disappeared.

Was the Benedict Arnold questline DLC? It felt off and disconnected to everything, the map was uninteresting and the missions were overall uninspired. I guess they put the bulk of their efforts in King Washington.

Three days left to do that. If there isn't too much padding, it shouldn't be too long to do, hopefully.

[Cazero]

I don't mean to be argumentative, but I've simply never had a problem with maintaining my upkeep from battles in any Total War game with any faction on any difficulty. I'm sure it would be different against human opponents, but the AI absolutely loves to throw tons of garbage armies at you.

Strange. In my experience, the AI makes sure to keep their even slightly weaker armies just a little bit out of reach, and is willing to abandon the defense of settlements to do so. It takes multiple turns or a lucky ambushing (sometimes failing even at 100% annouced success rate) to catch up.

[Theoboldi]

Just tried out Midnight Fight Express, a very cool top-down beat-em-up with gritty John Wick-style action.

The story itself and the kinds of villains you go up against are extremely over the top, but it does deliver on the fast-paced mixture of kung fu, guns, gun fu, and kicking chairs into people's faces. Everything that happens feels appropriately lethal and dangerous, and especially as you unlock moves you start feeling like an absolute powerhouse taking on incredible odds.

For the most part, the challenge is tough but fair. You don't have a lot of health, and enemies can overwhelm you quickly if there's many of them or they have weapons. But in return they also can't take much punishment, even (most of) the bosses, so you're encouraged to fight dirty and aggressive.

There are some issues though. Enemies with heavy guns, like assault rifles, sniper rifles or grenade launchers are appropriately dangerous, which often means they can kill you in a split second from across the screen before you get a chance to react. Dying isn't too punishing, but it's still frustrating, especially as the game later on expects to take on large groups of these foes. And if you don't already have a gun with you once they start firing, it just becomes a matter of rolling around, abusing invincibility frames and occasionally throwing a single punch until one goes down and you can take their gun. The usual methods of disarming you have also don't work well on them, because they leave you as a sitting duck and get you killed.

There's also some issues with stage hazards sometimes not being telegraphed, some lacking warning markers while some that have them allow them to be obscured by scattered objects. This can lead to the occasional cheap death as you don't see you're standing in the instant death zone.

Finally, there's some (thankfully non-recurring and rare) enemies and bosses with annoying gimmicks, who end up being immune to most of your arsenal of moves, leaving you to fall back on much less fun hit and run tactics that break the pacing. Sometimes they can be fun as the stage gives you interesting other ways to deal with them, but sometimes not.

Overall though, it was very fun. Would recommend checking this out if you're into over-the top action. The frustrating parts are worth dealing with, and for 20 bucks you're getting a good few hours of fun. Emphasis on a few, though.

[Rodin]

My favorite is finding the "Unseen Assassins" note in a grave, and then a few minutes later getting jumped in the woods near the Frenzied Village by one of the perma-invisible Black knife Assassins. Cue panicked screaming.

Nice! I've been lucky enough to not run into any of those yet, or any other world-bosses-with-no-healthbar except for the ubiquituous Crucible Knights.

I did manage to fully clear Limgrave and Weeping Peninsula without leaving those zones to hunt for better materials, which left me only one place to go: Malenia.

And I can't do it. She's heavily nerfed over her "true form" thanks to level scaling, but even so it's just not possible for me to take her down with only +2 Somber weapons at around level 40. I can survive Waterfowl Dance if she does it when I'm at full health, and I even managed to take her to second phase a couple times. But that's where it ends - Margrit's Bridge is a tiny boss arena and it's basically impossible to avoid the Scarlet Aeonia opener to phase two, and even getting her to that phase takes all 6 of my flasks.

I'm gonna have to break my rule and bypass her for now. What's really worrying me is that I don't know how to get into any of the other legacy dungeons other than Radahn's. Raya Lucaria requires the key, which has been randomized off somewhere. Volcano Manor needs either the parlor key (randomized) or getting eaten by an Abductor Virgin that won't be there. And Leyendell requires getting two Great Runes when the bosses might not drop them!

[Anteros]

Strange. In my experience, the AI makes sure to keep their even slightly weaker armies just a little bit out of reach, and is willing to abandon the defense of settlements to do so. It takes multiple turns or a lucky ambushing (sometimes failing even at 100% annouced success rate) to catch up.

It probably depends on if you're rolling in with what the ai perceives to be a death stack or something they can beat. I usually tend to favor multiple smaller armies since even on legendary the ai is so bad that you can reliably beat them while outnumbered 2 or 3 to 1. It means you can't auto Calc most battles, but the auto Calc is so unreliable I don't like to use it anyway unless I have a huge advantage.

Like seriously, the AI in these games is almost unacceptably bad. There's so many strategies that it just can't deal with at all. Anything more complex than just lumping your units together with the archers in the back will have it running in circles. It's especially bad in Warhammer. Kiting was already easy but adding flying units makes it ridiculously so. Or just lumping your enemies up on one target and blasting them all with a spell.

I know ai is hard, and some of the maps in this game are very complex which makes it even harder....but this ai is basically the same level as stuff you'd see 20 years ago. You'd think that after so many releases of basically the same game with different skins they'd have fine tuned it a little. I think it's actually worse than in their older games.

[Rynjin]

I did manage to fully clear Limgrave and Weeping Peninsula without leaving those zones to hunt for better materials, which left me only one place to go: Malenia.

And I can't do it. She's heavily nerfed over her "true form" thanks to level scaling, but even so it's just not possible for me to take her down with only +2 Somber weapons at around level 40. I can survive Waterfowl Dance if she does it when I'm at full health, and I even managed to take her to second phase a couple times. But that's where it ends - Margrit's Bridge is a tiny boss arena and it's basically impossible to avoid the Scarlet Aeonia opener to phase two, and even getting her to that phase takes all 6 of my flasks.

I'm gonna have to break my rule and bypass her for now. What's really worrying me is that I don't know how to get into any of the other legacy dungeons other than Radahn's. Raya Lucaria requires the key, which has been randomized off somewhere. Volcano Manor needs either the parlor key (randomized) or getting eaten by an Abductor Virgin that won't be there. And Leyendell requires getting two Great Runes when the bosses might not drop them!

Yeah, situations like this are probably why the Randomizer has a "disable Malenia self-healing" toggle.

[Mechalich]

I finally managed to beat the main story of Sands of Salzaar. The game's come a long way since early access with a lot of the nasty grinding removed (I think, I still conquered everyone to set up max legacy bonus, but I don't believe that's actually necessary anymore). It's a fun little army/party builder sort of game though it could use some more polish.

[Anteros]

Nice! I've been lucky enough to not run into any of those yet, or any other world-bosses-with-no-healthbar except for the ubiquituous Crucible Knights.

I did manage to fully clear Limgrave and Weeping Peninsula without leaving those zones to hunt for better materials, which left me only one place to go: Malenia.

And I can't do it. She's heavily nerfed over her "true form" thanks to level scaling, but even so it's just not possible for me to take her down with only +2 Somber weapons at around level 40. I can survive Waterfowl Dance if she does it when I'm at full health, and I even managed to take her to second phase a couple times. But that's where it ends - Margrit's Bridge is a tiny boss arena and it's basically impossible to avoid the Scarlet Aeonia opener to phase two, and even getting her to that phase takes all 6 of my flasks.

I'm gonna have to break my rule and bypass her for now. What's really worrying me is that I don't know how to get into any of the other legacy dungeons other than Radahn's. Raya Lucaria requires the key, which has been randomized off somewhere. Volcano Manor needs either the parlor key (randomized) or getting eaten by an Abductor Virgin that won't be there. And Leyendell requires getting two Great Runes when the bosses might not drop them!

Well, the bright side is radahn's fight should be easy since you'll fight a random boss there not intended to be fought with all those adds. I'd probably head there next and hope some of those other issues resolve themselves in the meantime.

You guys have got me tempted to try the randomizer myself. It sounds fun.

[Zombimode]

Like seriously, the AI in these games is almost unacceptably bad. There's so many strategies that it just can't deal with at all. Anything more complex than just lumping your units together with the archers in the back will have it running in circles. It's especially bad in Warhammer. Kiting was already easy but adding flying units makes it ridiculously so. Or just lumping your enemies up on one target and blasting them all with a spell.

I know ai is hard, and some of the maps in this game are very complex which makes it even harder....but this ai is basically the same level as stuff you'd see 20 years ago. You'd think that after so many releases of basically the same game with different skins they'd have fine tuned it a little. I think it's actually worse than in their older games.

Before you go on I would like that you ask yourself if there are maybe other reasons why the AI behaves they way it does.

Challenge is not the only goal and arguebly it is not even close to being the most important goal (for the majority of gamers). Spectacle and aethetics of play are likely more important.

"Staunch lines of spears" instead of Archer Checkerboxes exists because that is how people want to play.


A Total War battle AI that plays "optimally" would look completely different from the "big armies clash" spectacle that Total War wants to provide.


I've started a Vilitch campaign on the Realms of Chaos map. And I'm loving it. Vilitch is a cool lord with pretty funny skills. I love the changes made to diplomacy and war coordination.

[Rodin]

Yeah, situations like this are probably why the Randomizer has a "disable Malenia self-healing" toggle.

I saw that in the Randomizer options, and debated heavily whether to turn it off or not. Life steal is the sole reason why Malenia is so OP - Waterfowl Dance is just an HP check, and if she doesn't lifesteal there are many other strategies you can use to beat her. I settled on off, entirely out of hubris.

To be fair to myself, this spawn is the single unluckiest one. Even swapping only main bosses it's single digits. I'm not sure exactly what the criteria they used is - I counted on fingers and came up with 15+, but checking Fextralife several of those are listed separately to main bosses. That includes Margrit, which is why I'm in this pickle in the first place. According to the item randomizer it's around 30, or around 3% chance of getting this particular spawn.

It's also the ONLY spawn that blocks a legacy dungeon. The Draconic Tree Sentinel is a field boss - I fought and killed him already. The Red Wolf of Radagon may count, but he's towards the end of the dungeon. All the others are end of dungeon bosses.

Well, the bright side is radahn's fight should be easy since you'll fight a random boss there not intended to be fought with all those adds. I'd probably head there next and hope some of those other issues resolve themselves in the meantime.

You guys have got me tempted to try the randomizer myself. It sounds fun.

Radahn's fight would be the smart move. Unfortunately I can't go there yet based on my own self-imposed rules. When I watched the randomizer Boss Rush (speedrunning the game with randomized enemies and bosses and keeping score based on who you kill) he did what everybody does - he ignored all content in favor of sprinting over to Caelid/Dragonbarrow and grabbing all the high end Somber Smithing stones laying around out there. He also whacked the great white dragon (who doesn't get randomized) while he was at it for the massive souls boost.

I didn't want to cheese the game that way, especially since there's a randomizer option that lets you limit smithing stone availability based on level. The mines have smithing stones in the same place as normal and they're roughly the same level as normal. Higher level smithing stones are mostly prevented from spawning in low level areas. The easy way to beat the game would be to do a mine rush - hit the mines in Limgrave and Weeping Peninsula, then Liurnia, then Caelid, then use the power boost to get yourself up to Altus Plateau and hit the mines there.

The rules I put in place were that I couldn't go fishing for loot. I can't beeline it to the mine in each new area, I have to clear in a steady pattern across the map. I have to full clear each area before leaving, including defeating every boss.

I can now see that last was a bit optimistic. Malenia in Margrit's boss location is one of the worst, but there are others I can now imagine. How about Rykard in Rennala's library, and when you transition off him you've got Elden Beast in Rennala's second phase? Nightmare material.

So what I think I'm going to do is add a "best efforts clause". I've already been playing with boss skipping - I don't have to fight bosses I've no chance of winning against, I just have to come back later and beat them before being allowed to move to a new zone. I just have to add that for legacy dungeons and their bosses. If a legacy dungeon is inaccessible due to RNJesus then I'm allowed to skip it, and that includes getting bad boss luck as long as I put the effort in to beat the boss and fail.

The randomizer is incredibly fun, especially since it allows for tactics you normally can't do. I beat Elemer of the Briar because he spawned near a ledge and I was able to get under him and shank his ankles.

[Anteros]

Before you go on I would like that you ask yourself if there are maybe other reasons why the AI behaves they way it does.

Challenge is not the only goal and arguebly it is not even close to being the most important goal (for the majority of gamers). Spectacle and aethetics of play are likely more important.

"Staunch lines of spears" instead of Archer Checkerboxes exists because that is how people want to play.


A Total War battle AI that plays "optimally" would look completely different from the "big armies clash" spectacle that Total War wants to provide.


I've started a Vilitch campaign on the Realms of Chaos map. And I'm loving it. Vilitch is a cool lord with pretty funny skills. I love the changes made to diplomacy and war coordination.

Then why do difficulty levels exist? Why have a legendary difficulty if it's going to have the same terrible AI as the easy one? Right now the only difference between legendary and easy is that the AI cheats more so it fields larger stacks to run into your grinder. I understand wanting to make players feel like they're smart and good at the game, but harder difficulties at least should be challenging. I'm not asking for optimal play. I'm asking it to be smarter than games that came out 20+ years ago. I played a campaign last night where the AI kept attacking my fully fortified town with stacks of 3 units every turn. They don't even make it past the gate. They just run forward and then run away. Every turn. I can't auto-calc it though because the auto calc is so bad it thinks I should take "medium losses" despite them literally not even reaching my units, so every time I end turn I have to go through 2 loading screens and several minutes of the AI running forward and dying to towers before it runs away just to repeat with new units the next turn. It made me abandon the campaign.

There are plenty of games where I want to turn my brain off and smash things. Grand strategy games are not one of those niches. I would expect most people who enjoy the genre to agree with me, although I obviously can't speak for anyone but myself. In Total War it's so bad that it's not even turning your brain off. You have to intentionally play poorly and implement bad strategies to give the AI a chance. It's like a version of chess except instead of trying to out-position each other both players just mash their pieces into each other and see who has more standing at the end. If that's satisfying to you then great, but it's not what I expect when I buy a chess board.

You're right though. All of my ramblings are about what I think a strategy game should try to provide. What they're actually intending to provide is all the spectacle of the ocean with the depth of a puddle. I suppose it must be working for them given that players keep buying their "new' game every year. I shouldn't be surprised given the success of similar franchises like Fifa or Madden. Why bother making a game when you can just slap a few new skins on the old one and resell it for 60 bucks every year?

[Spacewolf]

Then why do difficulty levels exist? Why have a legendary difficulty if it's going to have the same terrible AI as the easy one? Right now the only difference between legendary and easy is that the AI cheats more so it fields larger stacks to run into your grinder. I understand wanting to make players feel like they're smart and good at the game, but harder difficulties at least should be challenging. I'm not asking for optimal play. I'm asking it to be smarter than games that came out 20+ years ago. I played a campaign last night where the AI kept attacking my fully fortified town with stacks of 3 units every turn. They don't even make it past the gate. They just run forward and then run away. Every turn. I can't auto-calc it though because the auto calc is so bad it thinks I should take "medium losses" despite them literally not even reaching my units, so every time I end turn I have to go through 2 loading screens and several minutes of the AI running forward and dying to towers before it runs away just to repeat with new units the next turn. It made me abandon the campaign.

There are plenty of games where I want to turn my brain off and smash things. Grand strategy games are not one of those niches. I would expect most people who enjoy the genre to agree with me, although I obviously can't speak for anyone but myself. In Total War it's so bad that it's not even turning your brain off. You have to intentionally play poorly and implement bad strategies to give the AI a chance. It's like a version of chess except instead of trying to out-position each other both players just mash their pieces into each other and see who has more standing at the end. If that's satisfying to you then great, but it's not what I expect when I buy a chess board.

You're right though. All of my ramblings are about what I think a strategy game should try to provide. What they're actually intending to provide is all the spectacle of the ocean with the depth of a puddle. I suppose it must be working for them given that players keep buying their "new' game every year. I shouldn't be surprised given the success of similar franchises like Fifa or Madden. Why bother making a game when you can just slap a few new skins on the old one and resell it for 60 bucks every year?

AI is definitely the biggest issue in TW games these days, in some ways it feels worse than it did back in Rome 1. They do keep adding things to make it look better but most of the workings are still from that generation and the things getting added just try to hide that fact. Though the issue with them attacking your city is more of an issue of how the game calculates auto battle results, the results are always skewed towards certain stats for example armour which is why AI dwarves usually do well. So in that case what the AI is doing makes sense because it can only really go off the auto resolve result though there should probably be some sort of flag that gets activated after afew attacks to say something is wrong so add more units.

The campaign AI does generally seem to be better since it has less to deal with, it'll try to pull your armies out of position to kite them around, armies with access to underway will use it to flank you, they recognise when your armies are out of position to declare war obviously it cheats like a bastard but no 4x AI doesn't cheat. It's the battle AI that really needs the overhaul.

[Resileaf]

Making an AI in a game with so many moving pieces is incredibly difficult and I think people don't appreciate that enough. With every faction having completely different units, completely different playstyles, completely different possible strategies, completely different spells, completely different counters, completely different lords and heroes, all of this with units that are composed of around a hundred different entities that have to move in blocks, on complex battlefields with various terrain types, all of this with the limitations of what a computer 'sees'... It's just way more complicated than people give it credit for.

It's impossible to make a computer act like a player. Players act on what they see on their screen, with limitations on what they're able to do at a given time (can't control two units and tell them to do different things at the same time, for example) while a computer doesn't even see anything, it has no visual references. A computer can only be programmed to react to events in basic ways rather than make complex strategies. Besides, it's not like the AI can learn through playing. Over time, you will always become better than the AI because you will learn how it works and what strategies it has difficulty countering.

You compared this to chess, but chess is incredibly simple to program for. Chess is 64 squares, 32 pieces, limited, specific movements unique to each piece. A computer can know every move in the game tens of steps ahead and plan ahead for it. Total War is not comparable to chess in the least.

As an aside, no one who cheeses or exploits an AI has a right to complain about how bad it is.

[zlefin]

Technically speaking, it'd be quite feasible to make a good ai for the total war games; I've looked a fair bit at game ai's. The real problem is that companies simply choose not to invest that much into making the ai that good, because doing so doesn't help their revenue much. People like to win; and the number of people who like a true challenge, or an adaptable challenge that improves with them, are fairly scarce. And even when they do exist, it's not like you can get a lot of extra money out of them.

A computer can certainly make extremely complex strategies, if it's designed to do so (or capable of learning). Most simply aren't.

[warty goblin]

The goal of game AI isn't to be good at the game, win, or even really present a challenge, it's to be enjoyable for the player to beat up. Most players don't finish most games, don't want to lose*, and play on Normal, spending lots of cash on better AI that only the hardcore who have played every game in the series for like 200+ hours will benefit from is probably a bad investment.

It's also a computationally expensive one. If the selling point of your game is big gorgeous armies hammering at each other, you probably don't want to skimp on graphical fidelity or army size to get the CPU time back for better AI.

Or looked at very cynically, about 100% of people who complain about AI bought and played the game. The game was therefore good enough. If they played for like 50 hours, I do t even think it's cynical to say the game was good enough. Games are illusions, state at them long and hard enough and the wheels come off.


*If they wanted a game they lost a lot, they'd be playing MP.

[Resileaf]

Well it's not like the series has only ever had bad AI. The AI in Shogun 2 was considered to be pretty solid and that was thanks to the lack of unit diversity making it much easier for it to formulate strategies, develop counters and react to battlefield changes.

Also it was a huge fan of camping hills.

[Zombimode]

You're right though. All of my ramblings are about what I think a strategy game should try to provide. What they're actually intending to provide is all the spectacle of the ocean with the depth of a puddle. I suppose it must be working for them given that players keep buying their "new' game every year. I shouldn't be surprised given the success of similar franchises like Fifa or Madden. Why bother making a game when you can just slap a few new skins on the old one and resell it for 60 bucks every year?

I understand your frustration but I think you're vastly underselling the ammount of effort each Total War title represents, especially the Warhammer titles.

Another point regarding challenge: What you call "intentionally playing poorly and implementing bad strategies" is in the perspective of other players "building a themed army", "making loreful choices", "intentionally ignoring exploits/cheese options" and "playing to match the desired aesthetic".
Under those "constrains" the moment-to-moment challenge actually increases.

And there is, generally speaking, a negative effect of higher levels of dificulty: that they may narrow down the possible avenues for success. Especially in a high variety environment like Warhammer Total War with hundreds of vastly different units, magic, lords, factions etc. etc.
The more difficult the game gets, the more the players is persuaded to pick the most optimal path to success.

Now, don't get me wrong: I get that none of my points will make you enjoy the game more. But maybe they will help you understand why some (many) players like the games how they are. And that it doesn't have anything to do with those players being dumb or incompetent

[Rodin]

AI is definitely the biggest issue in TW games these days, in some ways it feels worse than it did back in Rome 1. They do keep adding things to make it look better but most of the workings are still from that generation and the things getting added just try to hide that fact. Though the issue with them attacking your city is more of an issue of how the game calculates auto battle results, the results are always skewed towards certain stats for example armour which is why AI dwarves usually do well. So in that case what the AI is doing makes sense because it can only really go off the auto resolve result though there should probably be some sort of flag that gets activated after afew attacks to say something is wrong so add more units.

The campaign AI does generally seem to be better since it has less to deal with, it'll try to pull your armies out of position to kite them around, armies with access to underway will use it to flank you, they recognise when your armies are out of position to declare war obviously it cheats like a bastard but no 4x AI doesn't cheat. It's the battle AI that really needs the overhaul.

The auto battle for sieges being off is something I first noticed in Rome 1. If you take charge of a siege you are going to lose a significant portion of your army to the invulnerable arrow towers. It's an inevitable part of sieging that you have to put up with, and it made siegin have some stupid strategies. But no matter how you played you were going to lose a big chunk to attrition. No way around it.

The auto-battle didn't take that into account. It just looked at the two armies, slammed a force multiplier down on the besieged side, and then ran the auto-battle results. It bore no resemblance to what would actually happen if you took charge. A large army would crush a small one with no casualties, because the force multiplier wasn't enough to overcome the lopsided auto-battle. It was worse if you were defending, because the AI was stupid. It would stand in range of your arrow towers and take far more casualties than you would. And if you auto-battled those losses wouldn't occur.

It made it so that auto-battle for sieges was non-viable. You HAD to fight those battles no matter how lopsided they were, because you would lose battles you would easily win and get crushed in battles you would lose but inflict horrific casualties on the enemy army that made taking the city back easier.

They may have improved on the auto-battles in later games, but it was a sign for me for how slap-dash the process was for CA.

[Resileaf]

Auto-resolve is not the point of the game though? Like, it's basically there to not make you have to go through every little skirmish that you would crush the opponent in, not replace every battle. Auto-resolving was never going to be a complete recreation of a given battle, that would be pointless use of processing power, especially because all AIs use auto-resolve in their battles on their turns, so anything too complicated would result in longer turn times.

[Rodin]

Auto-resolve is not the point of the game though? Like, it's basically there to not make you have to go through every little skirmish that you would crush the opponent in, not replace every battle. Auto-resolving was never going to be a complete recreation of a given battle, that would be pointless use of processing power, especially because all AIs use auto-resolve in their battles on their turns, so anything too complicated would result in longer turn times.

Auto-resolve is meant to be a reasonable simulation of the battle given that you are not there using your superior human brain to direct things.

A basic scenario would be 1000 troops vs 1000 troops of equal power. Stick that into an auto-battle and the result is a toss-up. Put a human in charge and the human wins comfortably, because they're smarter than the AI.

And if that's what the TW auto-resolve was doing that would be fine. In the case of field battles it generally was. An army with a distinct advantage would win over a weaker army, with more casualties than you would expect if you were commanding it.

That's not what auto-resolve was doing in the case of sieges. 1000 troops vs 100 troops in a city is a certain victory for the 1000 troops. However, if you put the superior human brain in charge they would lose 200 troops by sheer inevitability from the invulnerable arrow towers whittling down your army as you marched them up to the walls, lost men on ladders, and then lost a decent chunk fighting on the walls with arrows raining down on them.

But if you do an auto-resolve? The computer would lose 20-30 troops. Because those arrows didn't exist, and it just mathed out the victory for 1000 v 100 with a slight bonus for the defenders. It punished the human player for taking control in the case of certain victory and punished the human player for not taking control even though victory was impossible and you just wanted to lose the city and re-take it some turns later with your reinforcements.

In short, the auto-resolve wasn't accurate. It wasn't even close. They didn't need to do a complete simulated re-creation of the battle. They needed to play simulated AI vs AI battles in the development stages and come up with a better formula that properly represented what a realistic result would be given the battle system that existed.


Edit:

Latest Elden ring weirdness: A trapped loot in Liurnia that's supposed to summon a half dozen phantasmal soldiers around you. One of those soldiers was an elder dragon. Not the first time I've been yeeted off a cliff in a Souls game, but "fell because an invisible phantom dragon landed on my head" is a definite first.

[Resileaf]

No, auto-resolve is most definitely not a simulation of the battle. It's just the game hitting the numbers together. It was never anything else and it was never intended to be anything else. A simulation of a battle would take much longer to go through because it would have to fake-load a battle map, fake-place all troops, fake-move and fake-fight them. Maybe not that much of an issue for a single battle, but over the entirety of the map? In Immortal Empires, it would take forever for the computer to go through those simulations every turn.

In any case, you should be happy that you get the foil the game's expectations by winning what seems to be a one-sided battle by taking advantage of your superior human brain to develop strategies that will save your grossly outnumbered garrison.

By the way arrow towers aren't invulnerable, they can be destroyed by siege weapons.

[Rodin]

No, auto-resolve is most definitely not a simulation of the battle. It's just the game hitting the numbers together. It was never anything else and it was never intended to be anything else. A simulation of a battle would take much longer to go through because it would have to fake-load a battle map, fake-place all troops, fake-move and fake-fight them. Maybe not that much of an issue for a single battle, but over the entirety of the map? In Immortal Empires, it would take forever for the computer to go through those simulations every turn.

In any case, you should be happy that you get the foil the game's expectations by winning what seems to be a one-sided battle by taking advantage of your superior human brain to develop strategies that will save your grossly outnumbered garrison.

By the way arrow towers aren't invulnerable, they can be destroyed by siege weapons.

I never once said it was anything other than the game mashing numbers together.

I said it was the game mashing numbers together BADLY.

And doing so in a way that makes auto-resolving a BETTER result than actually playing the game in many situations. If you think that's what CA intended, well....

[Resileaf]

Ugh, I meant to say "It's not meant to be a simulation" and not "It's not a simulation".

In any case, yes, I do believe that this is what CA intended, because it's a time-saving feature more than a gameplay feature.

[Anteros]

AI is definitely the biggest issue in TW games these days, in some ways it feels worse than it did back in Rome 1. They do keep adding things to make it look better but most of the workings are still from that generation and the things getting added just try to hide that fact. Though the issue with them attacking your city is more of an issue of how the game calculates auto battle results, the results are always skewed towards certain stats for example armour which is why AI dwarves usually do well. So in that case what the AI is doing makes sense because it can only really go off the auto resolve result though there should probably be some sort of flag that gets activated after afew attacks to say something is wrong so add more units.

The campaign AI does generally seem to be better since it has less to deal with, it'll try to pull your armies out of position to kite them around, armies with access to underway will use it to flank you, they recognise when your armies are out of position to declare war obviously it cheats like a bastard but no 4x AI doesn't cheat. It's the battle AI that really needs the overhaul.

I mean...then fix the auto resolve? I don't get how "well the auto resolve we made is terrible and the AI is based on that so of course it's also terrible" is a valid excuse at all.

Or looked at very cynically, about 100% of people who complain about AI bought and played the game. The game was therefore good enough. If they played for like 50 hours, I do t even think it's cynical to say the game was good enough. Games are illusions, state at them long and hard enough and the wheels come off.

That might be true for many games, but Total War releases so frequently that I've absolutely skipped several of their titles because I know exactly what they're offering. Or refunded ones that I've tentatively tried to see if they've improved at all.

I understand your frustration but I think you're vastly underselling the ammount of effort each Total War title represents, especially the Warhammer titles.

Another point regarding challenge: What you call "intentionally playing poorly and implementing bad strategies" is in the perspective of other players "building a themed army", "making loreful choices", "intentionally ignoring exploits/cheese options" and "playing to match the desired aesthetic".
Under those "constrains" the moment-to-moment challenge actually increases.

And there is, generally speaking, a negative effect of higher levels of dificulty: that they may narrow down the possible avenues for success. Especially in a high variety environment like Warhammer Total War with hundreds of vastly different units, magic, lords, factions etc. etc.
The more difficult the game gets, the more the players is persuaded to pick the most optimal path to success.

Now, don't get me wrong: I get that none of my points will make you enjoy the game more. But maybe they will help you understand why some (many) players like the games how they are. And that it doesn't have anything to do with those players being dumb or incompetent

It doesn't have to be cheese. The AI can't even handle simple concepts like flanking. It should be relatively simple to program it to designate a unit of pikes or whatever to protect their ranged units. They simply choose not to do so. I also think people vastly overstate how different most of the civs play from each other. They're mostly just reskins of the same units with slightly different stats. For the most part every race just has the basic mix of cavalry units, ranged units, and a few meat shield units. The fact that they're all different colors doesn't wildly change the way they play. At least not by the AI. It plays them all exactly the same. I'm sure a human player can do lots of fun and unique things with different races, but the AI does none of it.

Auto-resolve is not the point of the game though? Like, it's basically there to not make you have to go through every little skirmish that you would crush the opponent in, not replace every battle. Auto-resolving was never going to be a complete recreation of a given battle, that would be pointless use of processing power, especially because all AIs use auto-resolve in their battles on their turns, so anything too complicated would result in longer turn times.

That's not a valid point though. It's not uncommon for the AI to throw something like 10+ battles at you in a single turn late-game. These battles are unlosable to a human, but the auto-calc will lose every single one, or at least lose enough units that it loses the next one. You have to play them every turn, and it's a massive, repetitive, unfun, time sink. The auto-resolve is absolutely a massive integral part of the game, and the fact that it's so bad is a glaring problem.

[Mechalich]

That's not a valid point though. It's not uncommon for the AI to throw something like 10+ battles at you in a single turn late-game. These battles are unlosable to a human, but the auto-calc will lose every single one, or at least lose enough units that it loses the next one. You have to play them every turn, and it's a massive, repetitive, unfun, time sink. The auto-resolve is absolutely a massive integral part of the game, and the fact that it's so bad is a glaring problem.

Auto-resolve has been a problem for the TW series going all the way back to Shogun 1 - where it was essentially impossible to win an auto-resolve that required attacking across a river even if you had a 30 to 1 advantage. They've never fixed and its pretty clear they're never going to. Truthfully the series has over time moved further and further away from the 'grand strategy' setup and more into a just being a battle simulator.

[warty goblin]

I like the more recent Age of Wonders auto-resolve, where if you don't like the result, you can still manually fight the battle. It helps that the auto-resolve generally feels about right in those games, and occasionally gets better results than I can. If I have a really super-synergized army and use maximum cheese tactics I still outperform it, but for just getting a pretty good representation of a sort of average battle, it does fine. And even in a lot of cases where I would get substantially better results, the outcome of the battle is of limited interest and frankly not worth the time just to keep like 1 stack of weenie dudes alive.

Really, what all these hybrid strategy/tactics games need is a way to cut down on the number of battles per campaign. Maybe only let the player fight the ones with a leader/hero involved, or have some sort of sliding scale where as the size of armies increases, battles that involve very small forces are just auto-resolved automatically.

[Anonymouswizard]

Well SR:HK, once it gets going,, really has that 'just one more run' quality. So here I am at 2am the second night in a row, having played for almost six hours. I've completed all the optional runs and have started the end game.

I'm very glad to be playing a mage, Gobbet is a great character but she's the second crew shaman in a row, whereas going Street Samurai or Adept would require leaving Duncan or Gaichu off the team. As it is I've loaded myself with Heal Wound, Strip Armour, and s many combat spells as I can get my hands on. At 9 Willpower/Spellcasting I feel ready to take on the hordes of enemies!

So now it's time for two orks to attack their grandmother. This game didn't quite go where I expected (not helped by the implication that while Duncan didn't take Raymond's surname his sister did, blame my lack of imagination).

Looking back on Dragonfall and Hong Kong, I think they work because the developers just took functional systems and used them to prop up the themes and story of the trilogy. The combat is a functional X-COM clone, the dialogue checks are just thresholds, and character building is just a streamlined version of 3e. There's nothing really innovative or fexemplary except for the writing, which really does carry the games. The linearity itself is honestly a plus, I don't go into missions worried I haven't met the developer intended bumming about crafting tripwires quota. There's even the little touches of Shadowrun lore like one character being an elf poseur.

What I'm saying is that I really want Shadowrun Returns 4. Maybe I'll play Dead Man's Switch after this

[Anteros]

Well SR:HK, once it gets going,, really has that 'just one more run' quality. So here I am at 2am the second night in a row, having played for almost six hours. I've completed all the optional runs and have started the end game.

I'm very glad to be playing a mage, Gobbet is a great character but she's the second crew shaman in a row, whereas going Street Samurai or Adept would require leaving Duncan or Gaichu off the team. As it is I've loaded myself with Heal Wound, Strip Armour, and s many combat spells as I can get my hands on. At 9 Willpower/Spellcasting I feel ready to take on the hordes of enemies!

So now it's time for two orks to attack their grandmother. This game didn't quite go where I expected (not helped by the implication that while Duncan didn't take Raymond's surname his sister did, blame my lack of imagination).

Looking back on Dragonfall and Hong Kong, I think they work because the developers just took functional systems and used them to prop up the themes and story of the trilogy. The combat is a functional X-COM clone, the dialogue checks are just thresholds, and character building is just a streamlined version of 3e. There's nothing really innovative or fexemplary except for the writing, which really does carry the games. The linearity itself is honestly a plus, I don't go into missions worried I haven't met the developer intended bumming about crafting tripwires quota. There's even the little touches of Shadowrun lore like one character being an elf poseur.

What I'm saying is that I really want Shadowrun Returns 4. Maybe I'll play Dead Man's Switch after this

There are some excellent fan modules. I'd go so far as to say some of them are better than Dead Man's Switch. Antumbra chronicles or the caldecott caper are both in that category.

I always mean to try a mage, but sniping is just so satisfying.

[GloatingSwine]

That's not a valid point though. It's not uncommon for the AI to throw something like 10+ battles at you in a single turn late-game. These battles are unlosable to a human, but the auto-calc will lose every single one, or at least lose enough units that it loses the next one. You have to play them every turn, and it's a massive, repetitive, unfun, time sink. The auto-resolve is absolutely a massive integral part of the game, and the fact that it's so bad is a glaring problem.

That's like every autoresolve system though. They can't account for the impact of human skill, all they can do is math out the armies you feed into them.

It's definitely true that autoresolve in TW: Warhammer ignores certain things that it should account for (magic and aura effects especially), and the effect of defences in settlement battles are one of those things.

I'm not terribly worked up about the defences thing though, (In fact some settlements have deliberately undervalued defensive modifiers, the gate settlements in the Empire, to make the AI attack them instead of using underway to bypass them rendering them useless.) because in 90% of cases the difference in casualties will just be wiped out by replenishment.

[Anonymouswizard]

There are some excellent fan modules. I'd go so far as to say some of them are better than Dead Man's Switch. Antumbra chronicles or the caldecott caper are both in that category.

I always mean to try a mage, but sniping is just so satisfying.

Yeah, and I might try making my own scenarios. I guess it's at least partially that an official campaign would guarantee a certain baseline. I'll certainly try out your suggestions.

I would have gone Street Sam, but I was already planning to bring Duncan, who's a bit suboptimal in the role but makes up for it with 4/5 of his non-lethal tree (and the one that is questionable is right next to a boost to Full Auto fire, so I nabbed that instead). Unlike deckers going mage/shaman doesn't shake up gameplay much, mostly it unlocks buff/debuff spells and puts your abilities on cooldowns.

Yeah combat spells are fun, but I've got the most use out of those that drain AP (particularly the lightning spells). Both Duncan and Gaichu have abilities which benefit from stunning opponents, and spells are one of the better ways to do it