Fallen Enchantress - Legendary Heroes (1.2).

New Tactical Battle Maps. Battles are now much more intense with far more special skills available and initiative being far more important. New combat mechanics - such as swarming increase the strategic options.

Sounds good, so what's the problem?

Let me start first by saying that overall it's a great game, or I wouldn't have spent 90 hours on it altogether so far. That said:

It's Season 476. I am at war with Capitar, the sole remaining enemy faction. Their power is 810, mine is 285. Not only that, but Capitar has twice the number of cities, more income, more research, etc., etc.

Capitar is on the attack, loaded for bear with 8 armies en route to Fort Esiricin (a good strategic choice, in theory, as it's where I'm building all my troops).  5 stacks have already entered my territory, with a sprinkling of heroes, all better than my equivalents. Capitar's main builds are: Marksmen (initiative 16); Knights of Calder (initiative 22, att 72, def 10); Militia (armed with mauls; initiative 14, attt 144-168, def 24). The amies also include a few catapults. Most units have 6 members.

The plucky defenders comprise a sole slender stack: Ruler (lvl 15, Death/fire Archmage) Initiative 29; Lith Serratta (lvl 14, Defender) initiative 29; Ascian (lvl16) Initiative 30; Mounted Archers x4, (5 members, lvl 6-7, Ramshorn Bows) Initiative 24; Magush (lvl 10); last but not least, 1 Fell Dragon (lvl 13).

I can start by casting freeze, wither, pillar of flame, tidal wave, to weaken the stacks a bit.

The battle  tactics are so simple as to be non-existent:

(1) Dragon flames 4-5 Capitar units at the outset.

(2) Ascian mauls the nearest group. Doesn't matter which.

(3) Ruler casts mass curse (She has -1 time buff).

(4) Lith Serratta rests (or could cast haste, it doesn't matter).

(5) Magush rests -just along for the Exp. (or could cast Slow on a big-life enemy hero).

(6) Twang twang twang twang, the archers pincushion the remaining enemy.

(7) Dragon tail swipes the wreckage. (8) Ascian finishes off anyone still standing; if necessary, Ruler finishes the job with a flame dart or flame wave.

No enemy troop ever has the chance to use their splendid weapons.

Frankly, this is a serious AI flaw; why doesn't the AI (particularly one with such a huge lead in development) build at least to some extent to counteract the rather obvious tactic that a player with the 'archer' skill is going to build a lot of fast archers? And why didn't Capitar either build a dragon on its own dragon's lair (which I found much later), for a measly 250 gold, or at least provide some of its troops with fireproof cloaks? Better still, why not fan out across my lands to 3 or four different targets? If AI knows the best enemy units are concentrated in one stack, why not send, say, 2 of the 8 to four different undefended cities?

It's worth contrasting what would have happened in, say, AoW2.  First, AoW2 allowed stacks to 'gang up' so allowing a counterattack in massive force and from several directions in response. Second, AoW2 had a maximum turn mana use limit - not so important for the cursing, which only costs a few mana - but it would have made the freeze, wither, pillar of flame, tidal wave, etc either/or alternatives (compare poison plants, fire storm etc.) Last but not least, AoW2 attacked several strategic targets at once (and there was no 'cloud walk', at least until teleportation gates - a huge investment - were completed.) 

If there's one thing in FE-LH that needs a major overhaul, it's strategic and tactical combat AI.

There are so many strategies the player can use in the late game I'm not sure it's going to be feasible to come up with good AI responses to all of them. What I think could be improved is that the AI makes better use of what it's got. So yes, it should be building dragons etc., it should be focusing more on mounted troops with better initiative and the maximum amount of buffs possible, and it should be making much better use of tactical spells. But to be honest I don't think its troop builds (with the possible exception of the militia) were that bad. It could certainly focus more on buffing mounted troops to give them the best initiative (through magic items etc.), but what conventional army exactly were you expecting to stand up to three heroes average level 15, average initiative 29, plus a dragon? I struggle to think of an AI strategy which would beat that.

As you say, there may be scope for improving the use of cloaks and Nature's Cloak, and there may be scope for improving target selection when it can build a lot of stacks. This could lead to the alternative flaw that it doesn't concentrate its attacks well enough, but on the other hand, I'm not sure there are many targets which you can get with five stacks which you couldn't get with three. So maybe you could put a limit on the number of stacks it sends to attack one target, so it spreads its attacks out a little more. A three stack limit per target would mean it attacked three different targets in your case, which would presumably stretch your defences more.

But I don't think it would change anything very much. Even if it had bought a dragon, had max buffed troops, had a sovereign who cast good tactical spells, and attacked three cities simultaneously, I imagine you'd still win, it would just take longer and you might lose some units, burn more mana and have to run around a bit more. Unless you are resource limited and can't replace your losses after the AI attacks, you should still win quite comfortably.

Don't get me wrong, I'd prefer it if the AI provided more of a challenge when you're fighting epic battles against the strongest enemy faction in your game. But on the other hand, if you have a Stack of Doom, I don't think it should be a great surprise if it kills everything.

Well, yes, most of my strategy revolves around building a Stack of Doom. I know some similar games (Civilization) have tried to make the superstack a less obvious strategy, but when the entire game revolves around heroes, dragons and super spells, I don't think that would be either easy (or particularly desirable, for that matter).

Off the top of my head, the AI would have to be able to beat:

1) Summoned Skeleton Horde

2) Summoned Lightbringer

3) Fireball

4) Heart of Fire etc. buffed troops

5) Titan's Breath and similar spells which knock units prone (ok, that just means mounted units)

6) Summoned Grave Elemental (gives Death Ward to surrounding units)

7) Dragonbreath

8) Blizzard

9) Horrific Wail

10) Mass Curse

11) Ranged staff troops

And that's before considering basic Haste/ Slow/ Heal/ Shrink/ Growth strategies and strategic spells. A human player is always going to be able to work out some way of building a stack of doom when you have such a rich magic system. I mean, high initiative, warg mounted and buffed ice staff troops + a dragon + a tactically summoning hero + a troop buffing hero + a Fire/Death mage with damage spells... how exactly would the AI stop that?

There still seems to be plenty of scope for the AI cloaking its troops more, buffing its troops better, casting better tactical spells and co-ordinating its attacks a bit better, and I hope that will happen. I don't think the AI is ever going to be able to create and use a Stack of Doom as well as a human though.

On a separate note, I agree with Frogboy's stance on not placing high priority on player exploits.  If a player is saying "Why isn't this game harder...but I just can't help myself from training 3-man units and then upgrading them to 6 to save oodles of crystal etc..." then that player simply has a problem.   

And as for this:

You can avoid even getting yourself in a situation where your cities stop growing if you make sure to spend time researching the appropriate techs, and building the appropriate improvements in your Towns.  Most likely, you have it in the back of your mind that "Oh, I can put off building that granary because I can just queue a pioneer later if I need to..." so you put it off, and you research Blacksmithing instead of Agriculture, and guess what?  Your towns stop growing!  "Whoops, how did that ever happen?  Oh well, I'm sure I'll need a pioneer at some point..."  and so on...

I'd definitely rather see improvements made in the actual AI, and players left to their own self-discipline regarding intentional exploits in a single-player game.

I think with the current level of game development talent it's easier to change the game design so an AI can be written to play it rather than to write a competent AI for a complex game

whether or not that's good or bad for enjoyment depends on personal factors

no 4X game of this complexity has ever had decent AI. I don't know what people consider to be smart opponents, but even if it's quite good relative to other games (eg. Gal Civ 2) it's still extremely weak

Dude...it's a game.  And a SINGLE-PLAYER game at that.  If it FEELS like an exploit to you....don't do it.  If it feels like trying to win....well go ahead.  I for one have NEVER ONCE trained a Pioneer simply to keep my cities growing.  Nor have I trained a 3-man unit with the intention of upgrading them to a larger unit to save resources.  I just don't feel the need or desire to do so.  I DO save frequently, and reload if something goes drastically wrong.   By some people's rules, that would be cheating.  It doesn't bother me though, so I go ahead and do it.  Such is the freedom of a single-player game.

Moreover, regarding the pioneer thing, I'm not sure it's an exploit if you actually BUILD the pioneer.  Because then you are actually losing the population from your city.  If you CANCEL the pioneer after you've moved it to the back of the queue and built the necessary improvements to get more food, then it's an exploit.  And that type of behavior would never come about over the normal course of gameplay.

Anyway, this thread is supposed to be about the AI, not exploitative players, or self-discipline.

For me the problem with LH is that games end way too fast (disregarding the AI part). Playing on a huge map, I barely explore half of it and get past mid tech research, and the game is over (AI wins by map control, playing vs 10 opponents, we're left 5 when this happens).

My research pace (on expert world / AI) is on par with AI's, my MP is top mid of ladder, my economy / production / research are all up there, yet it seems like they play their own game without me being able to do much about it (even with all troops mounted, buffs and items you can only move & fight so much in a number of turns, even with no casualties).

I can't help weaker AI's against their attackers as I can't send them troops, give them mana or spells, no matter how many resources I pump into them, they won't be able to put a fight and stay in game. I can't have 20 armies of similar strength going in all directions to cut off the swarm of AI troops. If I make it to their area of control and start taking cities there, they just take them from the weaker AI's.

Another thing, the AI seems to disregard completely the distance between the player and it's own territory. They make DoWs (based on faction power) even if by the time their armies will arrive in the vicinity of my territory (if they can get past other AI's / wildlands / lairs etc) I will have researched and upgraded everything I have to the point he would get 1 shot.

Heh. I think I summed it up best by saying we disagree. I see your point, I just don't agree with it.

To give you some idea of my perspective the most fun I've had in strategy games in the last 5 years has been playing Civ 4 and Civ 5 game of the months from the CivFanatics website. Preset maps and starting positions which you can download and everyone tries to see how quickly they can win by the nominated victory method. Its great fun and all sorts try it, typically the fastest wins are around 200 turns but there are usually others who took 300-400 turns who still happily submit their saves and chat about the game. There are a few exploits which some of the fastest players use (mainly extorting money out of the AI at mid-high difficulty levels in Civ 5, although one could argue that is just a very effective strategy) but by in large Firaxis have closed most loopholes so it comes mainly down to strategy and tactics (and a bit of luck).

I'm not sure it would be possible to run similar GotM for LH simply because the number of exploits are so large and have such a huge effect on gameplay.

I think exploits are related to the AI. If there are lots of exploits then that works against the AI and makes it harder for the AI to provide a challenge. If Civ 5 had the number of big exploits LH does then the fastest people would be completing those GotM in 100 turns and complaining that the game was always too easy.

Some further thoughts based on another three games worth of trials and some of the above comments.

(1) The AI can cast area effect spells quite effectively when it has a large enough city to protect. It could do this more aggressively however. A further expansion of the same theme might be to have global effect spells continuously in effect throughout the whole of a realm's lands - e.g. a kingdom-wide Slow, decreasing the invader's initiative, or an empire-wide Curse reducing enemy armour to zero - anywhere within the borders of a particular realm. A variant would be to allow City defence spells to be cast without having to have an essence slot as already happens with city curses.

(2) Another way to mitigate deficiencies in AI might be to have more scenarios (when triggers can be preset) as opposed to stamps or maps.

(3) It would be helpful to be able to bolster allies' or friendly powers' defences by transferring units/cities to their control - particularly with Athica or Pariden which have useful special items in shops.

(4) Thinking in terms of Galciv 2, what would a Darklord faction look like? Super wraiths? Deorcnysse with arcane forge and 3 good buffs (e.g. initiative, fire damage and defence) could produce hordes of ferocious units even without access to ore/crystal.

The thing is from ur story is if you had lost that battle you would have probably quit the game because that was everything you had good.

Not at all. I've spent many nail-biting hours fighting off the hordes on the tougher difficulties in Civ5 for example, where just surviving to the end is a challenge in itself, or for that matter reloading time and again in some of the Wesnoth campaigns in the hope that some of my troops will get that all-important level-up, or for that matter spending 300 turns in a cave in the final level of Shadow Magic whittling away at the Shadow Lord stacks till it's safe to come out.

The problem here is that FE-LH is nearly at that level, bit not quite, because the AI basically never wins a battle. So if one is asked about AI love then my answer at least is that strategic and tactical combat AI is the area where a bit of affection might go a long way.

Civ5's qualities are not really the ones at issue here. I had posted this thread on AI issues here, I'd love to three quarters of that get some love.

EDIT: Argh sorry, didn't see that this thread is from the summer and had just been reanimated with necromantic powers. Sorry

Someone hasn't played Civ 5 recently...

Having been a bit critical of the A.I. in the past, I did feel at least obliged to play through 1.4 after the patch. To be fair I also tried the hard level and picked a leader/race without the 'cheap' buffs. In the meantime I'd also acquired the then 3 DLC packs*.

Introducing: Artamainen of the Kingdom of Norroway. Traits: Mancer, Hardy, Brilliant, Cautious, Air Apprentice (just to be certain of Tutelage), Might, Heroic (for extra XP) Light Plate (Master Chain is arguably 'cheap') Master Scouts (obvious upgrade for Mancers).

The other players: Altar (in case I run short of quests) Capitar (random), Umber (random), K'mool (random) (Trog Empire, led by warlock Hrakn; Juggernauts, Death/Fire Apprencice, Brilliant, Disciplined, Lucky, Master Chain, Quick, Tough). Key troops Ironic Bowyers (Fast, disciplined, light armed) Jugggler (Juggernaut with Best Sword/Shield, Banded Belt,Belt of Speed)

The map is Dust Bowl. Noteworthy features: Each player has their peninsula next to a sandy central area which is full of monsters and has most of the resources, esp Iron. There are no horses that I could find - a refreshing change actually, and no dragon statues. Extra monsters, More Random Events selected.

I start in the North-East. I wasn't particularly lucky with heroes (missed Ascian somehow, got  Iriel and Arcturus, plus trusty Bacco). Quite a few random events in the first hours; Blood Moon, Meteor Strike, a lot of skeletons, so basically plenty of pressure around the edges of the kingdom which kept me busy protecting resources for the first 200 turns or so. However the need for iron  from the dustbowl meant that my crystal builds in my own peninsula tended to get overrun by the wilds, an interesting strategic trade-off. 

By the 6 hour mark I've explored about 1/3 of the map including what the caravans to Altar, Umber and Capitar have revealed. My 'main' stack (just heroes with the occasional guest appearance from a sacrifical built unit) is essentially tied down with guard duty. 3 towns (Uxstead, the capital, Hornsby Abbey (no essence, and never able to develop an apothecary, whcih was a setback) and Fort Hedeby, fortunately a powerhouse for building units - eventually - for Hedeby has a linking via outpost with the capital, which also gets razed regularly. The open quests are Open Tomb, Heroes Tomb, and the farmhouse with spies, all a bit beyond the capabilities of my troops at this stage. My main units are militia and Townswomen (light-armoured girls with axes). K'mool is well in the lead (76-vs-63).  There is an utterly dire Death Demon in the south-centre of the dustbowl which rushes out to attack (and wipe out) any units that get within about 3 squares, so I've had to build roads round to the north.

Fast forward to the 12-hour mark. I've found a transport and seen nearly all the map either by land or by sea. Umber has paid for its temerity in always declaring war 2-3 turns after K'mool does, and I've added one of their towns to my set - not without cost I might add. Apart from that the political map is not that different. My main army in the centre can hold off the ravaging hordes of wild creatures and keep K'mool at bay, but their towns are still much too tough. The Ironic Bowyers turn out to be quite excellent against my light troops and the Juggglers also inflict losses because they are so quick. To counter act this, I build (1) Pledged of the Blood God - slow light-plate armoured hammermen who can absorb a bit of damage, and (2) Mages of Niflheimr - light armed mages with a bit of dodge (a successful build, I was still using the same guys even near the end of the game by which time they had a ton of health). I am now ahead of K'mool (137-vs-105). K'mool is stil competently raiding my central outposts though, and also has a habit of leaving troops to guard stolen outposts. I still have essentially one main army guarding the dust bowl.

The game was won somwhere between the 12 and 16th hour, by which time I'd allied with Altar and Capitar and started to grind down K'mool. I used Ascian's collar to capture the Krigoth, a useful substitute. I also used my vassal Kulan to capture Ssrip. Another good recruit, but why are they both so small?  Couldn't they be at least a bit bigger to reflect their putatively 'Large' size?

Other small points that occurred to me: why not add player-altered names to the repertoire of names in-game - at least the first 4 or 5 per newly created race; I noticed that some troops near the edge of the tactical map still have strange lighting effects (especially noticeable with the Mages because I gave them white robes); the game still tends to crash if the cursor is left over an enemy unit at the end of a turn (though not so often as before). Otherwise, pretty crash-free unless reloading the same batle several times (which ppl shouldn't really do anyway).

General assessment: I found the tactical A.I. greatly improved. Kulan tornadoed my army most effectively at strategically inconvenient times which resulted in heavy losses. I didn't see any pointless withered caravans either. Equally, K'mool gave me a hard time with attacks from 2 directions into my dustbowl outposts which were a real handful to cope with. Where I did see weakness was in Kulan's and Hrakn's apparent inability to field improved units in the middle game, responding to what they could see I was building. If lack of resources was an issue, either could have had access to the iron mines I was exploiting; alternatively, the map needs some redesign. Instead I saw the same thing as before, light armoured troops with mauls never getting in a hit. I also noticed that some AI cities were taking forever to build units - 20 to 30 turns, for example.

All in all, a massive improvement in AI in the early stages, hearty congratulations for that; work still needed to keep up the pressure later on, please. Last but not least, please consider adding a few pre-set scenarios with properly managed triggering events.

*(Loot pack is good, map pack is ok but expensive for what it is, quest pack is pretty thin.)