Right. I keep forgetting...

I did try out one of the new binaries that Frogboy had posted. I have to see it is a very large improvement.

Map Size: Medium

World Difficulty: Normal

AI Difficulty: Normal

Number of AI: 9 (I was pariden,  they were the other factions).

So i ran into Yithril on about turn ~30. I normally would have attacked at this point but the leader had 7 att and 3 def at that point. I just had 5 or so attack on my heroes (The enemy leader had like 30 hp so I wouldn't have faired well). At this point I decided to return to get some armor (Padded at this point). I returned and declared war. I noticed that Yithril had not produced any soldiers at this point but instead spent their money on equipping their sovereign and spouse. I ended up attacking and having to retreat because the Yithril sovereign is a beast.

Here are some city building notes on the AI

Yithril

1. They have used 3/10 available food from their capital. However, they are pop capped at their capital. Instead of building a house, they are building workshops.
2. Yithril did build 1 monument and the tower of dominance. The effectively gave them 2.4 prestige with their leaders garrisoned in the city (yay).
3. I would have expected at least 1 soldiers by turn 50 (when I launched my attack). Even if Yithril is focusing on the champions, they still would likely have materials left over (Unless they built a really late workshop).
4. They were building pioneers to settle gold and marble just outside of their influence. While this is normally a great move, when at war, colonizing should have a much lower priority.

Magnar

In my offical game save, I haven't explored them yet. However I did reload an earlier save while on their border so i could compare their city building technique. What I discovered was the following when comparing my city with theirs

Resource     Pariden     Magnar

Gold               9             8

Tech               3             2

Arcane            0             2

Prestige          2             3

Magnar's capital has a higher population than mine and is growing faster. They are getting the royalty bonus so he did find a prince somewhere in the wastes. Magnar's Queen was wielding an Oak Spear giving her 11 attack, a lot higher than my band of adventurers (at this time I had not gone back for better weapons yet.)

All in all, this is a vast improvement over what was released with 1.1. Great job frogboy. I had to retreat from a fight on normal difficulty (I normally play on ridiculous....the extra hp in this battle would have been RIDICULOUS lol). And i'm not exactly a 4x slouch (I've beaten Gal Civ II on crippling).

Anyways, have a great weekend

Jec

And where can this .exe be found please ?

much the same as others:

the ai doesn't build groups

it doesn't go for the right technologies first like a human player: food increases, housing bonuses, groups, armour and weapons. perhaps we're manipulating the game because certain techs are essential (overpowered), and in the long terms this might mean a rethink of technologies, but in the short term, it needs to learn to go for these techs and build those groups.

it builds too many settlements too close together when there is no resource advantage to doing so. it is rarely wise to build a settlement when there are no resources there. if there are two resources close together it needs to learn to leave some settlements as undeveloped resource grabbing outposts, then raze them one neighbouring influence is sufficient to cover them.

it build too many pioneers altogether. i'd suggest a reactive "there's a resource: build a pioneer" approach rather than the other way round, but can see how this would be tough

elemental right now is food, housing, population, money, troops (so long as you've covered your resources) and this is the way the ai needs to play.

It was download from Frogboy's dropbox account. I don't have the link anymore because mIRC doesn't have a chat history, sorry.

So I have a few more things to add about the new AI experience. First off I mistook Yithirl for Umber.

Altar

1. Very aggressive. conquered roslen and Kraxis.
2. Aggressive Unit Designs: Alter will have extremely high offense (21 mid game using a warstaff) and very low def (only using a light plate cuirass).
3. Topless Heroes: Ok Relias; we get it. Your a tough guy. Stop sending your sons/daughters into battle without shirts on. I'm not sure why but Alter's heroes are under equipped.

Umber

1. Recovery: They were able to found a new city with the soverign after I took their capital. I had not seen this behavior before with the AI. In the past, once they were out of cities, they were toast (Stuck at 1 power). Now Umber has stacks of units on my boarder (he's at war with Magnar and i'm in the middle of them).

General Thoughts

1. No Passage: The AI Will declare war on another faction based upon principle but with little to no regard for logistics. In this game, Umber and Magnar are at war. However, I control the only land between them (I haven't seen the AI use Raise Land yet). As a result, I have seen stacks of units stand on my border.
2. Unit Design: Alter desgined some units using only a chest piece of armor. If he had used all light plate, I would have had a problem on my hands (I had warstaffs but only leather armor at the time). So I am going to list some guidelines that I feel would help the AI design better units

• The Soldier should have all armor "slots" used (helmet, braces, leggings chest piece etc).
• The only time a soldier should not use the slots for armor, is if there is a magical item (traveling boots) that makes the design more complete.
• I'd image Alter in my game was using only light plate chest pieces to save on metal. However by not equiping the helemet, bracers and leggings, I was able to dispatch those soldiers with ease. I would either use leather armor for the remaining slots or all light plate but build less units total.
• Variety: The AI tends to build only the best units it has available. I've seen several times where Thousands of materials are laying around and because the AI has no metal, it has stopped all production. The AI should keep some Material Only designs so it could use this material. After all, a company (12) soldiers with a warstaff are going to be able to do serious damage even to plate users (the attacks sum in a group IIRC).

3, Trading: In Gal Civ II, trading gave a + to relations. It would be nice if this occured in elemental as well.

4. Stacking: The computer is getting better at stacking units. In my current game I did notice a stack of 8 "observers" from magnar. I would continue to encourage this type of behavior.

btw, i personally think the AI would do better with a bunch of premade units. i don't think there's really enough variation in the system at the moment to warrant unit design.

No problem, I asked in the IRC and someone gave me the link. Thanks for the reply though

Played for about 4 hours with the new .exe. Observed about the same behavior as you did.

I agree. Until more variety in equipment comes out, premades would reduce the risk of odd behavior in AI equipment.

I have some suggestions on how to add some flavor to the game. Its incomplete (in a balance since). https://forums.elementalgame.com/402521

With this post, I tried to point out that we could have 10 or so weapons that would continously get better (tiers would be mat used). Each weapon would have strengths or weakness (passive stat changes).

Kenata's Weapon Mod is also an excellent example. https://forums.elementalgame.com/400876 Her(?) example gives 1-2 abilities to each weapon. I think this is an excellent idea and if combined with tiers of mats, an excellent combat system would be created. This way the "Spear" could always be effective so long as you have the "up to date" version.

For those who have not tried this mod, I would highly recommend it.

Edit: Improved Grammar (wow it was bad).

Played the 11a binary. Still no AI squad building.

Looks like i'll have to pass this game again . 

There is a marked improvement with the AI. Seriously.

Apart from that, the game is actually quite fun to play now. You find yourself making actual decisions! If the two expansions address combat, the magic system and diplomacy, this game is going to have support groups set up for addicts.

There's going to be a lot of AI updates over the Christmas break.  I put out 1 on Friday. I took this weekend off to do other stuff but tomorrow I"m back on it.

Is there a reason why no one has posted a link in this forum or somewhere?  I'd like to try out the tweaked AI as well.

yes, one of the problems with weapons in the game at the moment is the lack of distinction between weapon type and weapon quality. it's not just a problem with unit design, but also with adventuring: the magical sword of the sovereign for example, is inferior to many similar (yet mundane) weapons that the character can research by the midgame.

spears for example, should always be a great infantry weapon even if they are a mundane tech. their quality comes from the fact that they are great defensively, especially against charging units. similarly, a lance is always a fantastic weapon for heavy cavalry, allowing the momentum of a charge to be focused. the base stats of weapon types need to be devised and hard coded to make them effective in the situations they're supposed to be effective in (i say stats because in my opinion, real game mechanics are much better solutions than gamey special rules that people have to look up). you can then start making magical versions with slightly superior stats, but their role remains the same.

stuff like

early tech (primitive):

dagger: low damage, iniative (attack speed) bonus, but largely ineffective beyond early game

spear: low damage, low cost, strikes first vs. enemies charging from front/front diagonals

crude bow: low damage ranged weapon

moderate tech (roman era):

short sword: moderate damage, small charge bonus

short bow: moderate damage, equippable by mounted units

high tech (medieval era):

longsword: high damage, charge bonus

longbow: high damage

lance: cavalvy only, high damage, huge charge bonus

late tech (late medieval/renaissance):

pike: moderate damage, two handed, strikes first vs. enemies charging from front/front diagonals

double handed sword: two handed, very high damage, charge bonus to damage

this way almost all weapon types remain effective throughout the game, yet there is plenty of scope for magical/improved versions of the same type.

The dagger could give the unit a damage bonus if the unit attacks the target from behind.

Frogboy, are we getting a new executable today, wrapped with a cute wrapping ?

There needs to be some kind of mechanism for distribution here.  It's frustrating to read about a exe with a tweaked AI without being able to try it.  I know it's not an official release, or even an offical beta, but I for one would like to try it anyway.

Maybe like an SVN server with an anonymous login or something?..

Hi Frogboy,

great to read you work on AI updates. Will we get these AI updates with v.1.2 in january or earlier?

If we see any movement on Elemental before January, I'll consider it a gift, or a Stardock employee with a mostly tested build and tons of free time over the holidays.

So the flavor of the week is to complain about the lack of access to internal builds that are not meant to be seen (even less played) by mere mortals like us?

Eh? No, I'm saying they're mostly on holiday after the big push, but they've likely been doing something since 1.1. So there might be changes going live between now and January, if it's really minor stuff and someone is still around to roll it into a new version.

No, no...sry. No complain. Just a question. I´m very pleaced with 1.1. And SD deserves holidays...

This and the AI needs to update its logistical rally points (used to form stacks of units from single units).  Basically the AI needs to have a front line.  If I break the front line it needs to re-establish a new font line faster than I can stay in its backfield picking off other units tryiong to group up.

For me, it is touch and go until I've research 'Heal'.  Before heal is research, getting killed by a group of monsters is really the only problem - if i paid more attention it probably wouldn't be a big deal.

Some stuff about my games:

1) I always play on the largest map size with the most AI, I can't remember difficulty, but nearby brigands and bears have 30+HP at start, so not the highest difficulty by far.

2) I cannot for the life of me remember the last time an AI unit (faction) initiated a war with me.  When I find them, they are always messing around - sometimes I have 3 cities and I can see they've just founded their capitol.  I just declare war and move into their capital - same turn usually.  If they have a big "influence bubble", I'll send in a caravan before declaring war so my army can run up the road into the city.  The enemy sovereign is usually nearby and alone, I kill them next turn.  I've maybe got two peasants wearing traveling boots by the time I start wiping out neighbors.  I may or may not have hired a traveling companion.  I'll kill a couple, then start ignoring the rest until I am a huge overpowered beast latter in the game.

I think the incredible variety of tactical options in the game makes it really difficult to program the AI.  For example, research of military tech seems a high priority for most "steam rollers".  I almost never do any military tech research until later.  Since no one ever tries to invade me, I just wander around doing whatever I feel like until I have so much tech, I make a 16 speed maxed out group with best armor / shields / magic, then start roaming around the map slaughtering.  It is weird that agility can be researched continuously...  I cast beserk on that group once and did more that 2000 damage!  It was sorta awesome.

Anyway, short list problems for the AI (my lame opinion of course - my mom says I'm smart... and handsome!):

-Players are untrustworthy - trade routes don't mean "I want peace" they mean "I want road to your capital for my army".

-Cities are distributed, travel is awkward.  Easy for human to gather troops for precise strikes against distributed foe.  AI needs a way to be more efficient than human at building / managing troops, since they will never be able to compete with pinpoint attacks.  In a game like this, with maximum thinking time available, AI needs to be a hardcore cheater.  Builds things faster, and they cost less to maintain. 

-AI seems to want to team up with player to kill other AI.  Does it also team up with other AI to try to kill player?

-AI do not seem to try to win game.  There are multiple win conditions.  When I first started playing I thought "oh no, while I'm fooling around with these cities, some AI is researching the master spell, or going on a quest spree for master quest ending".  It never happens.  They need to try to win, not just be flavor.  Some endings are pretty close.  If select AI went hard after specific ending, human would have to struggle to try to stop them, especially if they were a continent away.