I was easily able to smack down the AI.  For some reason, the AI units being sent home are not being put into the city, they're outside the city so their sovereigns can't defend.

Also, in tactical battle, the AI unit isn't using spells.

Note: At the basic level in tactical battles I would like to see the Sovereign AI;

a. Use Wind Shield if enemy army has archer units; if not then
b. Use Touch Entropy if adjancent to enemy unit with HP exceeding 50% of Intelligence and this touch will slay the unit
c. Use Spell Blast (assumes that researching this spell is a priority) with 30% damage if damage from this spell is greater than Arcane Arrow; if not then
d. Use Arcane Arrow (50% damage)

Note: In playing WOM I find that an army with a Sovereign, three archer units and two melee units is unstoppable as the AI is not programmed to counter archer units and it will frequently chase my melee units about the map while ignoring an adjacent spell casting sovereign.

Sure.  Also, I don't think v1.4 will be the last update I work on.  I am starting to think about a v1.5 also.  Just depends on where FE is in terms of needing me.

AI thought. If HP (Human Player) does not have (archers AND spell caster) AND AI has (spell caster OR Archers) = AI units should wait for enemy to approach, attacking HP with ranged attacks. This will weaken HP units before they engage in combat.

Please, please, please do this or something like it. I've always hated global mana for making all casters share a pool even for tactical combat. I see magic as a strenuous, difficult task that can exhaust a mage. I like the idea of a mage throwing everything they have at an enemy and running out of power. That seems much  more interesting and entertaining than a mage standing there a shrugging without any mana because some other major on the other side of the realm cast a spell.

I like having to manage mana as a key resource, just not globally for all spells.

If there is a v1.5 I would cast my vote to have it focused on adding a greater variety of tactical spells that the AI will use. This would go a long way making battles exciting and less cookie-cutter.

Example: Enemy AI calls forth a giant eagle that flies over your armies to attack your sovereign, disrupting his spell casting abilities until the Eagle is slain. This giant eagle can only be attacked by his target or ranged attacks from other units (i.e. you need a bow to shoot a bird out of the air).

My idea:

The channeler is the one who can cost globally- the only one who can cast strategic mana, which is what your mana points are spent on  (INT can affect strategic spellcasting power)  It takes a channeler to draw mana from shards, and channelers can grant attunement to magic.

INT should determine tactical spellcasting ability, from local essence not tied to shards.  Other mages can be influenced by shards, but cannot cast globally or draw from shards, which is needed for strategic spells. 

Mechanically: this means the following.

-  The mana pool is only used for strategic spells. 

-  A mage's int determines how many spell points he can use in tactical combat

-  shards  do the following: increase power of relevant spells, grant global mana for strategic casting.

- INT does the following: determines tactical spellcasting pool,  magic resistance

If you wish more complexity, I'd add these rules in:

Channelers gain a spell which allows them in tactical battle to replenish their mana reserves, at a cost of strategic mana.  The AI would use this spell whenever threatened (cost would be equal to the channeler's INT)

I posted this in the FE thread, but I think this system would work best for WOM.  It would require a reworking of mana costs for all tactical spells, though strategic spells would be unaffected.  You may also need a differentiation between say, a strategic summon and a tactical summon. 

Agree with the others that balance and Tactical Combat are the 'big' issues.

Glad to hear that a bell curve is to be used for calculating attack/damage rolls.This is a great move as long as the specific means and variances are adaptable.

Will modders have the ability to specify the mean and variance for specific weapons. Would be really really cool to be able to have weapons have different bell curve means and variances against different armour types etc. That would make the unit design choices much more strategic.

Personally always thought that artificially normalising hte random number generator with a bell curve felt like a bit of a cop out. I think you'd solve the problem a lot better in the long run by making characters and monsters behave more like units and rolling multiple dice per attack. Ie, so a hero with attack 10 rolls two ten sided dice and adds the results. Multiple attack dice are then added as the hero levels up. Every other game system, whether it's d&d or warhammer, does this; having one attack roll that gets ever huger and huger is just silly.

This way you' remove the problem of having to give heroes and monsters enormous attack values that are competetive with units who make lots of moderate strength attacks, AND you'd make damage results less random due to the use of multiple dice, instead of doing things "under the hood," which seems shady to me.

But I suppose it's still an improvement.

Hmph.

I had a big long response put up yesterday and it disappeared.

Let me summarize where things are as of this fine Monday morning:

The focus of v1.4 is:

1. Fix lingering stability issues. This should be something that is very apparent even if I released an EXE today. The game shouldn't crash. Ever. On any system ideally.  I found that v1.3 was effectively buggier because it ran a lot faster.  v1.4 should address this.

2. Fix bugs (see Heavenfall's thread).

3. Fix balance. Specifically, the combat imbalances, prestige imbalances, and training time balances.

4. Fix things that prevent the AI from playing well. Specifically, the path finding which was aborting quite often. Hence, AI wouldn't attack you, couldn't send units back to its cities to reinforce, couldn't expand as well as it should.

5. Fix sovereign behavior (this was a bug -- the combat manager was overriding the sovereign manager). Hence, stupid sovereign bouncing around the map.  Still have to get him to go get goodie huts better. Very frustrating.

6. Make AI perform better with magic in battles.  Sorry on this one, this is going to be tough to do in WOM as well as it should. The code is good (kudos to the Stardockians who wrote it) but I'd need a lot of APIs written to make the AI do better in tactical battles and all the coders are on FE.  So I can improve it quite a bit but I'm not going to rewrite it.

Anyway, in the long term, I hope to make more of this stuff DLLs (not in WOM but in FE) so I can hand stuff out to modders like I used to in the OS/2 days.

I think v1.4 will be a big improvement for most people.

They do now.  The code was there, just the logic was hard coded.

Here's the first reason : i'd like to point out that you can also have half a chance to obtain a roll between 15-16 if you add 14 to a D4 roll. My point is just this : On one hand you use a bell curve to reduce the randomness of your roll, while with the other you use an rocket-propelled variance which completely shots down your efforts to get more predictable results from your rolls.

Does this look like an intelligent point ? I hope i doesn't but i fear it does. Because it's still missing the most important mark : we're looking at one roll. While we ought to be looking is the random function resulting from all these combat rolls. And actually, it's the only thing worth looking at.

Here's roughly the idea, that's how it works right now :

-weapon and armor are of the same value there is more than half the chances to get "blocked" result. The higher that value is, the closer you get to half chances of getting blocked. It no longer resembles a gaussian distribution in any way, btw.

While weapon and armor are in the same area, getting more or less weapon and armor got a huge impact on the expected damage is tremendous.

If weapon value gets lower expected damages, are shot to nothingness.

If armor value gets lower the expected damage reduction from armor gets slightly less than armor value (translation : the difference between a lord's hammer expected damage on a defense 0 and a defense 5 is almost 5).

That the kind of figure we're looking at. Or we should be looking at. Now you tell me where you want to go, what edge do you want to give to technology/ranged weapon, what HP you want (because there's still the solution to gives units sky-high HP so that combats use so many rolls you get your bell curve back) rather than tell me "hey, i replaced my normal distribution RNG with gaussian RNG you'll see more mid range results"...

No, we're not.