Great questions, I also want them answered.

IMHO an important aspect of 6) Initiative is to determine who gets to act first? If two opposing units can kill eachother in two blows, then even +1 can be important. It could make the difference of being able to hit first, then receive damage and then hit/kill (instead of the oppsosite). This is simplified as ranged attacks or movement also affects who gets to attack first - but still shows its situationial importance. Initiative is therefore also related to, or judged by, the opponent's initiative. I ask myself; do I act first in most battles, then my initiative is good and it doesn't need much upping. If I'm often acting last and dislike it, then I need to rethink troops (e.g. less range with lousy Init) or perhaps even take some +Iniative traits when customizing my next faction/Sov. 

Inside your army, the relative speed is also interesting. My main tank (e.g. Bard-Defender tank) I often try to give the +Initiative equipment so he can ride into the middle of the enemy troops (and hopefully draw aggro) as early as possible.  Or I may want my sov to cast shrink early vs a quick-moving bossmonster. So I sometimes moved around the +Initiative gear for henchmen/Sov. A Bard healer can have lower initiative than your offensive troops - so there is something to heal. 

In longer fights I judge +1 initiative for an 20-initiative unit to be worth 5% (1/20). I can compare that with how much +1 Attack brings up the unit's attack, e.g. +1 attack on a 10 Attack unit is worth 10% (1/10).  IMHO this mathematical value is not as important as the relative speed vs opponent's Init - it's more used when caldulating if my Sov should cast Haste or not in boss-fight. I then compare what the Sov can do for damage, with what Haste can improve on the most offensive troop in unit.

Hope you can find some use in my thinking?

Thanks for all the responses.

I also wanted to add an eighth question:

8. What are your feelings about the various unit traits that you can add to trained units? You only have 3 free slots against many good traits so it is a tough choice. Iron Skin seems like a no-brainer to me. In the later stages of the game, Plate Ability would be a must have for the front line units. What are your thoughts on Reap (heals +1 HP per attack)? The traits that increase with the level seem good: Discipline (+1 accuracy and spell resistance per level) and Acrobat (+1 dodge per level). Spell Resistance (+20 spell resistance) and Precision (+10 accuracy) don't increase with the level but offer big up front bonuses. It is impossible to say anything useful about Defender (+10 Defense while defending) without knowing how defending actually works which goes back to question #3.

Good to know. Spell Mastery is the magical equivalent of Accuracy and Spell Resist is the magical equivalent of Dodge.

Well said. Offense is the best defense.

This confirms what I thought defending was. Is it merely attacking that takes a unit out of defense or can any action take a unit out of defense (casting spells, movement, etc)? Defender might have some small usage with an ability like Bash (available to club/hammer guys) that requires you to skip a turn after using but probably can't justify using up a precious trait slot just to get 1 or a couple turns of elevated defense depending on the duration of the battle (can't recall the cooldown for Bash).

I'm not following on the mages thing. Can you expand a little on the bolded part?

Bloodthirsty etc. all work on attack damage, not magic damage, so mages get no benefits from them at all. Trinkets are the key to mages. (+1 and +2 elemental damage available (fire and cold), and +2 Initiative amulet to pick from as well) I usually end up with Constitution (+HP) Dodge or something similar and Fast on my mages since they are lightly armored. The Soldier's boots (+1 Initative) are a very good pick as well. 

https://forums.elementalgame.com/447412/page/1/ I made a mod linked there that adds a trait that gives mages more damage with a trait from conclave buildings similar to Fortress ones. Adds a little more depth to mages, but really they are very good with the whole ignore armor thing and hardly needs any buffs. 

Guard Strike does pitiful damage usually because it's calculated before armor reduction so with armored troops you might as well not hit at all. I'd rather kill something soft to prevent damage instead of tanking it. There are exceptions of course, (Champion Defenders etc.) but dodge builds usually works better (especially with the new Bunny Slippers) This is just my personal experience talking of course, please show us some Guardian Strike/Defender combos Parrot if you have them ^^

The main problem with archers is their place in the tech three. You want armor, Civilization techs, Magic techs, and they all bypass archery techs. The Faction trait archers is insanely good really, (Ram's Horn Longbow -50% armor -4 ini is SICK, add Bloodthirsty to that etc. and they will do more damage than mages) but  when I get pushed army wise by AI I skip archers because I can't afford the research. If you don't have crystal, troops with hammers can do the trick as well with some Sovereign backup, so archers aren't exactly necessary either. Your Sovereign/Champion can compensate for lack of ranged more easily than spending tons of research in a dead end tech tree. 

However, most players do not play Ridicilous+ difficulty like me, so you can probably afford the research to work archers into your build there if you want. I usually end up going for Enchantment tech early because lightning hammers are awesome (and I can trade for crystal if needed with AI) and it's on the way to juicy magic techs. So just from enchantment I get #1 ranged much quicker than archers #2 a very good melee weapon #3 progress further towards key magic techs. This is instead of focusing on archery techs, so the cost/benefit ratio favors Enchantment over Archery.

TLDR: Archers are very good, just harder to get than mages. My most used tech path is Enchantment for Weapons, Civilization techs for research/production/food when I can afford to do it, and finally warfare for mounts and armor once I want to make the end-game troops. After that I could get Archery, but I would rather get +5 and +6 troop size instead of new shiny archer weapons because those upgrades work on all troops

I would say that Reap isn't really worthwhile because it's not affected by the size of your unit, Acrobat and other dodge bonuses are only really worthwhile if you're custom-designing a dodge-based unit (and dodge is a really weak defense mechanism later in the game), Discipline and Precision are decent but largely overshadowed by Fortress bonuses. Iron Skin is good early on for tank units, and can be useful for later-game tank units, or to pull the defense up a little on a weak unit. All of the spell resistance bonuses are generally too small to have an impact when facing a decently high-level caster, and most casters who aren't decently high level are often not much of a threat (exception made for Dark Wizards, Banshees, and Haunters if they can get their spells off), and are usually too specialized of a bonus to be worthwhile. Underdog/Brute are somewhat specialized and unreliable. Finesse makes the damage of your swordsmen and spearmen more comparable to that of your axemen and clubmen on a per-hit basis, and better on an over-time basis (disregarding special abilities, which favor axes and clubs due to the alpha strike). Initiative bonuses are almost always worthwhile. +X% damage or +X Attack bonuses are only useful on units with significant amounts of physical attack, and only as long as the defense ratings on the opposition haven't become significantly larger than the attack ratings on the troops you put these traits on.

On the whole, I'd prefer the initiative, defense, and health traits, and rely on trinkets which boost magic damage for offensive bonuses.

Bloodthirsty, Quick (+2 Initiative),  Finesse - for archers. I try to buff initiative to maximum.

Bloodthirsty, Finesse, Chain/Plate proficiency - for melee units, whom I try to arm with a quick weapon to have their initiative high.

Looks right.

I think it's more or less random what repeatable the AI picks for its research. Honestly, I kind of wish they'd get rid of the Refined Accuracy/Defense/Attack traits, because they screw with the late-game balance and seem more like a mechanic to make the guy who's behind fall further behind rather than a mechanic which really adds anything or encourages an arms race. Repeatable economy technologies are alright, but the repeatable accuracy technology makes dodge impractical (as if it needed a nerf against naturally growing accuracy that can be significantly boosted by Fortresses and design choices), the repeatable Defense bonuses can turn battles into slugging matches that just drag on, and the repeatable Attack technology has the opposite problem. The Attack/Defense technology pairing might have been an arms race if it weren't for the fact that the Attack bonus is slightly stronger in the long run, and if it weren't for that by the time I'm picking a repeatable tech I'm pretty much looking at not resetting my research for the rest of the game.

I'm not sure what you mean. I'm using only half the defense score when calculating the 50% nominal damage point (e.g. Attack = modified Defense => maximum damage = 0.5*Attack, which with "ignores 50% of the defense" means that this comes at Defense = 2*Attack), as I see no other reasonable point to pick for applying the defense reduction. This means that against a 0% resistant enemy, Yew Longbows are on par with the basic fire or ice staff against anything with 11 defense, and worse against anything with more defense, while Ram's Horn Longbows are on par against 24 defense and worse against anything better. Against a 50% resistance enemy, Yew Longbows are equal against 22 defense and worse against higher defense, while Ram's Horn Longbows are equal at 48 defense and worse at higher (discounting attack bonuses of any kind - if you want to account for them, you can go ahead and do the math, it isn't that difficult). This means that the standard end-game bow sucks compared to the most basic magic staff you can give your troops, and the RHL isn't terribly better off. If we switch up to the end-game magic staves, then Yew Longbows are simply worse in all cases, and a 50% resistant target can use the same comparison as the 0% resistance target for the RHL. With the right traits, it can be a different story - bows can be quite good - but otherwise they're kind of garbage (and yes, Finesse + Quick would be one potential set of "right traits").

Also, when fighting lightly armored troops is pretty much the only time that bows are really competitive with staves (especially standard bows - were it not for the fact that staves cost crystal, I would never design a unit to use standard bows unless it were some kind of restriction I put upon myself). Sure, bows get better attack bonuses, and the repeatable attack bonus doesn't affect magic damage, but magic damage is both harder to counter and much more reliable, and can't be trivialized by Refined Defense.

This would probably be one of the better uses of a +defense while defending bonus, yes, but you shouldn't necessarily neglect it for a tank line whose purpose is to protect ranged damage dealers (e.g. a Mage with Water IV and Evoker IV with lots of Water shards, or a bunch of archers, or some other similar things) - it just depends on how you want to play the battles, and what difficulty level you're on. You can make almost anything work on the lower difficulties; higher difficulties tend to require more optimized strategies.

Dense Monsters, Huge+ Maps (custom mod), Epic pace. Resources/Magic are usually normal, I feel the lack of them makes the game more interesting.

I meant that the damage bonuses is counted on their own. So Enmity, Bloodthirsty etc. and a +3 Physical attack skill will count against a 50% reduced armor. Not be reduced before the armor reduction. You are only looking at the base attack, not the bonuses, a RHLBowman will get +2 attack from Fortress +3 Attack from a skill (Brute or Finesse), meaning you will have 17 base attack, with full bonuses you can increase that to almost 30. I have done both elite mages and bowmen in real games, RHBows are sick and can compete with mages end game especially in ranged vs ranged fights. Mages are far superior and easier to get early/mid game, but RHB will get better later, and fire faster to boot. If you want to see some really sick ranged, get a Great Wolf to support them and it gets silly

@McBeef:

If you can get a regular unit's physical attack score much over 30 on a per-figure basis, you're either playing a very different game from what I'm playing, or you've researched Refined Attack way too much. And the Attack score used in the damage calculation is per-figure, rather than the combined attack score shown on the unit overview (though it does roll for damage for each figure in the unit). So a five-figure unit with an attack of 100 has 20 attack per figure, and will use 20 attack (+ bonuses) when calculating maximum damage, and each figure in the unit rolls to hit and to deal damage (everyone who hits contributes to the damage dealt by the attack). A better plot would be max damage vs defense at specified levels of attack, unless you also wanted to compare champion damage.

@sjaminei:

I still don't get what you're saying. Are you implying that physical attack bonuses ignore armor? Because that isn't in line with what I've seen.

What I've seen is more along the lines of

max_damage = (Base_Attack + Attack_Bonus)^2/(Base_Attack + Attack_Bonus + Defense*[1 - armor_penetration])

Yes, bows with proper traits can become relatively good, even perhaps better than staves if the AI hasn't boosted its defense up too much. But staves are much more reliable, and if the AI happens to decide to research Refined Defense instead of whatever its favorite Refined Tech of the Month happens to be, bow damage quickly falls off, and even at 30 Attack a bow really isn't that much better than a stave, because if we're looking at relatively end-game armor we're probably looking at 30-60 defense, bringing the 30-attack RHL maximum damage down to 18-14, which isn't terribly better than the 12 base of an end-game staff. Granted, it's better than the maximum damage of the staff, but not terribly much. (I'm also not counting trinket damage, because presumably if you're trying to maximize the attack you'd have damage trinkets on both unit types). If we throw in high-difficulty AI bonuses and it happens to decide to research Refined Defense for its repeatable rather than whatever it chooses currently, it wouldn't be unreasonable to see defense on the order of 300+ at the same time you can field RHL archers, in which case the RHL becomes considerably worse than the Incineration Staff or its cold damage equivalent, even if you're targeting 50% resistant enemies. Certainly, if the defense ratings stay at a reasonable level the RHL can be better than the end-game staves, but if the AI has bonuses that allow it to pull ahead in Refined XXX, who cares about physical attacks.

(Side note - Fortresses can give you up to +4 physical attack. Sacrificial Altar gives you 1, Weaponsmith gives you 2, Great Arena gives you 1. Sadly, Weaponsmiths are 1 per faction, Great Arenas require level 5 Fortresses, and Sacrificial Altars are Empire-only, so it isn't that practical.)

Thought of a few more questions:

There are a couple city spells that boost the stats of the military units that are produced in the city, like heart of fire or might. In order for a unit to get the stat boost, does the spell already need to be in effect when the unit was first queued or can it at any point prior to finishing the unit? Knowing this has two benefits: if you simply forgot to cast the spell on the city and the unit is already queued or if want to wring out a few more turns of mana production with meditation.

Some bonuses are based on percentages like the bloodthirsty +25% attack or the Ironeers +25% defense. Do bonuses based on percentages round up, or round down, or the nearest integer? For example, if you have a bloodthirsty unit with an attack of 6, and 1.25 * 6 = 7.5, so does the attack round to 7 or 8? What if the attack was 5, 1.25 * 5 = 6.25, so does this go to 6 or 7?

The stealthy trait makes monsters less likely to attack. Is this effect strategic or tactical or both? In other words, does stealthy make monsters less likely to attack a stealthy scout on the strategic map or less likely to attack a stealthy unit in battle?

Another non-combat question, but doesn't really merit its own thread: are there hotkeys for next idle city, next idle unit?

Yes, you can use "Tab" for that. It cycles through both inactive units and cities without production. (I can customize a few buttons on my mouse and for LH I use Escape /which is useful in almost all games/ and Tab.)

 Anybody know what other hotkeys exist?