Can you throw some numbers our way as to how damage vs armor works in 1.19? That knowledge is needed for mods, and for the hardcore players. #52 is extremely vague.

Let's say a unit with 12 warriors inside, wearing lord hammers (40 att) are attacking an enemy with 30 armor. What rolls are performed?

If nothing else has changed, then the max value of armor or attack is squared before the roll, then the roll is between 1 and max squared with equal probability for each result, then the result is square-rooted. For attack, and previously for armor, all results with less that half the max values were discarded and rerolled, and this has been removed for armor. 

So for instance, for 20 attack, the roll was between 1 and 400. If the result of the roll is, say 225 (for simplicity I took a perfect square), the final attack roll result will be the root of 225, which is 15. If it had been 226, it'd have been rounded down or up depending on what's in their algorithm. If the final result had been less that 1/2 max - so less than 10 in this example - it'd have been rerolled. This was also the case for armor, but this will be removed in 1.19, allowing for less-than-half final results.

What this means is that the higher the result, the higher the probability to get that result. The probability to have a final result of 1/2 or less (which was then previously discarded for armor and attack, but won't now in the case of armor) was 1/4th: for instance, in the case of 20 armor, you need to roll between 1-100, that's 100 possibilities out of 400.

What this means is that previously, the average result was around 80% max value (I did the math once), and will now be 70% for armor. That's a change, but we'll still see a lot of "blocked" results for similar tech equipment clashing.

Edit: Ah, so the low results are not rerolled, they're raised to 1/2.

Lets say the warriors are level 3, that gives them an accuracy of 24.

Lets say the enemy has a dodge of 12.

Each of the 12 warriors makes an attack against the enemy.  They roll 1-24 (their accuracy) to hit.  The Enemy rolls 0-12 (its dodge) to dodge.  Whoever gets the highest number wins.  The defender wins ties.

Of the 12 attacking warriors lets say 8 hit.  The other 4 miss and are no longer relevant for damage.

The 8 that hit go through the following to determine their damage.  They square their attack to 1600 (40^2).  A random number is selected between 1-1600.  Lets say they pick the following for the 8 hits:

1400

1300

1200

1000

800

600

400

200

Then these numbers are square rooted to:

37

36

34

31

28

24

20

14

Damage can never be more than half so those numbers are converted to:

37

36

34

31

30

30

30

30

Now the enemy gets defense against each of the attacks.  His defense is squared to 900 (30^2).  A random number is selected from 1-900.  Lets say the following are selected:

800

700

600

500

400

300

200

100

Then these values are square rooted:

28

26

24

22

20

17

14

10

Previous to todays change armor would also be mined to half its defense value.  So this unit would have a min of 15 absorb against each attack.  But that has been removed so we move ahead with these values.

As is the enemy takes 96 damage (258 - 161).  Prior to todays change the enemy would have taken 90 damage (since the 14 and 10 would have been raised to 15).  But we didn't make the change to modify these situations much (though it does make defense less powerful in these situations too).  We made it so that its possible that the good armor wont do anything (or very little) at all.  So that any attack could to damage.

A few common questions:

Q: Why do we square, roll and square root?

A: To make the random results less linear, we want a more predictable range with some variability.

Q: Why do we force min's on weapon damage?

A: For predictability, we want larger weapons to have predictably more damage. and we want players to be used to seeing the same damage ranges in attacks.  We dont want a unit to hit for 40, then hit for 2 the next round against an unarmored opponent.

Q: What are the problems with this system?

A: Its hard to easily communicate to the player exactly what the damage range is.  Technically its the entire range, but since its not linear the range of 0-40 dos do a good job of describing what a reasonable average or expected damage should be.

If you want predictability, why not average a bunch of rolls instead of skewing the distribution toward high values (and one-hit-kills)? If you average, say, the results of 5 attack rolls (preferably after considering the defence rolls), you will get results that are much closer to a normal distribution, and somewhat predictable (centred around the average).

Of course this already happens when large stacks are fighting each others, but a lot of the game is single units vs single units (champions, monsters, summons, and early game armies).

Using the same calculations, if the attackers had 30 attack instead, they'd have dealt 5 damage (166-161, because of the cap on attack). That's calculated using the same set of rolls than your 1-900, but raising the 10 and 14 to 15 (after square root). I'm... not sure it's enough, really. And if they had 20 attack...

I mean, full light plate is 14 defense, and a broadsword is 15 attack. It's not like equal defense vs. attack was rare or extreme.

Also, here, part of the problem is that you're doing (Sum of attack)-(Sum of armor). This actually makes squads weaker than individuals, because it smoothes lucky attack rolls out.

When 2 independant units roll, one great and one badly, the one that rolled great deals damage, but the one that rolls badly will just do 0. Here, the one that rolled badly actually harms the final result. If you had rolled 12 independant units, you'd have had a better result.

Example: 2 attack rolls, 15 and 10, against two defense rolls of 12. 

Independant: final damage = 3 (15-12, and the other deals 0)

Group: final damage = 1 (15+10 - (12+12)).

While I'm sometimes afraid of the power of groups and want them balanced with individuals, I don't think this work. Really.

"44. When casting a targeted spell on an army (such as imbue champion on a stack with more than one valid target) a dialog is presented to allow the player to pick the unit he wants."

YEEESSSS!!!!!!

I have to agree changing Janusk in some fashion.  Eunuch is a fine option! 

Can we get something to stop the heal spell being cast on a chr with full health.

More than? presumably less than a minimum - and 30 isn't half of anything

Try using a 10 percentile to 90 percentile range and then use the phrase "typically/normally X to Y" and then add a note somewhere under damage ranges - Damage ranges cover most (80%) outcomes but results may occasionally be outside either end of this range (10% high and 10% low).  As someone who gets to explain complex random numbers to a non technical audience this has worked across a wide range of people.

Thanks for the post Derek.

Follow-up question. When we use the gamemodifier DefendableDamage, does the game randomize that damage like it would a normal attack?

Sorry, you are correct.  The min on the attacks should be 20, not 30 since the attack was at 40.

I'm not sure off of the top of my head.  And Im don't have access to the source at the moment.  I would guess that it just allows for the defense steps I described but not the attack pattern.  But thats just by guess.

Thanks for the update!

Two UI suggestions that I would have:

 - could the shard resources be designated with a shard symbol and a number following like the other resources are.  

 - could two small tabs be added near the calendar (upper left corner) to activate or disable the hero & city column of the left side of the screen?  The tabs would be similar to the ones controlling the mini map etc. 

One other Tech/magic thought:

 - Could techs be added so that you could eventually learn the 4 magic schools that you choose from in the beginning.  Be nice to eventually be able to pick up the spell areas that are not critical at the start but might be fun later on in the game.

I am pritty sure it randomizes the damage as well. I've killed Unarmoured Janusk enough times to see that when using DefendableDamage with an ability that sets the damage "Value" (not MinValue/MaxValue), the damage done is in a range of 50% of the value to ~100% of the value. Note, roughly 100% as I don't recall seeing 100% damage, definitely 50% though.

One fix I would still love to see is a setting where I can change the network timeout value used when the game starts up.  It is painful having to wait several minutes for the game to load simply because the network is present but the game can't call home.

Here's a few more bugs I've come across with 1.11 that have to do with Player Ability Bonuses:

A_Administrative used by CivilizedBonus1 doesn't appear to do anything, I've tried using values up to 50...

A_Influence still doesn't seem to affect a settlements influence radius...

A_Sight and A_Agility both appear to be nonfunctional as well.

One UI suggestion of my own:

44) For targetted spells on armies, since you can cast unlimited strategic spells in a turn, allow casting on multiple targets at once?  Maybe like AOW2 have right click invert your choices?  (don't know how exactly you're designing this right now)

Very promising looking changelog right now.

Thank you, thank you, thank you for #44 after this patch i can finally play this game again! it was the only thing holding me back from it.

Wow, E:WOM is going to be really polished after 1.2 is released.

If there is one thing that bugs me right now about the dumb AI in tactical combat it's how it always uses up all it's troops movement points even if it puts them right in front of your troops without being able to attack. Then you have a huge advantage and can use all your points to attack. The AI needs to hold back and stay one or two squares away from melee units, then attack next turn. Thus forcing the player to either move back or to use up action points to close the distance.

This change alone would greatly increase the tactics in tactical battles. Also If the AI has a advantage in ranged/spell dmg they should go on defense and not rush you.

Happy days, #53! Glad to see return and teleport are getting a nerf.

Honestly I'd hardly call getting things to freakin work in the first place 'polishing'. Honestly calling any of the last few patches polishing is really insulting to those that bought the game at release

But I do have a lot of respect for Stardock for working on fixing a game that has a lot of potential while many others would just let this game fade into the mists of time. I understand that due to the massive changes made in December that the game is fundamentally different and a lot of bugs were bound to occur. I really am rooting for team Stardock and just being a grumpy impatient person, honestly I don't have any right to bitch as Stardock is making this game awesome and even handing out free expansions to the most patient of us.

Hopefully I can start recommending the game to others after the 27th. IDK about others but I'm a bit starved for a great TBS to take the place of Civ 4. This patch looks really promising as it addresses a lot of the issues I've been stressing about. I will be happy to grind out some more hours in it post-patch to do my small part to hunt out more bugs.

I think the game will be moderately polished after v1.4, but of course such guestimations are pointless. The primary determining factor of whether the game will be entertaining or not by v1.4 will be the state of the AI; there is no point in playing if the AI (on ridiculous) poses no threat.

So is the public beta still on track for 1-27-11? <3

Can't believe I didn't notice this thread until now.

Looking forward to this, esp. #36 and #44. Those two really bugged me for some reason.

Out of curiosity, is the new version going to include work for the strategic AI?

Also, while Frogboy was testing the AI during Christmas holidays, he mentioned that a bug where an AI faction never built even their first city wasn't an AI bug, but instead a map bug. A farm was situated within a swamp or something similar. Given that this bug was recognized by Frogboy, is it being fixed?

I didn't see any mention of the broken custom faction/sov traits being fixed.

Faction talent Arrogant reduces Champion cost instead of increasing it.

Faction talent Influential does nothing.

Faction talents that give a free tech give it regardless of faction type. (Kingdoms can get empire tech, and vice versa.)

Sovereign talent Stupid does nothing.

Also, don't forget that Spell Immunity is a 64 mana spell with no upkeep!

Does anyone else find themselves never using the Summon spellbook because it is so far up the tech tree and requires its own "tech"?  Seems like it should be included with something else, maybe unlock it when you get Arcane Mastery and the Abbey of Izil / Advanced Spellcasting for Fallen?