I'm trying to dig out how AI overland movement changed and I'll meanwhile list what seems to be different from normal and insecticide here and what seems wrong in both.
Note that this is mostly speculation at the moment, it'll take a few games and many attempted changes to see if I was right on them.

Change 1
When the AI targets a fortress, if it is the player's the original multiplies the value of units there by 1.25. I would think it assumes the player plays better so the rating is higher. Insecticide instead divides it by 4. If the fortress is another AI's, there is no change in either version. This might be why the AI keeps sending 1-2 crappy units to attack my fully guarded fortress, come on, no one sends 2 nagas against 9 pikemen. But if the pikemen's value was cut to 1/4th...see the problem?
This change seems to also apply to attacking other towns.

Flaw 1
There is no check to see if the player has an Alliance or Wizard's Part. The AI still considers them valid targets, just cancels the attack when it would actually be done, probably the same way the player cancels it when "You have an alliance, are you sure you wish to attack" is asked. Which is bad because the unit will stay there and keep trying and failing instead of doing something useful.

Change 2
When there is a war, there is a multiplier of 7 in place, otherwise it is 1. I assume this is the priority of attacking that target. Fortresses multiply this value by 500, other targets by 50, 100 etc. However, In the original, war was a multiplier of 10, not 7.
So the AI will actually consider attacking those they are at war with only 7 times more important than non-war targets instead of 10 times. To be honest I think the higher the better, not the other way. If the AI is already at war with someone, attacking another wizard is not a smart idea.
Additionally, peaceful and lawful wizards should not consider attacking without a declaration of war at all, that's just not right.

Change 3
In the original there was a check that skipped a target if a certain variable related to that wizard was less than 3. I assume this might be a value of how much the wizard wants to attack that other wizard. In Insecticide this check is removed.
There is specific code what to do if this value is higher or lower than 3 so skipping the entire process if it was lower seems a bug that needed fixing for sure.
btw if this value is exactly 2, then non-player towns receive half the priority of player towns. The priority seems to be based on the amount of people and buildings in the target town in all cases, except the fortress which is a fixed 500.

After this, nodes, towers and other dungeons are considered, I didn't analyze this part much as I don't think there is anything wrong with how the AI attacks these, it seems to work well enough.

...to be continued when I find out more.

In the original game, units with higher defense were less likely to get selected as the receiver of the damage when distributing the damage taken by the army between units on the survivor's side.

However, insecticide attempted to change this code and as far as I can tell it's somewhat messed up. If the unit does not have 50 or higher defense (no unit has that...), the chance of receiving the damage is just equal between units. I would think Kryub wanted to make this cost based (units over 50 defense get a chance based on cost) and did it only halfway or something.

What is your experience with this, if any?

What do you think would be reasonable? I've set my mod to do 50-(unit cost/10) but a minimum of 1 as the chance of the unit being the receiver of the damage, so more valuable units will take less damage and will get killed much later in the queue.

I can however justify the opposite just as easily :

A., Weaker units take more damage because they are weaker and less valuable, so the wizard will care about their loss less and will attempt to save the stronger units.
B., Stronger units take more damage because you normally try to keep weak units away from being damaged, and let the strong ones tank for them, their survival is much more likely even if damaged.

In terms of gameplay I think the stronger unit surviving is better, and allows the AI to build up armies more.


Another thing to consider.
Damage is being applied in chunks of 3 points. A unit that is selected to take damage will automatically get selected again the next time, until it dies or damage runs out. This means that after every combat there can be at most one damaged unit, everything else is either dead, or has full health. Not very realistic. I think it would be better if it selected a random unit after every 3 points of damage, with - or without - the already damaged unit receiving a boosted priority for the next selection.

I've applied this change to my mod and the results speak for themselves.
After the change, most units survive from the army of 6 attacking the "many" War Bears, but roughly 75% of the surviving units are wounded to varying degree. (Attacking army had mostly equal value units, cost 100 heroes, a priest and 2 cavalry, all cheap units)
Previously, there were fewer survivors but most of them not wounded at all.

With all the amazing new developments I was wondering whether it was at all possible to fix my biggest problem with the game which is that you cannot see enough of the map at any one time. I spend about a third of my playing time just scrolling about trying to remember where everything is.

Is it possible to increase the number of map tiles that you can see at once? What would need to be done to make it happen?

This is one part of the game which is famously obscure. Even in the official strategy guide, the author declines to reveal any of the machinations behind strategic combat. We know this is how the computers resolve combats between themselves, but I don't think any player has ever used it. The OSG suggests if you can't win a battle the conventional way, try strategic combat to see if it comes out differently. I've never done this, though. Is it random, or deterministic, how does it work?

On 2.5x mana, i keep running into nodes that have 4 great wyrms/great drakes/etc and even with two tricked out champions, i cant win.

Placeholder!

Game Rules and Settings:

Server: pitboss.watto.no:2061

Game tracker: http://www.civstats.com/viewgame.php?gameid=2904

Claim your civ: (the password is sirian)

Game Host unified tech thread (for all Caledorn hosted games):
http://realmsbeyond.net/forums/showthrea...pid=494726

Host: Caledorn
Admin: ???
Map maker: Commodore
Starting date: Today!
Mod: None - BTS
Leaders/civs: Pick 'em

Map Settings:
Cylinder, Monarch, Standard Size

Rules:
- c. Events Off
- d. Huts Off
- e. Speed Normal
- f. Difficulty Monarch
- g. Barbs On
- h. See starts before picking Yes
- i. Version BTS
- j. Scout start Yes
- k. Banned: Nukes, War Elephants (except Khmer), Blockades, Diplo Victory, Great Lighthouse
- l. Not Banned: Corporations, SoZ
- m. Espionage on but no spies (if you get a Great Spy you use him for a Golden Age, Scotland Yard, or settle)
- n. AI diplo (Don't use city trades to signify stuff, no gold countdowns)
- o. No double-moving (don't be a jerk, aggressor gets to decide if they are first or second in turn order as long as it's not a double move, anyone who camps the end of the turn timer so their aggressor can't get the second half to be dogpiled)
- p. No unit trading
- q. No city gifts
- r. No tech trading

Dantski
Darius I Ottoman

2metraninja
Julius Caesar Viking

Gavagai
Louis XIV Dutch

Old Harry
Frederick Egyptian

Alhazard
Isabella Sumerian

R.E.M.
Wang Kon German

Molach
Stalin Mongolian

Greywolf
Zara Yaqob Spanish

Ipecac
Napoleon Native American

TheWannabe
Elizabeth Arabian

Placeholder.

Have I mentioned how Fintourist is the brains in our outfit?

Password is OH

I've noticed that damage seems to be a lot less than it should be, according to the wiki's description anyway.

This is especially noticeable with high damage units like steam canons, fantastic units, etc. They seem to struggle with killing more than 1 of any multi figure unit, even if they are heavily buffed and attacking a low level unit.

I've tried killing engineers with elite mithril steam canons many time, i even gave them the long range ability via the MOM tweaker. But even though they have 15 ranged, a 30% hit chance and am attacking engineers with two defense...they almost never kill more than one figure.

Why is this?

The maths is :

On average 30% hit - 4.5 damage, 0.6 damage blocked, roughly 4 figures should be killed with every hit.

That is obviously not happening in game though...i know that the next model gets to make a new set of defense rolls, but with only two defense, how is it stopping the steam canon from killing a 2nd figure almost all the time? Is there some kind of bug where left over figures get to total up their defense instead of making individual defense rolls?

I'm using the latest stable insecticide patch by the way.

Physical ranged damage has a distance penalty, magical ranged does not. It's also easier to get missile immunity compared to magic immunity, and there is no magical equivalent of warp wood (dumb spell, no resists rolls to make a unit useless is incredibly OP).

A 1 on 1 match with, say, a priest vs a normal unit (halbrediers, cavalry, etc) willl always end up in a lopsided win for the priest and its especially obvious if you are fielding garrison stacks of them. You just kill pretty much everything before they can reach you (excluding the end tier units obviously).

Does anyone have any clue why there is such a massive power gap? Does anyone even build non-unique bowmen? Even with admantinium weapons, the distance penalty is just so god awful.