Higher province number moves first?

[Micah]

Better map move is already a pretty huge advantage.

[chrispedersen]

Originally Posted by Illuminated One

Ok, a test with 3 nations

Arco (nation nbr 0)
Ermor (1)
Lanka (something highish)

All using vastnesses thanks through modding.

4 provinces
120 and 121
90 and 170

First trial (9 times) Arco and Ermor moving armies from 120 to 121 or vice versa.
The armies never miss each other. Sometimes Arco is the defender, sometimes Ermor.

Second trial (about 20 times) Arco and Lanka oving armies from 120 to 121 or vice versa.
The armies practically always miss each other. When they fight sometimes Arco is the defender, sometimes Lanka.

Third trial (7 times) Arco and Ermor moving armies from 90 to 170 or vice versa.
The armies sometimes miss each other. When they fight Arco is always the defender.

What are the province sizes, visually, and are any of them tagged as large or small. Also, what are the sizes of the sizes of the army?

Finally, my personal guess goes along these lines:

All units are given a unique number. When an 'army' moves it is composed of a 'group' of army commanders.

Suppose Army A:

Army commander 1,
Army commander 2,
Army commander 3


Suppose Army B
A Cmd 4
Army cmd 5
Army cmd 6.

I believe that army combats are triggered by any commander activating. Combats can be done in sequential order, or reverse sequential order. When a commander triggers, he triggers all commanders associated with him.

I believe this fits observed behavior pretty well. When numbering commanders, I believe that this is done in nation order. So, in the first turn, Arco will build commander x, ermor will build cmd x+1 etc.

[thejeff]

Ok, assuming it is based on the commander numbering thing and it follows the usual forward/reverse random sequence, an army with a very early commander and a very late commander should always move before an army with only a middle commander.
If this determines where the battle happens it should be testable.


As for commander unit ids, they are handed out in order, but I suspect as usual this might be done in nation or reverse nation order. No evidence for that, just an assumption. It also reuses numbers of dead commanders so it gets more confusing as the game goes on.

[thejeff]

OK, so the test:
I started a tiny game with Arco & Man, starting with 9 provinces so they'd share a border without having to fight indies.
First turn moved the starting troops to the border and built a commander(M) there for Man.
Second turn built a commander(A2) in the border province for Arco.
Third turn both attacked across the border.
Arco led by one of his starting commanders(A1) and A2, Man led by M.
All ordered to retreat.
The battle took place in Man.
I repeated this with the same commanders. The next time it was also in Man. The third in Arco.

This led me to think it's not that simple. Then I ran another try hosting with debug on. The following was interesting:

host: move
  findpath from 6 to 8... distance 7 ms 514 bestland 8 ok 1
from 6(sea0) to 8(sea0) (water-1 boat0)
  findpath from 8 to 6... distance 7 ms 514 bestland 6 ok 1
from 8(sea0) to 6(sea0) (water-1 boat0)
  findpath from 6 to 8... distance 7 ms 514 bestland 8 ok 1
from 6(sea0) to 8(sea0) (water-1 boat0)
      moves validated
  Move interrupted from 6 to 8
  attack move: Bune moving from 8 to 6

Bune is the Man commander. 6 is the Arco province, 8 the Man one. This battle took place in 6.

The next try had the same findpath and move validation, but ended with:

  Move interrupted from 8 to 6
  attack move: Leonteus moving from 6 to 8
  attack move: Kake moving from 6 to 8

Those are the Arco commanders. The battle took place in 8.

This doesn't seem to be based on commander id or the order of the moves.

[chrispedersen]

Originally Posted by thejeff

OK, so the test:
I started a tiny game with Arco & Man, starting with 9 provinces so they'd share a border without having to fight indies.
First turn moved the starting troops to the border and built a commander(M) there for Man.
Second turn built a commander(A2) in the border province for Arco.
Third turn both attacked across the border.
Arco led by one of his starting commanders(A1) and A2, Man led by M.
All ordered to retreat.
The battle took place in Man.
I repeated this with the same commanders. The next time it was also in Man. The third in Arco.

This led me to think it's not that simple. Then I ran another try hosting with debug on. The following was interesting:

host: move
  findpath from 6 to 8... distance 7 ms 514 bestland 8 ok 1
from 6(sea0) to 8(sea0) (water-1 boat0)
  findpath from 8 to 6... distance 7 ms 514 bestland 6 ok 1
from 8(sea0) to 6(sea0) (water-1 boat0)
  findpath from 6 to 8... distance 7 ms 514 bestland 8 ok 1
from 6(sea0) to 8(sea0) (water-1 boat0)
      moves validated
  Move interrupted from 6 to 8
  attack move: Bune moving from 8 to 6

Bune is the Man commander. 6 is the Arco province, 8 the Man one. This battle took place in 6.

The next try had the same findpath and move validation, but ended with:

  Move interrupted from 8 to 6
  attack move: Leonteus moving from 6 to 8
  attack move: Kake moving from 6 to 8

Those are the Arco commanders. The battle took place in 8.

This doesn't seem to be based on commander id or the order of the moves.

Very interesting.

It does look as if moves are validated by commander, although what determines the province the fight occurs in, and the order of the battles is not clear. ( But it isn't cmd order).

Its also interesting that the distance isn't 1, its 7. Does this correspond to the relative sizes,do you think?

[thejeff]

I doubt it. I'm not sure what the 7 means (or the 514 for that matter.)
You'd need to look at a larger more complicated sample.

[Azriel]

I've noticed this as well. However, I have also noticed one other behaviour that may either help clarify things, or help confuse them. When I am moving units from one of my own provinces to another one of my own provinces, it always happens before combat. Every time.
But when I move to a province occupied by an enemy, the sequence seems to be random. I also have tried to chase enemy kamikaze armies across my territories in a futile and frustrating attempt to stop them. Only rarely do I catch up to them, and when I do I have no way of knowing if I caught up to them because they stopped moving or because I moved "first."
My conclusion is that there is some sort of initiative assigned to the respective armies and they move in in that sequence in order to determine in which province battles are fought. And in these cases, moves within friendly provinces seem always to occur first, as if they are treated separately, even though the friendly province you moved into is attacked the same turn. As you all have shown, initiative does not seem to depend on the composition of the army at all.
On other thing, maybe you have realized this by now or not, but you can immobilize an enemy army by attacking it with more than one army. I have found that whenever more than one army attacks the same province, whatever enemy army is in that province always defends. Maybe this is statistically untrue, but empirical evidence is overwhelmingly conclusive. There is one exception here, and that is if at least one commander is ordered to defend, then there is a chance the other commanders in the attacked province may move prior to combat. Oh, by the way, when I say more than one army, I mean more than one group of commanders. When you select multiple commanders and give them all an order, it appears as if the game treats them as one army. I have noticed a difference between grouping them thus and when I group them differently or even send them separately.

This is just one more little detail that makes Dominions III such a great game. The less that players can master the game system, the more the contest becomes a test of skill and strategy. For what it's worth, I hope this is one mystery that is never solved.

[Frozen Lama]

I hate to say "your wrong," but i think you may be a little confused or have just seen some very wonky data. I'm pretty sure that its accepted that you cannot "catch up" to armies with just one army chasing them. those rare times must be when the army is not ordered to move, or when they are ordered to attack your pursuing army. don't know about the multi army thing, but i'm almost positive that chasing armies doesn't work.

[Zeldor]

Yeah, for some unknown reason some people strongly believe that you can chase another army and get a battle. Only way to do it is to move to a province that is being attacked.

[thejeff]

Movement between friendly provinces is handled first. That is known, well observed and even documented in the turn sequence section in the manual.

From my simplified debug example, it seems pretty clear that individual commanders are handled separately, whether they are selected together or not. The multiple army thing is probably a statistical artifact. Chasing armies simply doesn't work, unless he stops or turns back and attacks where you are coming from.

The sequence in the debug log seems clear. All movement orders are processed, then checked to see if any are interrupted by attacks coming the other way. Exactly what determines that is unknown.