Battle Plans Episode IV: Front Interpolation

[FStefanak]

@bitmode I remembered this today and created a bug report for that straight line interpolation based on your findings:

https://forum.paradoxplaza.com/forum/threads/hoi-4-front-is-interpolated-via-a-straight-line.1466347/

I have been thinking about this and I've come to believe that between this interpolation alone can be a strong factor for why the AI is so ineffective. Going via the straight line "distorts" the plan particularly on edges, where the front has to move much more than in the center. Because of this, edge provinces are much more likely to get higher weights, and AI has to work so much more to push those edges according to "plan". The advance in the center is slowed by having to wait until the edges "catch up".
(It probably won't be THIS bad in practice, as the interpolation is recalculated just as the front shifts, right? So they never actually go for the straight line, they just make the first step towards the straight line. But the strong bias towards expanding the edge is still there.)

 

[bitmode]

I have been thinking about this and I've come to believe that between this interpolation alone can be a strong factor for why the AI is so ineffective. Going via the straight line "distorts" the plan particularly on edges, where the front has to move much more than in the center. Because of this, edge provinces are much more likely to get higher weights, and AI has to work so much more to push those edges according to "plan". The advance in the center is slowed by having to wait until the edges "catch up".

I you look at Episode V, I doubt that reshaping the arrow alone will change much from the AI's perspective.It typically draws generic middle-of-the-road attack plans that don't particularly exclude either end of the front. As soon as the whole current enemy front line is contained in the interpolation, the actual attacks are roughly the same. It is not even clear whether the AI designer intended to have it do anything else than zerg rush the entire front.
I see the straight line more as an obstacle if you want to draw some particular shape as a human. It is like the proverbial wrench thrown in the works.

Also it is very likely that the straight line was put in to shore up some long-forgotten edge case and in my experience the QA person checking the bug won't be able to confirm either way whether it is still necessary. Usually you need to be able to demonstrate that something clearly was not designed that way because otherwise it is "working as designed". Though maybe the battle plans will see some larger change at some point:

In general tho its an area we want to improve in the future. A lot of the heavy weights are because of no intermediary filter system on orders that can look at risk/range etc etc

 

[FStefanak]

I you look at Episode V

Nice! I missed that. Great work as always.

You seem to be right, the weights make it so bad that interpolation probably doesn't matter as much.
Still, I am a firm believer in fixing all systems, especially those closer to input than systems downstream. Otherwise one has to pile workarounds and tweak other systems and everything else just becomes that much more expensive to fix over time. Dealing with small tangible bugs seems like a no-brainer.