Slitherine

I don't believe I've ever seen the AI take the bonus movement after destroying or repulsing one of my armies. Is this by design, and if so, why?

triari wrote:I don't believe I've ever seen the AI take the bonus movement after destroying or repulsing one of my armies. Is this by design, and if so, why?

It does advance if it can conquer something, City, Capital, Mine, ...
Yes, it is a design decision. An advance could in fact block the other units spot for follow up and it can also put a unit on front sticking out and being easier to counter or surround.

There are cases it doesn't properly consider.

1) rivers - I have seen the Germans unable to cross the river into Russia by Odessa solely because they never advanced after combat.

2) Armoured / Mot units that end not adjacent to any enemy - these will have to pay Oil to move and again to attack.

3) Atacking without moving gets a bonus.

Never advancing is as bad as allways advancing.

I don't agree. It is better for the AI to not make a clever move than to make a stupid move.

iainmcneil wrote:I don't agree. It is better for the AI to not make a clever move than to make a stupid move.

The problem is when not making the clever move IS the stupid move.

I have seen the Germans held up on a river line for MONTHS because they would NEVER advance into the hex(es) they were clearing of Russians.

Even a simple 50% chance of advancing across a river, -10% per step below 10, would be better.

sagji wrote: The problem is when not making the clever move IS the stupid move.

I have seen the Germans held up on a river line for MONTHS because they would NEVER advance into the hex(es) they were clearing of Russians.

Even a simple 50% chance of advancing across a river, -10% per step below 10, would be better.

I have to agree w/ sagji here. There are a few instances were the AI would have given me a real fight if it had taken the bonus movement. Such as when I was closing in on Moscow, I moved adjacent to Moscow. On the AI's turn, it destroyed my army by Moscow, but did not take the bonus movement to buffer the capital. Therefore, on my next turn, I was able to move into the hex and attack Moscow. Because of other forces I had in the area, I was able to prevent the destruction of this army.

A value could be added that considers rivers, capitals, and terrain where the AI would benefit from the movement.

The other thing they don't even do is advance to free encircled units, which could be re-supplied if they took the combat advance. I think adding a "should I advance or not?" AI subroutine would be a valuable addition to the AI.

I'm with Iain here.

Just as the first rule of a healer is "First, do no harm."

So should the first rule of an AI be, "First, do nothing stupid." Errors of commission are always more egregious than errors of omission.

Programming an AI to consider all the factors that go into the "Do I advance or not?" decision is not an easy task. Can you honestly say you have never hesitated before making that decision yourself? I certainly stop and think about it quite often. It's a complex decision, guys, and there often is no "right" or "wrong" answer, just a choice of how to play it.

Often when I refrain from advancing, it's so that I can move another unit to that just-vacated hex in order to attack.

I'm not denying that it's possible for the AI to do better. I'm saying that our clamoring for an improved AI needs to be tempered with large measures of patience, and understanding of just how difficult a task this is.