loki100 wrote:yep, committing the German 6 Army to drive deeper into Russia has usually led to a major victory
ROFL! That one made my day!

Kensai wrote:LOL!
Dunno if it is a bug, but I am almost certain the "outnumbered" surrender is present in PON regardless if an army has supply carts or an unblockaded port. Have seen it happening more than once with outrageous differences of troop numbers. Am I confusing it with an experimental version of the engine? Probably, I had started a game in the autumn of 2011, many versions have passed since then.
Anyway, be careful HerrDan!
Kensai wrote:Do you have an open port and valid supply chain to the isolated area? Those troops may run into trouble if they stay long enough without a steady supply...
loki100 wrote:what is worrying is that with the return of good weather the Russians are starting to win battles and to inflict heavy losses. I guess that this has increased frontage deployable (probably bad news for you) and partially negated your advantage in close quarter combat (better morale. probably better cohesion) so they are doing well at long range.
this promises to be grim, and thus great fun to read about ... sorry
interesting to see the US-Spanish war start, wonder how well the US AI will cope with a complex naval war though
Kensai wrote:It is always fun to see the human player challenged. And Germany going against Russia is a classic!![]()
loki100 wrote:well done at Koenigsberg, that is the advantage of luring the AI onto a large fort ... it will attack with full commitment and that means you get the sort of battle sequence where heavy losses feed into lots of lost elements, thus pruning a huge stack down to a manageable size.
wee grumble though, I've a relatively slow internet connection and if all your images are .png loading is slow, can you convert to .jpg, makes little practical difference unless you are doing really clever graphical tricks
loki100 wrote:well done at Koenigsberg, that is the advantage of luring the AI onto a large fort ... it will attack with full commitment and that means you get the sort of battle sequence where heavy losses feed into lots of lost elements, thus pruning a huge stack down to a manageable size.
Kensai wrote:
Yep. But this is an extremely disturbing behavior by the AI that I hope catches Pocus' attention. I really don't understand why Athena does not hold on the siege if it has problems with cohesion or worn-out elements. This battle was a turning point indeed, but also a good opportunity to catch and fix certain bizarre behaviors.
The AI blew its chances here and dropped the ball. I am kind of disappointed.
It is so obvious what went wrong here. If only it waited one or two turns to get reinforcements in its units and regain cohesion. With so much artillery, it would have annihilated the defenders, even if they didn't surrender on their own due to the huge numbers difference.
Honestly, in other AGEOD titles I don't see such careless behavior, perhaps there are inherent limitations in PON (or it simply needs some extra rules). In SCW I enjoyed so much the small campaign of the North which I lost (as the Republicans) for few VPs while managing to keep on my objective regions, but the AI behaved remarkably well at mid-range aggressiveness settings, scoring victories after victories on the battlefield.
Jim-NC wrote:It also appears that the Russian army may be running out of replacements. At some point, the AI can't keep up with the replacements necessary for such a venture (attacking a fort then retreating when it goes wrong).
Kensai wrote:Sorry if I sound like I want to diminish your victory. That wasn't my point. Studying other guys' AARs is an opportunity for me to spot limitations in the engine and contribute to improvements. That's my other job.
The AI can commit errors in "reading" your intentions. It can move to a wrong direction, for example. But it should NOT commit errors when it has the initiative. Choosing between assaulting or not should be a very careful business for the AI. I have studied too many battle logs these past three years (especially in PON) to know when a battle is doomed. These catastrophic results (400K dead, injured, and prisoners is probably one of the highest values I have ever encountered!!) are evident of too low cohesion and element hitpoints. If the Russian AI has run out of reinforcements, then it should somehow not attempt to assault and stay on besieging.
I wonder if this can be hardcoded to the AI's behavior somehow. When cohesion and/or hitpoints of the stack are low --> the AI should NOT be allowed to assault and simply maintain the siege. If this is too deterministic, perhaps there should be a percentage of attempting the assault, obviously extremely low (not more than 10%, just to abstract an impatient General).
loki100 wrote:I think the issue with the assault is to set it up you need to fully commit, so if the first attack goes wrong, the next will hurt you deeply, till you run out of organisation to continue.
I think this is a PoN issue for 2 reasons. First PoN has large forces (incl the AI monster stacks) and relatively few provinces - compare a PoN Austro-German war to Rise of Prussia. Second, PoN uses the simple command model, so you don't get the full army organisation of say Revolution Under Siege. This links to the mega stack issue as the AI can't build those in the games with the complex command structure, its not an issue in Wars in America (all the armies are relatively small) or in AJE (attrition and supply prevent the use of single monster stacks, and, again, in many scenarios the armies are relatively small).
I agree that if we could cure the mega stack then wars would be more interesting, at the moment there are a number of tricks to cull that sort of force, most are unrealistic but the player has no other choice but to use them. I'd rather a clash of multiple armies, with the risk of not committing/strange stack targetting etc, the result would be more dynamic and realistic wars.
At the end of the day, the AI is an AI, that in PoN is pretty good, and there are plenty of constraints on a player (if you use restricted SoIs). We want the game to have some dynamism and for the player to do better (so as to retain interest), so it doesn't worry me too much.
As to .png, one thing you can do is to take the image, open it in an editor (paint.net is great, and free) and save as .jpg. You'll see the resulting file is about 25-40% the size of the .png which makes a difference when loading a lot of images.
Users browsing this forum: No registered users and 2 guests