A miracle?

[Space Otter]

Can somebody please tell me how on earth did I win this? My very first battle in the game, and it turned out... well, rather epic, judging by the casualty counts

[yellow ribbon]

first, you would need to show us the attacking troops (level, time, equipment up to date) to give us proper insight

in generally:

unfortunately not... i reported this far too high entrenchment bonus since July... players tried to change the system, but came to no satisfying end.

you had luck. Also cav., supporting units, HQ etc do, so i believe, not participate in battle but are, looks like in you game either, among the taken POWs.

i once attacked with all disposal units (380.000) of the USA a single city of the CSA, entrenchment was lower, my army badly beaten up, worse than yours...
even in sea battle this kind CAN happen, its not systematical but sometime one side does loose no one, while enemy looses even his last dinghy boat

[Space Otter]

It was a Japanese assault on Port Artur, fresh from the transport ships.
Yes, it seems I got lucky. Judging by the reports Japanese command lost control very early into battle, and at the same time Russian fortress artillery performed remarkably.
Meanwhile, won the russo-japanese war and starting to love this game: so much more plausible compared to, say, Victoria. Absolutely brilliant.

[Jamitar]

seems pretty obvious to me. I often get though the casuality faces are on the other side. ( mine)
you weren defending in an entrenched position probably fort or city and they rushed in like idiots and got mown down by artillery machine guns and hidden men.
total advantage, the somme was 20 times worse

[yellow ribbon]

no! Jamitar, one gets this situation even when you face musket fire, in my case i lost over 200k in my example against a NOT full entrenched foe...

one problem is that there are two bonuses... for my taste a participating troop is either in a fort OR in a trench. not getting a bonus for both at the same time.

it gets even worse with this problems, a besieged troop which tries to break out has an entrenchment-bonus as well as the besieging troops when they were there long enough (pioneer bonus)...
even entrenched garrisons fight entrenched garrisons when ACW starts and both already have a bonus of 200 just after the event fires.

however, the SOMME lasted from July to November, the Germans were outnumbered not more then two to one and the rate of mortality for all involved troops: "only" about every second dead... the losses of the Britons were even barely one-sixth the first day.

...and the screenshot shows ONE DAY

[Jamitar]

im just putting the facts. in this game defender gets huge advantage and thats quite correct. musket fire or not , if only your heads popping out , your at height if you need and you arent tired, the guy running and gunning first wont be able to aim while running correctly, second will be tired and scared, third will be having his whole body.one runs, everyone starts running is also a problem

[Kensai]

A really bad dice roll by the AI. I think you were lucky.

Unless, you have reloaded the battle and always got the same result...

[pesec]

I had same results and in my experience it is all due to fortress artillery. It can annihilate enemy so bad it is not even funny. Check the battle results: your infantry did not land too many hits and would certainly get wiped out, but fortress artillery (NOT regular) landed a ridiculous amount of hits each round.

In some cases, I had assaults where I happened to destroy fortress artillery early on and those assaults had much, much, much smaller losses.

[Kensai]

What's the best way to destroy fortresses? Bringing your own big guns, I guess?! Is there a recipe of winning?

[caranorn]

What's the best way to destroy fortresses? Bringing your own big guns, I guess?! Is there a recipe of winning?

Siege and shelling by ships. That's how I made Port Arthur assaultable every time after the first I played this scenario. The problem is that the ai assaults when it's not at all ready to do so. You first have to cut off the defenders from supply, create a few breaches, hopefully deal some damage via the siege and naval bombardment. Only then do you stand some chance of success...

[Space Otter]

Now that I've played some more it really seems a design oversight. Either Port Artur artillery is to good or AI needs to be taught not to rush onto it.
Here is a screenshot by my friend, he claims he got this kind of outcome twice.
Unfortunately I had no time to try and replicate this myself, but interestingly enough I had no problem taking korean fortresses after a couple of turns of siege.



This is just wrong - almost the number of casualties Japan historically suffered in this whole war.

Oh, and guys. I'm not going to make a separate thread for whining, I understand that the team behid the game is small and there are technical limitations and all, but is there any way to speed up turn calculation? For me, it takes nearly two minutes, with a game hanging dead twice, to process a single turn, on a quad core sandy bridge machine with 8 gb ram. The game seems to use only a fraction of CPU resources. I've honestly tried to play at this pace - build a factory, then wait eight minutes until there is enough money for another one, repeat, and understood that I simply have no patience to get to anything meaningful. Any promise of improvement? Any kind of upcoming patch? Anything?

[yellow ribbon]

ok, i dont speak for the devs, they may disagree...

1.) as i said, this kind of battle event is a NORMAL happening which is just not systematical.
the only major thing changed to your own example is the doubled entrenchment level 200 vs 378 what underlines my point i mentioned above and for 4 months...

but true, with all that siege bonus, AI behaves like an child with too high blood-sugar level.
however, since casualties depend on tactical decisions, they are not historical.

*******************************

2.) there are numberless threads about turn processing time. it clearly depends on the computer, mainly HDD and you shall be happy with 2 minutes.
there is no need to rerun the whole discussion.

i run the 1850 GC late in game with about 5-6 minutes per turn in the 1890s on an old PC, but started with 2-3 minutes in 1850s. others got even up to 10 minutes each turn.

i understood game was never designed to use multiple cores, they mentioned they will try to rework it. due to the time this needs, i guess no one will state if, how and when it is available.

[Space Otter]

Yes, seems to be the problem with AI behaiviour. Should be easy to fix, though - otherwise AI seems to be very smart.

>you shall be happy with 2 minutes

Well, pity then, I simply don't have this kind of time. I looked forward to playing the GC so much.
Anyway, not complaining: the small scenarios are well worth the price I paid IMO

[Jamitar]

hey ! stupid generals exist its even hugely common in those days. the who was born rich and spoiled gets it. it might have been your brother/ father /cousin

which one seems less likely to win?

plus maybe they didnt know the number of troops you had

though the part where every single man dies whereas theyre in a huge assault is problematic.
guessing the survivors just fled

[caranorn]

hey ! stupid generals exist its even hugely common in those days. the who was born rich and spoiled gets it. it might have been your brother/ father /cousin

which one seems less likely to win?

plus maybe they didnt know the number of troops you had

though the part where every single man dies whereas theyre in a huge assault is problematic.
guessing the survivors just fled

The problem is that it makes a scenario like this unplayable from one side (as the Russians). What's the point of continuing to play if the ai commits suicide with it's main army right from the start? I believed I tried this scenario three (as the Russians, several much more enjoyable games as the Japanese) times so far, always with more or less the same outcome, wholesale destruction of the Japanese army within the first five turns of the scenario. Were this just an occasional outcome (and not quite so dramatic), I could put it away as a stupid command decision, but like this :-( ...

[McNaughton]

In each scenario (other than the GC) there are certain AI commands as well as situations. In reality, the Russians have a major advantage in this scenario, that being large numbers of troops, and a centralized position. The Japanese land on the Port Arthur Peninsula, and are in somewhat of a tenuous position. They have too few troops to really cover everything they need. Their quality is very strong, but the Russians can easily afford casualties and wear down the Japanese.

The situation we see above is dealing with problems with the AI unable to realize it put itself in a bad situation against a human player. What the AI would be keen to do would be to secure its lines of supply, keep a covering army at Mukden, and a siege army at Port Arthur.

However, the AI commands are somewhat limited, it is hard to get the AI to do things in a particular situation, at a particular moment.

Some key commands for the AI in war are as following...

AI.SetAggro
How aggressive an AI is in a particular theater (group of regions)

AI.SetLocalInterest
What regions an AI will go after

AI.SetTheaPrev
What theaters the AI will secure in order

The problem with these, is without a series of reactive events, the AI starts a plan, and sticks to it, with reacting only to where their enemy puts troops (it cannot predict where their enemy will go, and to change focus from attack to defend).

In this situation, looking at the scenario files here's what you see for Japan...

Aggression in Manchuria - 400 (average is 100)
Regional interest in Port Arthur - 180 (average is 100)
Regional interest in Mukden - 200

Now, the problem results in, the AI should be aggressive against the Russians in Mukden, but, would be much better off being passive and siege Port Arthur. This unfortunately is impossible, as both Mukden and Port Arthur are in exactly the same area, meaning that aggression is shared.

Lowering the aggression of the Japanese in Manchuria to 200 should probably keep them aggressive enough in the rest of the theater not to be quickly outnumbered by the Russians (Japan would best strike hard, quick, and repetatively to keep the Russians off balance), but keep the siege forces from attacking.

Another problem too can be the leaders. The vast majority of Japanese leaders are 'reckless' and 'hothead', given that they were very aggressive. This can change the set aggression of a stack even if the human or AI switches it to be passive. The Japanese did act aggressively, and did experience much greater casualties than Russia in this war.

[caranorn]

Yes, I can imagine programming the ai can't be easy (one area I've never dabbled in). I assumed this was the case or else these hasty assaults on Port Arthur would long have been dealt with by the PoN team. I guess for now this is essentially a Japan vs. ai and/or two player scenario only...

[McNaughton]

Yes, I can imagine programming the ai can't be easy (one area I've never dabbled in). I assumed this was the case or else these hasty assaults on Port Arthur would long have been dealt with by the PoN team. I guess for now this is essentially a Japan vs. ai and/or two player scenario only...

For now, yes, I and another, are working on some events for the AI dealing with diplomacy and combat, mainly for the GC, but easily should be ported to the scenarios. The RJW is probably a relatively easy fix (at least a fix to stop/slow Japanese attacks on Port Arthur), which I can look at... However, if people here do not mind to be testers a mod could be 'perfected' relatively fast.

[McNaughton]

I am working/testing this scenario, but it is awfully hard to get the AI to be both aggressive in Manchuria, as well as not kill itself on the walls of Port Arthur, or to not totally ignore Port Arthur.

I have got much better results, making playing as Russia much more exciting, but the Japanese still occasionally abandon their siege of Port Arthur (when they should keep going), or for an unknown reason just try and assault it well before they should. Will post my findings next weekend after further tweaking of the AI files.

[Kensai]

Good to know, McNaughton. Thank you!