Killing the Myth: Battle difficulty in campaign

[player1]

Is battle difficulty bugged in campaign games?

I used to think so.
Not anymore.

With testing custom battles I agreed with other testers, that it's bugged, since harder battles are faster not biased toward AI, in same way as easier battles are just slower and not biased toward player.

Well, the problem is that we all jumped the gun and "concluded" that same works for campaign games. Even whole debates about is M or VH battles more challenging arised.

Well, it's not true!

I did a dozen of tests on campaign maps, when I set up several one on one battles, and my conclusion is that battle difficulty works in campaign game fine.

On easy, it's easier for player to beat similar unit, on harder it's easier for AI to beat similar unit.


I hope someone else could do some tests and confirm the issue, so we could put this myth at rest.


P.S.
I modified descr_start.txt files for campaign, so I could set up several one vs one battles. Also I tried when attacking to keep same elevation with defender as possible to prevent any bias.

[Oaty]

Also I tried when attacking to keep same elevation with defender as possible to prevent any bias.

Why not just use grassy flatlands, no chance of a terrain differential there. And if you use 1 on 1 unit the A.I. marches straight for your unit.

Wether the A.I. gets extra bonuses or not is not too much the issue, it's the fact of increased kill rates that allows the human to roll on unit over unit so quickly that makes very hard so easy.

[player1]

Maybe because test is done in campaign, not in cutsom battles, which is obviously buggy.

I say, and have done tests to conculde that it is not buggy while playing campaign.


EDIT:
Haven't noticed AI routing faster at VH, since it was me who was getting routed in one on one battles, always.

[Cheater]

I once heard that it didn't really matter and that it was harder to play it on normal or hard rather than very hard. I honestly don't care much for this. I put them both on VH. It doesn't matter for me since I cheat and my enemies never have a chance anyway! :P

[Puzz3D]

I say, and have done tests to conculde that it is not buggy while playing campaign.

I still think CA should check the final release, and make sure the difficulty settings are working properly in the campaign. It's extremely difficult for the players to check this in the campaign.

[player1]

Probably the reason why nobody bothered to check it campaign and assumed that it automaticly didn't worked, if it doesn't work in custom battles.

[Puzz3D]

Probably the reason why nobody bothered to check it campaign and assumed that it automaticly didn't worked, if it doesn't work in custom battles.

That's a pretty good assumption to make isn't it? CA has clearly stated that they will not make known the combat algorithms. So we don't know the mechanism, and we can't repeat tests under the exact same conditions in the campaign in order to minimize statistical error and arrive at accurate casualty ratios.

I find the concept of giving the combat bonus only to the attack value to be faulty. This concept accelerates combat which alters the tactics. If the bonus was split between both the attack and defend values, combat wouldn't be accelerated as much. Since combat is so fast in RTW, putting all of the bonus on the defend value might even be preferable.

[player1]

I find the concept of giving the combat bonus only to the attack value to be faulty. This concept accelerates combat which alters the tactics. If the bonus was split between both the attack and defend values, combat wouldn't be accelerated as much. Since combat is so fast in RTW, putting all of the bonus on the defend value might even be preferable.

It's not so bad, if you consider that it's bonus to AI only (in campaign, or fixed custom battles in BI).
That makes player units rout and killed faster making lower casualties for AI.

[Puzz3D]

That makes player units rout and killed faster making lower casualties for AI.

Increasing the defend value would also make lower casualties for the AI without making the routing faster. What's so great about fast routing? I don't hear any players saying that the routing isn't fast enough and should be faster.

Of course, what did CA do in the v1.2 patch? They increased the rate of fighting fatigue to compensate for the fast routing instead of slowing down the rate of combat! Why? Because the people making these adjustments don't understant what constitutes good gameplay in this system. I'm supposed to simply keep increaseing the speed with which I move the mouse until I'm ready to punch the screen? I have 20 units to control. Stop increasing the speed of the gameplay!

[player1]

Increasing the defend value would also make lower casualties for the AI without making the routing faster. What's so great about fast routing? I don't hear any players saying that the routing isn't fast enough and should be faster.

Well, they are increasing defend value in expansion.

[Kraxis]

Well, they are increasing defend value in expansion.

True, i seems they have listened a bit. But it is still not all that great. But I must admit that certain units in the Demo did last far longer than anticipated despite losing very badly.

Perhaps this can be credited with the comment in one preview (can't remember which) where the writer said that he felt the battles were 'more right' this time. Of course he had played Rome a good bit so he knew it, but certainly not the details like we do. But if he feels the battles are 'more right' then it isthe right course of action to move towards defence.

[Puzz3D]

True, i seems they have listened a bit. But it is still not all that great. But I must admit that certain units in the Demo did last far longer than anticipated despite losing very badly.

Well it is Barbarian Invasion, so having the stinking barbarians dying like flies the way they do in RTW wouldn't do.

[sik1977]

Whatever you say, a bonus to attack or defense values at H or VH for the AI is nevertheless 'cheating' to compensate (quite heavily) for the lack of AI's ability to conduct decent battles. Whats needed is a thorough improvement in the AI routines in order to give meaningful battles. Simply increasing the AI stats doesn't bring strategy to the battle in question, simply means the AI will make the same suicidal charge etc. and perhaps kill or rout more enemy units because of the blatent cheats.

Having said that, I understand that programming AI must be hard and time consuming. I simply hope that a good portion of the development time and resources have been devoted to AI, with tangible results.

There is no reason (with all things being equal) why AI's Hastati should kill/defeat my Legionaries on VH unless they have them flanked (or else) due to a superior tactic/strategy. Thats when its fun to lose to the AI, not because you know AI cheated, which leads to frustration.

[Puzz3D]

Whatever you say, a bonus to attack or defense values at H or VH for the AI is nevertheless 'cheating' to compensate (quite heavily) for the lack of AI's ability to conduct decent battles.

Even with a better AI this method of making the game harder by increasing the combat stats of the AI units will still be used. However, in MTW the cheat was quite modest. It was a 15% combat advantage on hard and a 30% combat advantage on very hard. RTW is considerably higher than that with a 46% advantage on hard and a 95% advantage on very hard.

Talking about better AI: I've been playing a campaign in WE/MI and the AI has set up ambushes in three different battles where it split it's forces and hid the second force in trees off on a flank. When I advanced the AI held these concealed units until I engaged the AI units in front of me, and then it attacked me with the flanking force. Thinking back, I remember one battle in STW where the AI hid forces in trees on both flanks, and, when my army moved down the middle passing between the trees, the AI attacked me simultaneously on both flanks.

[player1]

I did few more tests now, with defending army on cordinates 88,90 and attacker at 88,88 Both non-barbaric factions with single unit of warband merc.

I was only doing sinlge attack command.

At medium and easy player always won, at hard and very hard AI always won.

Easy win was easier then on medium, while very hard was bigger loss then on hard.

[Horatius]

The slower it goes the easier for the player to controll, so making it fast mode really does make it much harder.

[Red Harvest]

I did few more tests now, with defending army on cordinates 88,90 and attacker at 88,88 Both non-barbaric factions with single unit of warband merc.

I was only doing sinlge attack command.

At medium and easy player always won, at hard and very hard AI always won.

Easy win was easier then on medium, while very hard was bigger loss then on hard.

Did you do this triarii vs. triarii? That was the test that showed us the system was FUBAR. I'm not convinced. From what I could tell the deciding difference was often morale. Speed was clearly different. I'm very skeptical that campaign and custom would use a different combat model. Seems like about a 1 in 10 probability to me that these differ.

In tests I ran I found the custom battles scaled well with what I saw with the campaign, but that was 6 months ago.

Puzz is on the mark about defense. In the combat model RTW did a whole lot of zigging where they should have been zagging.

[Red Harvest]

The slower it goes the easier for the player to controll, so making it fast mode really does make it much harder.

Not once you figure out the snowball technique. Faster does have the distinct impact of making battles unfun.

[player1]

Did you do this triarii vs. triarii? That was the test that showed us the system was FUBAR. I'm not convinced. From what I could tell the deciding difference was often morale. Speed was clearly different. I'm very skeptical that campaign and custom would use a different combat model. Seems like about a 1 in 10 probability to me that these differ.

It's kinda difficult to do triarii vs triarii test in campaign.
You have allies you know.
What about me doing samnite vs samnite test?

In tests I ran I found the custom battles scaled well with what I saw with the campaign, but that was 6 months ago.

Tests I did for know show that there is definetly difference between campaign battles and custom battles.

Puzz is on the mark about defense. In the combat model RTW did a whole lot of zigging where they should have been zagging.

True.
Current model (the working one, not bugged), makes first strike and flanking too important at higher levels, since they are as effective as on medium (due to same defense of AI). The rule of the tumb is to keep combat short, so since otherwise AI would win in test of endurance.
Also archery rules in harder levels, if it's only you how users missles.

[player1]

By the way, if anybody is interested, the best battle conditions in campaign I found, you could get with enemy army at x 88, y 88, and you attacking from south-east (like when adding your unit to Flavius army).

[Puzz3D]

The slower it goes the easier for the player to controll, so making it fast mode really does make it much harder.

This is rediculous. The player is supposed to be able to control his units. Making the gameplay so that you can't control the units defeats the purpose of the game. This is Whack-a-Mole mentality.

[Zawath]

The battle difficulty is messed up in campaign. Very hard in 1.0 is like 50 times harder than very hard in 1.2.

[player1]

You haven't battled romans, didn't you?

They lose -4 attack from fixing primary/secondary bug.

Anyway, if it is messed up, VH is still harder then H, and that is still harder then M.

I've now done tests more then several dozen times.
Just to make sure I'm not deluding myself.

Cutsom battles are completely differnt topic...
They behaive differently then campaign battles.

[TB666]

Well finally some confirmation about something I have felt a long time ago.
I play the Darthmod and I have lost far more battles with VH then with M.
So it is good to see that it is true

[sunsmountain]

It might be true, it might not be.
From what i've experienced, my roman light cavalry was able to charge and kill the enemy in VH, patch 1.2, but not in VH, patch 1.1. I don't know if the attack values are or are not increased.

To me though, it's a side issue that doesn't really matter until the AI is further developed.

Talking about better AI: I've been playing a campaign in WE/MI and the AI has set up ambushes in three different battles where it split it's forces and hid the second force in trees off on a flank.

And how many months did WE/MI receive in development? How many features did the MTW AI have to take into account? Rome doesn't just look better, it's richer in ideas as well. I cannot stress this enough: A good AI takes time. So far, we've seen 1 real patch since release, with nominal improvements.

You can only really start improving the AI in the game AFTER it has been finished. Up until then it has been implemented, but not necessarily smart. Smart takes time to think about it, in rest, without deadlines breathing down your neck. In part, this is impossible nowadays as every industry performs on schedule. But i truly hope future patches will continue improving this wonderful game, that keeps us posted & posting.

[Puzz3D]

And how many months did WE/MI receive in development? How many features did the MTW AI have to take into account? Rome doesn't just look better, it's richer in ideas as well. I cannot stress this enough: A good AI takes time. So far, we've seen 1 real patch since release, with nominal improvements.

That feature was in the very first game, STW v1.0, along with better use of height. The suicide general was addressed in a patch. In MTW, the suicide general was taken care of in the v1.1 patch, and cavalry was given a better flanking AI. An attempt was made to correct the suicide general in RTW v1.2 but it failed. I'm not aware of any other improvements to the battlefield AI in the v1.2 patch. CA has stated that there won't be any improvements to the battlefield AI in the BI add-on.

[player1]

CA has stated that there won't be any improvements to the battlefield AI in the BI add-on.

First time I read something like this.
Care to share a quote?

[TB666]

CA has stated that there won't be any improvements to the battlefield AI in the BI add-on.

Where on earth have you heard that ??
CA has stated in all the interviews they have done that the AI has been improved on all fronts in BI and that BI will bring the AI up to the same level in RTW.

[Puzz3D]

Where on earth have you heard that ??

It's in the FAQ.

"The basic battle game won't change that much; this is an expansion, not a re-imagining of RTW."

[Kraxis]

That does not equate to "We won't fix the sucidal generals, nor will we improve the AI at all." It is just, as usual, a very conservative and secure statement that we can't use for much.
When they want to they do say that they won't improve the AI, so it isn't as if it is set in stone.