Campaign Question(s): AI Unit Size Scaling, and Union Victories

[TortugaPower]

Greetings all. This game is absolutely fantastic. I love a good challenge, and this delivers (and then some!). Coming off a UG Gettysburg playthrough, I definitely had my sights on this next. A few questions about the Union campaign.

[Having read the AI scaling topic on the front page, there are some adjustments made here...]

Short version:
#1) Is there a way to limit the AI scaling? Does the AI also scale its unit quality (e.g. the guns used) to match the player?
#2) Is the Union expected to lose certain battles in the campaign, i.e., are some battle intended to be considerably more difficult?

Long version...
In my first campaign as the Union, I was playing on an earlier (and maybe more difficult) version and I didn't realize your units in the tutorial carry over and intentionally experimenting a lot... So it was no surprise that I lost (the fatal blow at Crossroads). But I felt very ready to start a new campaign and do better.

In this one, I was doing great. All wins except for a [close] draw at 1st Bull Run. But then... Crossroads.  That was the one that killed me before so I was hungry for revenge. Because I had done much better this campaign, I had many more troops who were also better equipped. But instead... I was crushed again. I noticed that the CSA were fielding a larger army this time. I come on to the forums and read here that the AI will scale its forces with the player. Well, that's a little disappointing. I can only tell you that it was absolutely impossible to hold.

#1) I have read that AI scales its size with the player. If so, it somewhat diminishes the value of doing well in the beginning. I also understand there has to be a small amount of recoil to player success/failure in order to keep the game from snowballing too much.  But the CSA army I faced was ungodly large, and the scaling really caught me by surprise.  I also really want to know, does the AI scale the quality of their forces to match the Union as well?

#2) Is the Union supposed to lose certain battles early on? This would make sense historically.  For me, it seems like Crossroads is borderline impossible.

And other factors might be in play here, such as level of Reconnaissance showing you the numbers indicates whether you should accept a battle or not. Maybe you need to use your Recon in order to choose which battles to fight. I'm sure there are still other things I haven't considered. I'll be very interested in the responses.

Cheers,
Tortuga

 

Edit: I'm attaching here a screenshot of my second, most recent defeat at Crossroads.  In the first defeat (not shown), the battle seemed more manageable. That's because a larger size army for each side actually hurts the Union, because there is only so much cover defensively, and the larger your army size, the less of your army that gets cover. That creates a weird system, at least for this battle in particular, well you are punished for doing well on the earlier maps. I still did well (about 1.5 kills-per-death). I doubt Lincoln would have considered this battle a defeat when a Union force, outmatched 2-to-1, inflicted a number of casualties on the CSA. It doesn't feel like it should be a loss

Edited December 2, 2016 by TortugaPower

[Nick Thomadis]

Hello, 

This battle gives you the ability to theoretically have about 20K soldiers (10 units in total) and the average player army measured for this battle is measured about 12-15K. So the real problem seems to be your very small army. If you didn't use all your resources or previous negative outcomes resulted to losses that you could not recover, then this battle can be very tough as you report. Otherwise the AI army would not be much bigger or it could be smaller so you would have many more chances to win.

In our next update the battle outcomes will affect AI opponent in various sectors so Campaign AI can be weakened significantly if you play well.

[Koro]

1 hour ago, Nick Thomadis said:

Hello, 

This battle gives you the ability to theoretically have about 20K soldiers (10 units in total) and the average player army measured for this battle is measured about 12-15K. So the real problem seems to be your very small army. If you didn't use all your resources or previous negative outcomes resulted to losses that you could not recover, then this battle can be very tough as you report. Otherwise the AI army would not be much bigger or it could be smaller so you would have many more chances to win.

In our next update the battle outcomes will affect AI opponent in various sectors so Campaign AI can be weakened significantly if you play well.

Won't this make the battles much easier though? Too easy perhaps 

[TortugaPower]

1 hour ago, Nick Thomadis said:

This battle gives you the ability to theoretically have about 20K soldiers (10 units in total) and the average player army measured for this battle is measured about 12-15K. So the real problem seems to be your very small army. If you didn't use all your resources or previous negative outcomes resulted to losses that you could not recover, then this battle can be very tough as you report.

Hi Nick, thanks for the feedback.

Your quote seems unlikely to me (based on some quick math), but there is a very high chance I am misunderstanding something. In the battle, it only allows one corps to join the fight, and also lists a maximum of 9 brigades. Most people won't have unlocked Army Organization for 2000-sized brigades, but pretending they do, that puts the maximum number at 18k.  If you bring any supporting units, as most people do (artillery, cavalry...), then this max number drops. If you only have 1500 brigade size (as I did), the most you can get with a pure infantry army is 13.5k.

For the average player army, is Golem Labs able to track each person's progress?  That would be very good to help them (I support it if they do), but if not, I'm curious where this average number comes from.

As for using my resources and negative outcomes, I had used all of my resources, and I had won all engagements except a draw at 1st Bull Run. So the problem is somewhere else -- perhaps, on my choice to buy more elite troops rather than building a large untrained army? That's most likely, but I can't say for sure.

I have attached my savegame here "grly8mp0.ja1", perhaps it can be of some use to the developers, or if anyone wants to check my situation. From here, you can see I only have $1300 in the bank, and you can see in Career my victories. (Side note: I hope the save names are unscrambled in the future, haha...)

Cheers,
Tortuga

grly8mp0.ja1

[Koro]

Do you have any unused men? Generally you should try using the men given to you to avoid situations like this.

Oh, btw, my crossroads video: 

A bit different troop numbers though still hard:

 

[TortugaPower]

It could be very interesting if you took my savegame and showed what your strategy could be with that!

[Hitorishizuka]

Can't download your save to check what your campaign is, sorry, getting an error.

For reference, I loaded my old save and this is what I had when I won Crossroads. Assume weapons are Springfield 1842 unless otherwise stated.

Division 1: 2* 1000 Infantry w/ Harper's Ferry, 1* 1500 Infantry, 300 Cavalry w/ Palmetto M1842, 1* 8 Gun Artillery w/ 12PDR Napoleon

Division 2: 1* 1500 Infantry, 1* 1500 Infantry, 1* 8 Gun Artillery w/ 10PDR Ordnance, 1* 140 Skirmishers w/ Sharps

Division 3: 1500 Infantry, 1500 Infantry, 1500 Infantry, 13 Gun Artillery w/ 6 PDR Wiard

I think I only took one of the Infantry brigades from Division 3 to hit the 9 brigade limit. So I brought 7140 men, 300 cavalry, and something like 400 men w/ the Artillery. I'm pretty sure I lost most of the cavalry (maybe the entire brigade) trading against their artillery but otherwise it wasn't -too- bad.

[TortugaPower]

Thanks Hitorishizuka.

I will likely restart my campaign and make it a focus to go with a large army. Your army is much larger than mine, so either a victory at 1st Bull Run is crucial and is making a difference between my army and the victorious armies (like yours and Koro's), or I'm just playing sub-optimally in some other fundamental way. I'll have to sniff around and see which it is.

What were your numbers at the end of the battle? I'd be intrigued to see the scaling, especially for Koro.

Edited December 2, 2016 by TortugaPower

[Hitorishizuka]

Did a real quick replay. Didn't do quite as well as the first time I played from memory, but here you go. Had a little money left so I upped my cavalry size from what I reported right before I did the mission.

 

Edited December 2, 2016 by Hitorishizuka

[Lincolns Mullet]

5 hours ago, Koro said:

Won't this make the battles much easier though? Too easy perhaps 

I think it'll be just fine with the intended game design.  You're right that if a good player beats every battle, their army will become even more lopsided if the enemy army is shrinking and/or becoming less effective.

However!  The goal then would be to ensure each difficulty levels offers an adequate challenge.  The core concept of the campaign is recreating, or closely approximating, the historical odds for each battle.  And with that design concept comes with the stated design that not all battles can or should be won.  That it's OK to take a loss or withdraw if things are not in your favor.  So the second part to this concept is to ensure all levels of difficulty are addressed, and then the dilemma of compounding victories making the game too easy will only be a problem for the best players, or those playing on easy or normal but should be playing on Hard.

In this way, if side battles affect larger battles AND side battles are challenging enough that an average player may lose some, then the benefit of winning a side battle will be weighed against the risk of losing the side battle (or choosing to withdraw once you see you can't win).  If a very good player is steamrolling the AI on normal, then they need to go to Hard.  If most players can steamroll the AI on Hard, then Hard needs to be made more difficult.

That's my take on it, anyway.  The key will really be about making each difficulty level appropriately difficult for different grades of players.  Whether or not the side battles diminish the enemy army isn't the issue, in my opinion, it's whether the difficulty levels are scaled appropriately.  How that is done can be worked through design, AI, or straight up bonuses for the enemy like morale and damage increases.  I think the ideal situation is constricting the amount of men/gold you get on Hard and Normal so that you have to husband your troops effectively.  I'm not a big fan of adding artificial morale and damage bonuses to the AI, which is why I typically play on Normal and still get a challenge.  On Easy, resources should be very forgiving so you can send your men to the meat grinder fairly regularly and still get enough resources to replenish losses.

[TortugaPower]

Nicely played, Hito, but also very confusing. You brought more Union troops in but faced less CSA than I did.  Is there an RNG element to the AI army size? Otherwise I can't understand this scaling.  I finally managed a victory, same setup (my third attempt). I don't like to save scum, and I won't do that if I'm recording for my channel, but I just wanted to see if it was possible to win. It was, but it was *damn* difficult.

Tortuga's totals: 8076 Union vs. 15682 CSA (a  1.94 to 1 CSA advantage)
Hito's forces, total: 9485 Union vs. 13914 CSA (a  1.47 to 1 CSA advantage)
Koro's totals: 12480 Union vs. ~15473 CSA (a ~1.24 to 1 CSA advantage, going off the recon in his video)

So I'm still quite puzzled.  Perhaps a bug, where the AI size scaling algorithm hits a weird regime if player numbers are too low? I brought in the least forces and yet faced the most.

Thank goodness you can view your results afterward in the "career" screen -- a very nice feature.  The screenshot is below.  So the AI scaling remains a frustration and really an unknown, unless I am just really unlucky and the AI hates me in a special way.  I wish the savegame worked, it was quite an intense fight.

Edited December 3, 2016 by TortugaPower

[Hitorishizuka]

Best guess is that I brought more artillery and seem to have ticked over a tier that appears to have forced them to use more budget on their own artillery and less on infantry. Ditto with cavalry. Also, one of my brigades was skirmishers which may have helped pull down the overall numbers.

Edited December 3, 2016 by Hitorishizuka

[Lincolns Mullet]

I'm not entirely sure how it works, but there's supposed to be a "trap" mechanism that can trigger or not trigger on any given battle so that your original recon #'s are lower than the actual result.  You go into the battle confident there's only 10k troops, and then there ends up being 14k. Tortuga, you don't happen to know what your recon told you about the enemy army size prior to starting the battle did you?  Do you remember if it matched to this final result?

[TortugaPower]

I believe I had 0 or 1 points in recon, so I had no intel.  I was also wondering about the Recon stat.  If they put a mission in with (to exaggerate) say 100k Rebel troops, and you were supposed to know not to fight that by using your Recon, I'd be okay with that. But that doesn't appear to be the case here.

[Hitorishizuka]

I generally have 0 points in Recon. As it stands, so long as you match Brigades in required slots you can win every map. Don't need to know how many enemies are on the map so long as you play smart.

[Pema]

Playing for the Confederacy,not in one battle,was not the benefits of the Union,forces, or equal to,or my advantage.Sorry if questionable pumping some fans of the veterans,the game is much easier.Regards,and sorry for my Google English.