[IMPORTANT] Can Union AI hold Cemetery Hill on first day?

[David Fair]

The game design is limited because it is focused on tablets.

It is also limited because the development crew is ready to release the game.

I'm not optimistic that any changes will be made to artillery.

 

 

But let’s go through your list of fixes:

 

The number one issue most people have had with artillery is identifying targets/understanding where to position artillery.  

You might consider adding this to your list.  

An icon that identifies potential targets would allow new players to understand where to put artillery without the current learning curve.

I’d suggest making this your first point.

 

1) Deploy by brigade:

Adjusting the artillery to historic performance sends many gamers (and the design team) into hysteria.  

It might be possible if the artillery were in a brigade to mask the fact that ACW artillery inflicted about 6% to 8% of battlefield casualties.

Guns are a powerful symbol on battlefields.

My belief is that high explosives have simply tainted the reality of black powder artillery.

Players want to see massive casualties from artillery regardless of the anachronism.

 

2) Limber/Unlimber:

The design team is limited on the number of different types of units they can have.  

They have pushed back on every suggestion for depicting "limbered" artillery.  

 

My thought is that they don't need a different representation - they can just make the current sprites move faster – and abstract limbered artillery.  

It wouldn't look great - but neither does the current system with guys pushing the guns slowly over the map.

 

3) Higher artillery vulnerability to infantry fire:

Harnis, batteries that had infantry support were difficult to target - which is why the artillery had such low casualty rates. 

It is also why the artillery of both armies was virtually intact on July 4^th at Gettysburg.

Batteries that suffered high casualty rates were those ordered to “hold at all hazards” without infantry support.  These batteries were not charged and destroyed by melee.  Rather, the CSA infantry sent forward skirmishers that were too small a target to warrant discharging the guns.  The skirmishers gunned down the horses to immobilize the battery.  Bigelow’s battery at Gettysburg is the poster child for ACW infantry vs. unsupported battery tactics in the ACW.

 

Charging batteries with formed infantry was an expensive proposition - charging batteries that were supported properly with infantry was usually a disaster for the attacking force.  

 

From a design standpoint I’d suggest that you want the artillery to be the throttle on the battle (this is the role artillery played historically during the ACW).   “Winning decisively” was difficult during the ACW.  Defensively the combination of infantry properly supported with artillery, on superior ground was decisive (see Marye’s Heights, Malvern Hill, Gettysburg, etc…)

The trump card for this defensive troika of infantry, guns, & ground was breastworks.

I’m not a fan of higher artillery vulnerability because it is not supported by ACW fact.

 

The UGG team is pondering how to get the Union to hold Cemetery Hill on Day 1 - the real answer is get the mechanics of the units correctly aligned with historical reality and the troika of gun, infantry, & ground should be sufficient to hold the line.  Breastworks were the trump card in defensive tactics during the ACW.  Culp's Hill defensive works precluded a successful assault by the CSA.  It is true that some of these works were accidently abandoned and occupied by the CSA.  The Union had a very difficult time trying to drive out these intruders.

 

4) Aligning artillery impact with ACW reality.

General Gibbon stated the effect of artillery was, “more moral than physical”.  If you consider Alexander, Hunt, and the dozens of other artillerists during the period it is clear that artillery stiffened the defensive resolve of the infantry that was supported with artillery.

In UGG I’d suggest that infantry closely supported gets a morale boost.

 

Also, it was fearful for infantry to charge / attack infantry that was properly supported with artillery (Gettysburg, Fredericksburg, Malvern Hill…) and had a negative impact on morale of the attacking forces.

Why did Union XI collapse?  First they were flanked, second they were missing one of the key elements of our combined arms troika – specifically ground.  Third, XI Corps was covering too much ground with too few resources.

 

Note that the balance of infantry and artillery was key.  At the Peach Orchard the Union tried to bolster III Corps with guns from the artillery reserve to cover the extended front.  This left a line of 32 guns on the left flank of the position.  These guns were the target for the CSA attack which smashed the flank of the Union Emmitsburg Road position.  They were not “charged” – skirmishers drove these guns out of their positions and they were forced to retreat.  General Humphreys tried desperately to find some infantry to put into the line on this flank.  His goal was to remove some of the guns because they were out of proportion to the infantry.  General Humphreys was correct and the CSA proved it by smashing this line of guns and rolling up the flank of III Corps.

[long5hot]

Nick, I played versus Cunning and again took the Cemetery and the other objectives, this time inflicting over 16000 casualties on the Union and obliterating 8 brigades after I had them surrounded north of the Cemetery.

http://i.imgur.com/ggXjDMB.jpg

Unfortunately none of the screenshots I took during the battle were saved. The Union really didn't do much wrong, its the initial setup that places them so firmly on the back foot. The only way the Union can counter a strong attack on its left flank is to pull right back to the Cemetery, straighten its line and thus free up reserves that it can send south. It does not do this, and I'd argue that it probably shouldn't as such a maneuver would be worthy of Napoleon.

The Union starts out defending a bulge, with only one brigade holding Culps and a hanging left flank that's so easy to roll up, and only one brigade (Bucktail) in reserve. That's a recipe for disaster.

The battle in brief - I took the town and pressured the northern brigades. I advanced Avery to the foot of Benners, upon which the Iron Brigade advanced a little and began to trade volleys, then marched the Tigers and Smith around behind onto Culps and attacking the Iron Brigade from both sides. This along with the artillery from Benners forced it to rout and I had Culps. Bucktail was committed to help and this time did not charge at my sharpshooters but engaged Smith, but it wasn't enough (2 brigades with 1 routed versus 4).

In the south I rolled up the flank with ease as per usual, routing brigade after brigade. I sent 3 fresh brigades deep south to take the Ridge & hold it against the reinforcements, and to chase down brigades trying to rally. The rest surrounded the town and obliterated every brigade within the noose - none escaped alive. I've practiced this 7 times now and I've become very efficient at it.

A skirmisher unit escaped through my lines un-noticed and took Seminary Ridge near the end of the battle, but I had a couple of recovering brigades near enough to run them off before the battle ended.

All in all I really didn't notice any difference between this version and the previous one. Also the victory was achieved purely through infantry movements - artillery were in a supporting role. I didn't use panzer formations (although I certainly could have).

I'll run the "All-out attack at Seminary Ridge" scenario again, as that revealed some poor AI behaviour in the last version including lousy line formations and terrible use of the abundant reserves the Union has at its disposal. If this patch does indeed improve the AI, this will be the battle that shows it.

[Nick Thomadis]

Longshot,

 

Are you sure you tested on 0.95? We had to pull it out and revert to 0.94 due to a dll bug that made for some users impossible to play the game.

 

We are going to re-update to 0.95 very soon.

----------------

 

David,

 

Thanks for your ongoing feedback. About artillery, we have to always remember that in this specific game and any game, there are limitations. So the design approach uses these limitations to reach a specific gameplay goal. Surely the artillery now is not so powerful as it was in old beta versions for which you rightfully complained. Artillery still is very helpful to fortify positions and can inflict enormous casualties by canister/shell if supported properly. If there is long range artillery fight, then the casualties are as low as you describe. Statistically these nbrs you use include all-day barrages and not 1-2 hours battle engagements. If battles where enduring in that intensity and proximity all day, then artillery casualties would rise a lot more.

[long5hot]

Longshot,

 

Are you sure you tested on 0.95? We had to pull it out and revert to 0.94 due to a dll bug that made for some users impossible to play the game.

 

We are going to re-update to 0.95 very soon.

You're right, its 0.94. Interestingly though I replayed the All-out attack at Seminary Ridge (July 2 Afternoon) vs Cunning. In my previous play-through the AI did not move any reserves and basically fought like General Jack Daniels was in charge. In this one, the infantry reserves were committed (the artillery were not, they stayed motionless) and it generally put up a better fight, though it left the south flank exposed which was its undoing. Might be worth checking out in case there's something you haven't noticed/fixed yet:-

http://imgur.com/a/rDsZk#0

[Nick Thomadis]

Eager to read your test after 0.95. Melee balance will play big role in the outcome as in 0.94 is overpowered and favours the CS. AI in 0.95 should also deploy better his troops. But all will be reviewed again. I will not forget the matter.

[Harnis]

I understand there are limitations in a game, but what limitations are currently preventing a more realistic approach to artillery? And how would making artillery perform and behave realistically be somehow contrary to the gameplay goal?

[Nick Thomadis]

I understand there are limitations in a game, but what limitations are currently preventing a more realistic approach to artillery? And how would making artillery perform and behave realistically be somehow contrary to the gameplay goal?

Limber: If we cannot have limber animation and switch states between limber-unlimber then we cannot just increase the artillery speed and assume it runs with horses. Seeing man feet to run as benny hill would look not only hilarious but would seriously unbalance game watching artillery moving fast everywhere without having pauses.

 

Captured guns: the best would be to have a mechanic to capture guns and so casualties of artillery and loss of guns would be very realistic. If we cannot have that then we cannot just simulate immune artillery, because players want to see a result after an action of charge to an artillery for example.

 

Artillery lethality: Already replied above to David.

 

Not sure what else can be the questions for artillery. We are surely going to evolve artillery mechanics in the next game and probably into this with a patch but not before the full release of UG:G.

[David Fair]

Nick,

 

Honestly I'm not clear on your comment.  But please - let's not make up illusory caveats to disregard the data regarding artillery.

 

My point was that the moral effects of artillery were greater than the physical - this is not correctly represented in the game.  

Particularly at close range canister had a tremendous impact on the morale of attacking troops.

Guns, men, and ground were a powerful defensive combination (physical and morale) that sapped the morale of attacking units.

The game is too easy because this mix is not properly implemented in favor of the defense.

 

The net result is every game is a Battle of Cannae - with the utter destruction of one army. 

 

As a result of not having the game mechanics correctly balanced the the AI (and some players) are having difficulty holding Cemetery Ridge on Day 1.

I don't think this this is an AI problem - it is a game mechanics problem.

 

 

I don't know if the information below will be helpful - but it might be a response to your post above?  

 

The casualties inflicted by rifled musket represented over 90% of the total casualties during the ACW.

The casualties inflicted by artillery - regardless of the ammunition type, range, or duration of period firing - were between 6% and 8% at Gettysburg.

 

Here is the data from Gettysburg on the ammunition, duration, and location of the major artillery actions.

 

The CSA carried 44,000 rounds of artillery ammunition to Gettysburg.

 

Brigade-level artillery action at Gettysburg was concentrated in just a few actions that lasted:

 

Benner's Hill - 45 minutes.

Peach Orchard - 2 hours.

Longstreet's Charge (Pickett) - 90 minutes to 2 hours (depending on the source).

 

The CSA fired an average of 80 rounds per gun most of this ammunition was devoted to one of these three major artillery actions.  

 

In combination the CSA fired about 22,000 rounds - about half of the ammunition they brought north for the campaign.

 

The Union fired an average of just over 90 rounds per gun most of this ammunition was devoted to one of these three major artillery actions.

This was 32,781 rounds - roughly the amount of ammunition carried by the artillery reserve.  

The Union had more than 60,000 rounds at the remainder of the battle.

 

Note that in each of the actions the artillery spent at or near 40 rounds - this is a few rounds short of the total capacity of the artillery ammunition supply that a battery and caisson combination carried into battle.

 

The artillery at Gettysburg was not firing continuously.  The guns were silent enough on the morning of July 3 that General Meade personally scolded an individual battery for "wasting ammunition".

 

I suspect that you are trying to make an argument that the statistics don't include canister?

If this is the case then I can only assure you that the AoP surgeons who collected the data and the projectile fragments understood the difference between a minie ball injury and other forms of injuries.

[David Fair]

PS - on the issue of the gun looking silly if it were moving faster.

 

If you changed the icon from men pushing their guns around to a gun and limber combination with horses it would look much better, move more accurately, and have a minimal impact.

[long5hot]

Ok, 0.95 Cemetery test versus Cunning:-

http://imgur.com/a/Lkapo#0

Things the AI does better than 0.94:

• Moving troops to respond to threats.  It covered Culps after the Iron Brigade was drawn away.

Things it does worse than 0.94:

• Cunning is more aggressive than ever, committing huge forces to attacking the town and immediately marching the Iron Brigade off the hill (which is why it had to move another brigade to defend it).
• Reserves are committed immediately.  Bucktail was seen in the front-lines of the northern forces contesting the town, and Ames straightaway marched to help the left flank.  Nothing held in reserve, in other words.

Things it still does badly:

• The left (southern) flank received no additional help (it couldn't as all the troops were in the town) so it was still rolled up with ease.
• Culps was still taken, for the same reason, and because the Iron Brigade abandoned its post.
• The AI doesn't move artillery or HQs at all.
• Four brigades spent literally half the battle chasing my sharpshooters away.  Culps could have fallen if they'd counter-attacked it instead, but no the sharps were more important.
• Stannard still stuck on his hill, no real effort to take the Ridge even though it wasn't strongly held.
• When the line needed to be shortened by pulling back from the town, the AI ... didn't.  It still looked like General Jack Daniels was in charge of its battle-strategy planning.
• It doesn't like enemy artillery in the front-line and will charge it whenever possible regardless of the infantry support near said artillery.

Other notes:

• A few artillery batteries in the Cemetery were incredibly difficult to move.  They were unsupported by infantry, yet took little damage from ranged attacks and routed brigades that tried to charge them.

In short, the AI behaves differently but has become overly aggressive.  I felt like I was playing against Determined.  It still lacks common sense and makes terrible head-scratching decisions that lead to disaster, and there's work to do with artillery & HQ usage.

 

Edit: Just starting a game against Determined, and its throwing everything it has at me in a full-on attack from the outset.

[David Fair]

Regarding Benny Hill artillery.  I played around with a couple of tests and the current icon doesn't look that bad when the artillery moves at roughly the same rate as the infantry.  If you wanted to represent "Limber to the Rear - Gallop" then the artillery would move at roughly the same rate as the skirmishers on "Run".

 

If the infantry & skirmishers don't look like Benny Hill to you why would the artillery?

[Harnis]

Yeah they could either just make the artillery dudes run at the same speed as skirmishers, or swap out the sprites for a little horse and limber. It doesn't really seem like a huge limitation....

 

And as for casualties in artillery, fixing their mobility to historical levels would fix this. There would be no need to deal with captured guns and stuff. In the whole battle of Gettysburg I don't believe even a single gun was captured and turned against the enemy. So it's not a big deal. Capturing guns is simulated by if you catch arty in melee. Most of the time the guns would be disabled before being captured.

[fierman]

I noticed friendly artillery does sometimes refuse to change face when ordered to do so. They try, and then they change back to the original facing in a few seconds. Especially happens when two artillery units are close to eachother.

[Nick Thomadis]

Regarding AI over aggressiveness it is a known issue now and will be addressed for the release or with a hotfix asap.

 

Regarding Benny Hill artillery   I am going to revert to this matter to you a little later.

[David Fair]

Harnis,

 

Smith's Battery lost four of its six 10 pound Parrotts to capture.  One of these guns one was spiked and later recovered by the Union - the other three guns were used by the ANV at Gettysburg as replacements for three 10 pound Parrots that had been knocked out of combat during the artillery barrage before the Peach Orchard attack on Union III Corps.

[RichardMcCor]

i played a bunch of SP on scenario 1 against CSA with different AI types. cunning, determined, risky etc. All with me as Union with no clicks at all.

 

four big notes

 

1. overall they seemed to do better with co-ordination of attack, i.e. more than one brigade attacking at once. Got union off oak ridge a couple of times. Still did too many single brigade charges against set opponents though. Still charged from very  long distance on their own with predictable results . Petigrew into paul & iron brigade on the ridge for example. Yuk! Also after charge and retreat, they then tried again in the same place with the same result.

 

2. leaders did not help recover their brigades well. leader kept helping little archer while not helping the large brigades of davis and petigrew like a human player would

 

3. CSA arty much better. on 0.94 arty would not get into position and just watch. Now in 0.95 the CSA seemed to get into much better position and supported the attacks better

 

4. cover helps more than before. Now that cover seems more important to help brigades the AI needs to use it more. The union on Oak Ridge  can hold out for a long time in cover. No point charing paul when he is at 95% cover on Oak Ridge. Same for putting a brigade in the trees. Does the AI know that cover is good to use? It should use it more.

[David Fair]

Men, guns, and ground - the holy trinity of the ACW.

[long5hot]

Regarding AI over aggressiveness it is a known issue now and will be addressed for the release or with a hotfix asap.

 

Ok, so ... no more testing then?

 

How about the other shortcomings of the AI?  Its scant regard for objectives or for cover?  Its chasing unimportant units, especially skirmishers and sharpshooters, while elsewhere the battle is being lost for lack of troops?  Its poor use of artillery and HQs?  Its habit of charging any artillery that sets up too close, regardless of context?  And the way it tries to charge over & over again with brigades that are demoralized, attacks that are doomed to failure & only ensure that the brigades rout & never recover?

 

To me it seems there's a lot that needs fixing, but I do understand you need to get the game finalized.  It'd just be a pity to see its primary selling point, its AI, going through to release with these issues.

[OLee O'Hara]

Release or not.. I'll post my testings :

ver. 0.95

 

I could see many improvements and in many situations the AI is doing very well... but there are still some issues:

Could Union AI hold Cemetary Hill on first day?

 

On dynamic:

Culps hill was totally undefended:

https://www.dropbox.com/s/r9tsgbl4n4dukks/2nd%20battle%2Cculps%20hill%20again%2C%20on%20dynamic.jpg?dl=0

the counter attack.. done by only one brigade without securing the flank. The Union brigade just ignored the presence of Gordon and tried to attack Doles. This attack didn't last long:

https://www.dropbox.com/s/xohgao97voi1ew5/2nd%20battle%2Cculps%20hill%20again%20and%20then%20the%20counter%20attack%2C%20on%20dynamic.jpg?dl=0

and this happened at Lutheran Seminary at the same time, the AI attacked but finally failed, the AI should know that the 1000 vicory points of Seminary ridge aren't as important as Cemetery Hill and Culps Hill, At some point the Union troops will have to withdraw from Seminary Ridge anyway.

https://www.dropbox.com/s/s0hck7cf1endwoa/2nd%20battle%2Cculps%20hill%2C%20and%20this%20happened%20west%2C%20on%20dynamic.jpg?dl=0

 

On determined:

Culps hill was defended by only one brigade:

https://www.dropbox.com/s/gvax2eby7dl6be0/culps%20hill%20on%20determined.jpg?dl=0

I'd like to believe the succes was based on my awesome general skill but...  I fear it's different, the AI just did a few mistakes:

https://www.dropbox.com/s/1auammtx6fgxq2m/culps%20hill%20on%20determined%20end.jpg?dl=0

 

-> I always used Rodes troops to conquer Culps Hill and thus Cemetary Hill got lost for the Union as well, and the next battle was always "Union disaster at Cemetary Hill"

 

 

Typical mistakes of the AI that I watched in my last games (on dynamic):

The AI orders Biddle to attack the flank of Achers brigade, but opens the flank to Davis at the same time... next moment he got shot to the flank and the men routed:

https://www.dropbox.com/s/wgcsxycpyysxk7t/1st%20battle%2Cshowing%20the%20flank%2C%20on%20dynamic.jpg?dl=0

 

The AI orders the Iron Brigade (and cavalry!) repeatedly into suicidal frontal attacks, and to attack McPhersons Hill doesnt make much sense, it's well defended and no victory points there... and Oak Ridge is already taken:

https://www.dropbox.com/s/wpejeiznqei50sc/1st%20battle%2C%20frontal%20attack%20of%20cav%20and%20iron%2C%20on%20dynamic.jpg?dl=0

 

The artillery isn't positioned very well, Stevens cannons cannot shoot at Pettigrew but Pettigrew can shoot at him:

https://www.dropbox.com/s/59bvog3ebwsavaj/1st%20battle%2Cbad%20artillery%20positioning%2C%20on%20dynamic.jpg?dl=0

https://www.dropbox.com/s/a5843qoy4csqd8c/1st%20battle%2Cbad%20artillery%20positioning%2C%20on%20dynamic%202.jpg?dl=0

 

Other little issues or maybe just my personal preferences:

I know it's some kind of arcade strategy game, and I know most of it wont be changed anymore but I just note it down.

 

- Artillery units retreat very fast, a charging brigade has to run after them for a very long distance, the cannons are 100m away and the brigade reaches the heavy(!) cannons after 400m..something like this

- I can never be sure if an artillery is shooting or not. There's often an enemy brigade marked as red aim but is in fact out of range and there's no info about the status of the artillery. Thus I have to watch a (long) moment to see if they shoot or not.

- Personally I think the game is 5-10% too fast

- A brigade can only shoot to one direction, that's not so much a problem on the open field cause brigades used concentrated firepower... but a defending unit that's outnumbered by the attackers gets an 'additional' disadvantage, like if only one brigade has to defend a forested hill for example... Ofc a defending brigade that's outnumbered would lose after some time in any case, but it could hold the position a bit longer and reinforcements would get the chance to arrive in time.

- Brigades cannot entrench, again there's no problem on the open field, no problem at all cause brigades wouldn't entrench while they are in battle anyway... but it's another big additional disadvantage for a defending unit, even more if it's outnumbered! And if forests got a high defend bonus to simulate a heavily fortified unit would also mean that any attacking unit could easily retreat into such a fortified forest, either way it affects the balance and possible strategies.

 

 

However, it's a great game

[Nick Thomadis]

AI will get new mechanics with the coming hotfix patch (today/tonight will be ready if nothing strange happens) that will radically change its defense behaviour and will cure its overaggressive manner.

Thank you all for the continuous feedback. We will need your AI tests and after the new hotfix. The game will be supported and improved after release, so anything you report matters.

[fierman]

I just blew away the confederates on the first scenario. went around the lines with my videttes, after which the AI thought of them as a bigger danger, so Davis started running after the videttes all over the field. which left everything open for the union units to go wherever they pleased.

 

please: this issue had been discussed and noted earlier,. but there never was a reply on it from the devvers. I am close to just giving up with the game as a whole now.

[Harnis]

I just blew away the confederates on the first scenario. went around the lines with my videttes, after which the AI thought of them as a bigger danger, so Davis started running after the videttes all over the field. which left everything open for the union units to go wherever they pleased.

 

please: this issue had been discussed and noted earlier,. but there never was a reply on it from the devvers. I am close to just giving up with the game as a whole now.

with how overpowered Videttes are, the Confederate General's response was perfectly reasonable

 

The issue in that case is not the AI, it's the Videttes.

[fierman]

with how overpowered Videttes are, the Confederate General's response was perfectly reasonable

 

The issue in that case is not the AI, it's the Videttes.

 

Well, true.. i guess

[David Fair]

The issue is with the dev team - not the videttes or the AI.

 

There has been 10 months of complaints about the videttes and they have driven many people away from this game.

 

They aren't implemented in anyway that aligns with history.  

They aren't consistent with a brigade-level game.

They shouldn't have a Division Commander that roams the field like a Corp Commander.

They ruin the first phase by attacking glacial artillery that doesn't seem to be able to fire at them.

Their horses seem to be armored - you get them in range and only hit 5 of them with a blast from a 1,200 man brigade.

 

Horses could not survive on the ACW battlefield.  

Which is why Buford fought his men dismounted.

It is also why the cavalry fought on East Cavalry Field.

 

When cavalry actually fought on horseback, as they did at South Cavalry Field, they were wiped out - hence Kill Cavalry Kilpatrick.

 

The Union had a significant intelligence advantage at Gettysburg.  The Videttes could play this historically accurate role - but they don't.

[RichardMcCor]

0.96 + hotfix.

 

Seems to have made other AI improvements over 0.96.

 

AI on offensive where he seems to fight the union in day 1 scenario 1 the best.

 

Gets a minor victory now vs draw and worse previously

 

CSA commander now looks after his troops better.

AI brigade supported each other better

arty moved up in support much better

 

Still tried a few long range changes of poor Buckenbrough  right into Iron Brigade and Cutler with v bad results