Does the Soviet AI need help? (Full Version)

All Forums >> [New Releases from Matrix Games] >> Strategic Command Series >> Strategic Command WWII War in Europe



Message

elmo3 -> Does the Soviet AI need help? (11/4/2016 7:52:51 PM)

I've been watching the video series on Youtube by Night Phoenix. Very entertaining. However it looks like the Soviet AI really needs help, especially in the Ukraine. So far it has left Chernigov, Gomel, Krivoy Rog and the nearby mines, Bryansk, Poltava, Sumy, Kursk, Kharkov, Orel, Kaluga, Tula, and even Dnepropetrovsk which is a NM Objective all undefended!

In short there were virtually no Soviet defensive forces south of Moscow all the way to the Crimea. Not sure if the AI is not building forces, or if it is building them and then moving them off to the middle of nowhere. Hopefully this has been addressed already, or if not then by the time of launch.



ILCK -> RE: Does the Soviet AI need help? (11/5/2016 2:29:23 AM)

He was destroying massive numbers of units, my impression was there was nothing for the AI to play there and it focusing on the locations that mattered most.

Comparing his steamroller with what happened to ParaDogs in his review it feels like the AI plays pretty well. ParaDogs is in a lot of trouble because he was unable to focus on Russia and didn't get that massive wipe out of units and seizure of MPPs early and the AI is punching back nicely.



Biker1984 -> RE: Does the Soviet AI need help? (11/5/2016 10:37:43 AM)

There are some good things and some bad things I have seen by the "Soviet" AI.
The good things is that they focus their attacks to weaken the units and trying to destroy it (combined with the airforce), so they did a good offensive in the spring 1942 almost destroying a panzer core. Second they fortify there defense line very good as such they are hard to beat after 1942.
At the same time I have also seen constantly attacking "the supposed weaker unit" as a weakness. I have seen the soviets attacking a well defended infantry division behind a river only inflicting damage between 1 to 3 points and not doing any damage to the Germans for 5 separate units. Even if they have done 1 strength point of damage it would not add up to the 5 to 10 points they have lost during those attacks. In addition they use strategic railway movement way to much about 2-3 times per turn, which cost them significantly MPP. I don't mind that they use strategic railway movement, but if the cost would be the same as the Germans a lot of valuable MPP is lost.
I'm also not that impressed by the USA navy at this point of the game. I woudl say if London has fallen, the USA declares war on Germany and sends its complete fleet to hinder the German invasion. Sealion against the AI is to easy at this point.



Toby42 -> RE: Does the Soviet AI need help? (11/5/2016 1:38:24 PM)

From the latest Paradogs video I think that the Soviet AI is doing fine! I think that the Axis is in trouble in Russia?



elmo3 -> RE: Does the Soviet AI need help? (11/5/2016 7:28:26 PM)


quote:

ORIGINAL: Treale

From the latest Paradogs video I think that the Soviet AI is doing fine! I think that the Axis is in trouble in Russia?


That could be but it looks to me like Night Phoenix is playing a newer version of the game so I was wondering if something changed in the Soviet AI that was causing it problems.



Ancient One -> RE: Does the Soviet AI need help? (11/5/2016 10:07:10 PM)

quote:

ORIGINAL: elmo3


quote:

ORIGINAL: Treale

From the latest Paradogs video I think that the Soviet AI is doing fine! I think that the Axis is in trouble in Russia?


That could be but it looks to me like Night Phoenix is playing a newer version of the game so I was wondering if something changed in the Soviet AI that was causing it problems.


Honestly, it looks like most of the difference is that Night Phoenix is a more skilled player than Paradogs at this game (no offense to the latter).



ILCK -> RE: Does the Soviet AI need help? (11/6/2016 12:44:36 AM)


quote:

ORIGINAL: Zagys

quote:

ORIGINAL: elmo3


quote:

ORIGINAL: Treale

From the latest Paradogs video I think that the Soviet AI is doing fine! I think that the Axis is in trouble in Russia?


That could be but it looks to me like Night Phoenix is playing a newer version of the game so I was wondering if something changed in the Soviet AI that was causing it problems.


Honestly, it looks like most of the difference is that Night Phoenix is a more skilled player than Paradogs at this game (no offense to the latter).



NP was more ready to launch Barbarossa was a big part of it. USSR declared war on PD and he wasn't ready in the east. Good example of the risk of SeaLion.

That said, I had hoped PD was on the newer version because the UI in the PD video looks a lot better than the one in the NP videos.



Ancient One -> RE: Does the Soviet AI need help? (11/6/2016 1:08:59 AM)


quote:

ORIGINAL: ILCK
That said, I had hoped PD was on the newer version because the UI in the PD video looks a lot better than the one in the NP videos.


I hope not, because it doesn't look like that one has the "MAX" button for reinforcements.



elmo3 -> RE: Does the Soviet AI need help? (11/6/2016 1:36:09 AM)


quote:

ORIGINAL: Zagys

I hope not, because it doesn't look like that one has the "MAX" button for reinforcements.


That and the better looking research screen is what made me think Night Phoenix is using a more current build.



Hubert Cater -> RE: Does the Soviet AI need help? (11/6/2016 12:49:35 PM)

Thanks for the questions everyone and I thought I'd post a quick comment on the Axis invasion of the USSR and what to expect from the Soviet side, i.e. against the AI or a human player.

Currently the campaigns have been designed/balanced with the historical timelines in mind and this just means that a well thought out and planned Operation Barbarossa should see the Axis player eliminate scores of Soviet units, with limited to no losses to their own side, and ideally reach the gates of Moscow by the first winter of 1941. At that point the Axis should suffer some effects of overstretching their supply lines, the Soviet Winter effects, and the Soviet lines should somewhat stabilize with the arrival of the Siberian units as well as any repurchases on their own end.

Under the right circumstances, and at this point in the game, you should see some ebb and flow on that front until the end of the war with either the Axis finally able to tip the balance in their favor, or the Allies tipping the balance and pushing back on all fronts from their side.

We've seen examples of two games, one from PD, and one from NP, and if Barbarossa is not initially a very strong push from the Axis, then the slow build up and eventual push back will happen earlier from the Soviet side, while if Barbarossa is more like NP's game then what you see there is typically expected and if the Soviet AI has run out of defenders from the early attacks then the Axis may find some locations undefended as it pushes forward to its historical objectives. Again this would be the same in either a human vs human or versus an AI game as the Soviet side only has so many initial units, with so much territory to cover, and it takes time to rebuild from your losses in game.

Of course this all being said, we also have to take into account different skill levels and some players will find the Soviet theater playing out just perfectly, while others will be better off playing at one of the higher difficulty levels and providing the AI with some help like additional MPP collection and/or experience and so on, especially after a few games and once the game mechanics and strategies settle down for you etc.

For the excessive operational movement, this was a bit of a bug in these betas where the AI was constantly attempting to redress the "weak fronts" by sending units back and forth in an attempt to balance them out, and yes this was an unnecessary expense that ultimately hurt the AI in terms of MPP costs, so consider this fixed and much improved from these beta playthroughs.

For the US naval side, it is still early in game and the US navy does take into account the loss of the UK home island and is planning accordingly as it not only has the responsibility of counterattacking, but also to screen and shield any potential US infantry transports or amphibious assault vehicles... so there will eventually be some action but again here the AI plays for the long game much like how a human player will do as well.

Not saying that there is never any room for improvements, just outlining some of the thought process that goes into the AI as there is always so much for it to cover from game to game and from outcome to outcome.

Hope this helps,
Hubert




apec -> RE: Does the Soviet AI need help? (11/6/2016 12:54:51 PM)

There is the excellent Barbarossa AAR by HvS, I suspect he is using a more recent beta version. He is playing very well, nonetheless the AI looks more challenging campared to what I have seen in the video.



BillRunacre -> RE: Does the Soviet AI need help? (11/6/2016 2:52:43 PM)

Yes, HvS's version is more recent than the video series. He is also an extremely good player!



elmo3 -> RE: Does the Soviet AI need help? (11/6/2016 3:51:09 PM)

Thanks Hubert.

I have not followed the HVS AAR in detail, just browsing, but will go back and read through it since he has a newer build.



ILCK -> RE: Does the Soviet AI need help? (11/6/2016 4:01:04 PM)


quote:

ORIGINAL: Hubert Cater

Currently the campaigns have been designed/balanced with the historical timelines in mind and this just means that a well thought out and planned Operation Barbarossa should see the Axis player eliminate scores of Soviet units, with limited to no losses to their own side, and ideally reach the gates of Moscow by the first winter of 1941. At that point the Axis should suffer some effects of overstretching their supply lines, the Soviet Winter effects, and the Soviet lines should somewhat stabilize with the arrival of the Siberian units as well as any repurchases on their own end.



Trying to see this in the videos and haven't been able to get a good view. As you cut off pockets of units they suffer some supply penalties depending on if they are either in a city of the pocket has a city but are there any other effects in terms of readiness and morale to being out of supply?



BillRunacre -> RE: Does the Soviet AI need help? (11/6/2016 9:04:34 PM)

Being low on supply reduces readiness and morale and the ability to reinforce, so overall it makes these units much easier to destroy, while suffering fewer casualties yourself.



Biker1984 -> RE: Does the Soviet AI need help? (11/6/2016 9:57:52 PM)

quote:

ORIGINAL: Hubert Cater

Of course this all being said, we also have to take into account different skill levels and some players will find the Soviet theater playing out just perfectly, while others will be better off playing at one of the higher difficulty levels and providing the AI with some help like additional MPP collection and/or experience and so on, especially after a few games and once the game mechanics and strategies settle down for you etc.

For the excessive operational movement, this was a bit of a bug in these betas where the AI was constantly attempting to redress the "weak fronts" by sending units back and forth in an attempt to balance them out, and yes this was an unnecessary expense that ultimately hurt the AI in terms of MPP costs, so consider this fixed and much improved from these beta playthroughs.

For the US naval side, it is still early in game and the US navy does take into account the loss of the UK home island and is planning accordingly as it not only has the responsibility of counterattacking, but also to screen and shield any potential US infantry transports or amphibious assault vehicles... so there will eventually be some action but again here the AI plays for the long game much like how a human player will do as well.



Thanks a lot for addressing my concerns here Hubert and I'm absolutely going to purchase this game as I like the dept and the overall handling the AI specific problems. Great to hear that the operational movement for the AI has being addressed.
I may have missing it but I have not seen the Soviet winter in PD games. Are there certain prerequisites that would trigger this?
What I also have seen is that the Sealion is a success and the UK has no more ground units in Britain, the US will send in a lot of aircraft. I would expect this to happen if the UK has still a lot of ground units (so during a historical accurate gameplay).
Is the AI counting that a possible Axis Sealion is successful and if yes how would the US react to that?



ILCK -> RE: Does the Soviet AI need help? (11/6/2016 11:25:01 PM)

I saw the Soviet Winter in one video. Movement grinds to a halt and Germans units take actual damage as well. Not sure if there are readiness and morale penalties along with it.



Jim D Burns -> RE: Does the Soviet AI need help? (11/7/2016 7:47:39 PM)

quote:

ORIGINAL: Hubert Cater
and if the Soviet AI has run out of defenders


The notion that the Soviet Union would be devoid of units at any time in the war is just silly. Between June-December 1941, the Soviets mobilized almost 10 million reservists and volunteers. They lost about 6 million in the same time period so they almost doubled the size of their at start forces in 5 months.

The soviet Union didn't suffer manpower shortages until near the end of 1943, and that wasn't even close to a situation where the country was devoid of units. The U.S.S.R never came close to a situation where it couldn't keep a solid front line after the first month or so of the war.

http://operationbarbarossa.net/wp-content/uploads/2014/06/Soviet-FILARM-General-Structure.pdf

quote:

Between 22nd June and 31st December 1941 the USSR called up 5 500 000 of its reservists
and conscripts into active service. In addition another 4 000 000 men and women ‘volunteered’ for
militia or volunteer units, and most of these ended up in the Red Army. This easily remains the
largest and fastest wartime mobilisation by any country in history. The overall Soviet forces
mobilised in the second half of 1941 were actually larger in personnel terms than the entire Soviet
armed services that existed on 22nd June 1941 (although in equipment terms, especially tanks and
aircraft, they were considerably weaker). This was just as well because by November 1941 the Red
Army and Air Force that existed on 22nd June 1941 (then already both the largest in the world) had
been all but destroyed. In this chapter we will focus on the immense Soviet ground forces that were
mobilised between 22nd June and 31st December 1941.



Edit: A good way to possibly represent the massive manpower pools and faster than normal mobilization would be to make Soviet Corps free auto-builds. So if a Corp dies it automatically goes on the production track and perhaps make them arrive a turn or two early. Making the Corps slightly weaker than most countries Corps would help mitigate the free builds. The goal being to allow the soviets to throw cheap/weak defensive units in front of an attacker without gaining too much attack potential.

Jim







Scook_99 -> RE: Does the Soviet AI need help? (11/7/2016 9:29:37 PM)

Thank you for your insights to the AI, Hubert, there is a noticeable difference between this version and SC2. I was thinking about this...does the AI break down its actions by country, or as a group. Because, I was wondering if there is the ability to set the AI for each country individually? I would expect not, but just thought I would throw that out there to intrigue and terrorize your thoughts simultaneously.



BillRunacre -> RE: Does the Soviet AI need help? (11/8/2016 2:55:57 PM)


quote:

ORIGINAL: ILCK

I saw the Soviet Winter in one video. Movement grinds to a halt and Germans units take actual damage as well. Not sure if there are readiness and morale penalties along with it.


Yes, unit morale and readiness take a hefty penalty too. [:)]



BillRunacre -> RE: Does the Soviet AI need help? (11/8/2016 3:02:12 PM)


quote:

ORIGINAL: Jim D Burns

Edit: A good way to possibly represent the massive manpower pools and faster than normal mobilization would be to make Soviet Corps free auto-builds. So if a Corp dies it automatically goes on the production track and perhaps make them arrive a turn or two early. Making the Corps slightly weaker than most countries Corps would help mitigate the free builds. The goal being to allow the soviets to throw cheap/weak defensive units in front of an attacker without gaining too much attack potential.

Jim


Hi Jim

Soviet Corps and Armies are inferior and cheaper than everyone else's. That's an interesting idea to make them quicker to build too, I'll give it some thought. Thanks. [:)]

Bill



ILCK -> RE: Does the Soviet AI need help? (11/8/2016 5:47:29 PM)


quote:

ORIGINAL: Bill Runacre


quote:

ORIGINAL: ILCK

I saw the Soviet Winter in one video. Movement grinds to a halt and Germans units take actual damage as well. Not sure if there are readiness and morale penalties along with it.


Yes, unit morale and readiness take a hefty penalty too. [:)]



So you are saying this is .... bad? [;)]


Just as a hypothetical how do morale and readiness affect damage? Obviously lower will lower effectiveness but in practical terms is a unit at 50% readiness and 50% morale 50% as effective as a 100% in both counterpart?



BillRunacre -> RE: Does the Soviet AI need help? (11/8/2016 8:00:25 PM)

Yes, units with low morale and readiness will fare significantly worse in combat, with these %s being proportional to their fighting ability all other things being equal.

It is bad for the Axis, and a good time for the Red Army to consider counter-attacking, or it can just enjoy a break from the Axis onslaught to improve its defences.



Hubert Cater -> RE: Does the Soviet AI need help? (11/8/2016 8:05:14 PM)

quote:

The notion that the Soviet Union would be devoid of units at any time in the war is just silly. Between June-December 1941, the Soviets mobilized almost 10 million reservists and volunteers. They lost about 6 million in the same time period so they almost doubled the size of their at start forces in 5 months.


To clarify, and I'm also not disagreeing with the above, mine was just a commentary on the current in game situation explaining what might be happening for the AI.

However, in game terms, any lack of "on map" defenders after the initial Axis assault is really only temporary, and more of a functional abstraction representing the core loss of effective combat units in the short term. Their delayed re-appearance, as well as newer units, back into the game reflects the time needed for reorganization, restrengthening, and replacement of units back into the field. This abstracted disarray on the Soviet side (after the initial wave of the Axis assaults) also allows for the Axis to maintain their timelines, i.e. their blitzkrieg forward into Soviet territory as otherwise it would be very difficult to reach historical objectives by the end of 1941 at the current game turn lengths.

Essentially the Soviet manpower is still there, it is just in this game represented by not only units on the map, but also units being rebuilt and built and reflected in the delayed time it would take to get them back into the field. For example, many destroyed units can be bought back at reduced cost and at a reduced production time to reflect this.

That being said, there are ways to do it differently, however it would require probably 4 times as many turns in a year (if not more) as well as combat values adjustments and then there would be no way to have the overall game length play out in quite as an enjoyable and reasonable time frame. You would have increased unit concentrations, not necessarily a bad thing as this does eventually occur by war's end, but I doubt you'd have as much of the "tactical" feel that the game offers, and at this strategic level which many players seem to enjoy quite a bit as well. For example, that blitzkrieg re-creation feeling you get in game at the start of Barbarossa as you sweep through the USSR and so on.

Sometimes it is a careful balance, to get playability vs history just right, but I think you'll find that it plays out quite well overall.



Hubert Cater -> RE: Does the Soviet AI need help? (11/9/2016 1:26:18 PM)


quote:

ORIGINAL: Scook_99

Thank you for your insights to the AI, Hubert, there is a noticeable difference between this version and SC2. I was thinking about this...does the AI break down its actions by country, or as a group. Because, I was wondering if there is the ability to set the AI for each country individually? I would expect not, but just thought I would throw that out there to intrigue and terrorize your thoughts simultaneously.


Thank you and yes the AI can indeed be set for each major power and in fact is in many cases in game.

There are large portions of the AI where I simply let it assess the situation on its own, i.e. most land combat OFFENSIVE/DEFENSIVE planning is set this way outside of specific GUARD scripts that I run to hold certain areas for X number of turns under Y sets of conditions, but at other times we can fine tune specific behavior such as certain OFFENSIVE actions we'd like to see on land if needed.

However, outside of land combat, FLEET, TRANSPORT, AMPHIBIOUS, PURCHASE, RESEARCH and DIPLOMACY planning is much more specific and on a per major basis.



Jim D Burns -> RE: Does the Soviet AI need help? (11/10/2016 12:27:06 AM)

quote:

ORIGINAL: Bill Runacre
That's an interesting idea to make them quicker to build too, I'll give it some thought. Thanks. [:)]


There is plenty of historical justification for this. If my memory hasn't completely failed me I think they instituted a 2 year mandatory military service policy in either 1936 or 1938. So by wars start there were tens of millions of young Soviets that had already been through their two year service obligation and creating combat formations would be far faster than normal conscription for most nations since basic and advanced training courses weren't needed. A brief refresher period to get the new formation acclimated with itself would be all that was needed.

This is why the Soviets were able to grow their military so fast. Sure the militia units and some of the fresh conscripts needed regular training before they could be considered capable regulars, but they had millions of already trained reservists to draw from as well and I think it probably was a big help in saving Russia from utter collapse during the big defeats in 41 and 42.

Jim



jjdenver -> RE: Does the Soviet AI need help? (11/16/2016 3:14:14 AM)

One thing that I noticed while reading the Barbarossa AAR that's going up now is that it seems like the Axis don't take much attrition when attacking. Historically even unorganized Soviet formations that were encircled, pierced, and generally torn to pieces by the Wermacht in 1941 were able to inflict heavy losses on German units. This combined with the difficulty the Germans had while running their tanks across the SU on the offensive, led to German units bled white by late 1941. This happened again in 1942 and 1943. Even when winning the Germans took terrific losses.

I don't see this in the AAR. I think that a successful attacker seems to emerge mostly unscathed, maybe dropping 1 or 2 strength points occasionally. This seems to lead to a lot of problems simulating the war in the east as the attrition is extremely one-sided. This also I think snowballs by 1942 (and I'm guessing 1943 also) because German units build up more and more experience while Soviet units aren't able to do that. By end of 1942 historically the Soviets were able to aquit themselves quite well against German formations but I don't see this in the AAR. I think this lack of operational toll on the Germans doesn't "feel" right.

About the suggestion earlier that Soviet corps get immediately put back into production, perhaps as weaker units might only exacerbate this problem. The Soviets need to be able to inflict _casualties_ on the Wermacht, to grind it down as they did historically and cause the Germans to have lots of problems replacing those losses, to have to choose to rebuild/refit only parts of their army in Russia, to slow down as the year campaign season progresses and losses mount.

I think there needs to be a re-think of this area in the design because currently it seems the defender doesn't cause many losses. They can only cause losses when they counterattack - which just isn't historical.

My 2 cents from reading the AAR. I want to love this game but I'm worried. :)



HvS -> RE: Does the Soviet AI need help? (11/16/2016 6:16:19 AM)

Hi jjdenver!

What you write about the German losses in my AAR is part of my strategy. It is possible to do panzer sweeps and encirclements in this game, but I usually advance without these things. There are a couple of reasons, and increasing losses are one of them. The engine allows to do these things, but they lead to the same developments as in RL: higher losses for the attacker.
There is one major reason why I have less losses than the Wehrmacht in RL: I concentrated my forces (no attack on Wjasma and Moskau in 1941), and I stopped at Orel. I always pay attention on supply and unit attrition, so if a unit drops below a certain strength, I refresh it. This is did not happen in RL. My advance in 1941 was not as far as in RL, but my troops were in better condition.
But the main reason for my low unit attrition is my attack doctrine. There are certain parts of the front line (Northern Finland, Smolensk, Orel) where I don't attack since months. I just use the Prepared Attack bonus to decimate enemy units. To be more precise: I only execute attack with own loss forecast zero. If the forecast is 1:4, I don't attack. So there are very few losses on these fronts.
The second reason is the use of Bombers on parts of the front where I attack. I designate some key units (usually tanks) and send my bombers first, and when my ground units attack, the Soviets are already at half strength.
I will post Turn 47 later today, and if you look at the German units around Stalingrad, you will see a couple of German ground units at strength 4 or 5. There are losses when attacking, but usually I was able to refresh these quickly. But it does not work at Stalingrad, because my bombers are too far away to execute my usual strategy.
One last point: you mentioned that the Soviets are able to fight back against the German units in late 1942. I think this is modeled when the Soviets get Infantry Weapons 3 which happened in Turn 46... they will more difficult to destroy.



GavCol -> RE: Does the Soviet AI need help? (11/16/2016 11:33:15 AM)

Likewise I'm worried by this. In previous SC games once a steamroller of tech and experience was built up it was nearly impossible to stop.
The Barbarossa AAR shows the same tendency with minimal attack losses and excellent supplies where has are.
The AAR has reached a tipping point due to this and that is despite the computer being handed support points in the set up.

Appreciate the focused attacks by hvs but no supply restrictions or attrition.

Worried on this front....rem that Kursk needed a supply concentration that took away from other fronts

Still a buyer but did want the agony of picking focal points
Thanks



BPINisBACK -> RE: Does the Soviet AI need help? (11/16/2016 11:55:14 AM)

I don't see the point.
As HvS has said he is advancing very slow. When i saw him playing i was wondering myself for the reason of that behaviour... As he has explained before, he could go really faster... but, as he said: that means less supplies, the HQ would be far behind his lines, the frontline would be very irregular... so, more losses.
All we have the RL in our mind...but this is game... Futhermore, a game between an experienced human and a machine... [:'(]
You can always watch the Paradox videos... There, he is advancing bit a bit... The games are completely different but in Paradox game, the reds seems to have a huge amy somewhere behind the lines waiting for the CA.
And, as Hubert said, the AAR from HvS and Paradox has been played with an early beta (if i am not wrong).
So, finally, of course that a experimented human can kick the "ass" of the computer!!! But, let's see what happend with Human with Human AAR. After them, it is always "easy" to adjust the game.
I will buy the game, for sure! [&o]

Excellent reports, BTW!!! [&o]



Page: [1] 2   next >   >>

Valid CSS!




Forum Software © ASPPlayground.NET Advanced Edition 2.4.5 ANSI
1.75