already requested Will there be AI/gameplay improvements to enable COIN-oriented gameplay?

[Temetre]

Its a bit hard to see how Afghanistans' focus on COIN gameplay is gonna be realized, with the current limitations of DCS' AI, ground units and air defense, as well as the very static/scripted mission structure.

(Wags said Afghanistan is intended to be a counter insurgency focussed maps, see the recent interview below:)

Spoiler

 

 

Edited April 1, 2024 by Temetre

[zerO_crash]

Afghanistan isn't intended to be COIN, it is COIN purely historically. There has never been a war in Afghanistan that was a country vs. country initative, in recorded history. This, due to the complexity of the region, chief amongst the complex socidemographic and geographic aspects. Resources is another important aspect which, to add to Wags interview, is also making the map relevant for Chinese (besides Russia and US). ^1 My point is, if to simulate historical incursions, the map is already there. Now, while simulating a proper scenario is a multi-aspect task (AI being component), there are already units which allow you e.g. visually create groups of insurgents (insurgent troops, insurgrnt vehicles, some badic buildings, albeit most not fitting this region, etc...). Most of it, will have to do with the creativity of mission designers. 

I will aldo point out that as per military/professional practices, missions typically only simulate that, one exact mission. It is rather modern in professional simulation to start creating complex mission scenarios which have manu components in one go. I mention this, because until multicore support, the amounts of units in a given mission, affected the performance severely. Now, however, I've tested a mission with more than 5000 units without any significant performance impact (highest difficulty AI). The only thing I noticed, was that the initial mission loading time was slightly longer. I will have to check this out more thoroughly, but it indicates that with time, we will be able to simulate individual battles with realistic force strengths across a map, which is seriously needed. I will add one thing here for ED, and that is the allowance of multiple factions in a mission, not only two. While we have few units to represent too many factions, such is the situation in real world as well (locals were often used to identify a friend or foe on the ground). 
 

The badic tools are there, it's really a matter of mission design. It pays to be clever. As an example, I'll add that some of the AI issues, can be alleviated with different AI difficulty. For example, if a given place has an enemy unit with APCs & tanks, and mission design doesn't predict any ground-battle, then you can lower the AI-level of those APCs and tanks in order to have them react slower and with less accuracy to air attacks. It's not perfect, but it works.

 

Also, have a look at the attached maps, it just goes to show how interesting the terrain is. Based on that information alone, one can create interesting missions.
 

^1 https://www.voanews.com/amp/afghan-oil-production-jumps-with-49-million-chinese-investment-/7473728.html (There are unspecific reports of Chinese offical delegates visiting Afghanistan regularly. Time will show what that develops into. Natural resources is one point that makes the region attractive, strategic is another.)

Map of ethnic groups anno 2005: https://www.loc.gov/today/placesinthenews/archive/2008arch/20080115_afghanistan.html

Edited April 1, 2024 by zerO_crash

[IMGX]

The problem with DCS is that it is not fun, it is a frustrating game, the AI is unbeatable, it has extreme aim, an amazing reaction capacity, from the ground they kill you just by looking at you, there is no visibility either, in multiplayer you cannot see the planes Enemies, I repeat: they are not seen until they kill you.

 The only fun thing in DCS is learning how to fly airplanes and their systems.

 Then, when you learn a module and want to put it into practice by playing, it turns out that playing is not fun, because you constantly get killed without even knowing who shot you.

[markom]

1 hour ago, IMGX said:

you constantly get killed without even knowing who shot you.

War sucks, doesn't it?

[IMGX]

6 hours ago, markom said:

La guerra apesta, ¿no?

If you like it like that...

But we must not forget that this is also a game, with limitations, visibility on a screen is not the same as the real world, neither are head movements, neither is spatial perception... For these reasons, these deficiencies must be compensated.

All you have to do is look at the online servers with a very low number of players, at the best time of the day how many are there? How many PVP servers exceed 30 players, 4 or 5? Here you can see that the multiplayer (which is a fantastic tool for attracting players) does not quite work.

[Temetre]

vor 10 Stunden schrieb zerO_crash:

Afghanistan isn't intended to be COIN, it is COIN purely historically. There has never been a war in Afghanistan that was a country vs. country initative, in recorded history. This, due to the complexity of the region, chief amongst the complex socidemographic and geographic aspects. Resources is another important aspect which, to add to Wags interview, is also making the map relevant for Chinese (besides Russia and US). ^1 My point is, if to simulate historical incursions, the map is already there. Now, while simulating a proper scenario is a multi-aspect task (AI being component), there are already units which allow you e.g. visually create groups of insurgents (insurgent troops, insurgrnt vehicles, some badic buildings, albeit most not fitting this region, etc...). Most of it, will have to do with the creativity of mission designers. 

I will aldo point out that as per military/professional practices, missions typically only simulate that, one exact mission. It is rather modern in professional simulation to start creating complex mission scenarios which have manu components in one go. I mention this, because until multicore support, the amounts of units in a given mission, affected the performance severely. Now, however, I've tested a mission with more than 5000 units without any significant performance impact (highest difficulty AI). The only thing I noticed, was that the initial mission loading time was slightly longer. I will have to check this out more thoroughly, but it indicates that with time, we will be able to simulate individual battles with realistic force strengths across a map, which is seriously needed. I will add one thing here for ED, and that is the allowance of multiple factions in a mission, not only two. While we have few units to represent too many factions, such is the situation in real world as well (locals were often used to identify a friend or foe on the ground). 
 

The badic tools are there, it's really a matter of mission design. It pays to be clever. As an example, I'll add that some of the AI issues, can be alleviated with different AI difficulty. For example, if a given place has an enemy unit with APCs & tanks, and mission design doesn't predict any ground-battle, then you can lower the AI-level of those APCs and tanks in order to have them react slower and with less accuracy to air attacks. It's not perfect, but it works.

 

Also, have a look at the attached maps, it just goes to show how interesting the terrain is. Based on that information alone, one can create interesting missions.

Aye, Im aware Afghanistan has a lot more going on than just more recent COIN stuff. Wags was however highlighting that its the modern map and will likely be mostly about COIN missions; if people will follow that logic or not is ofc another question, but it threw up the question.

The problem with mission design is that DCS only really supports heavily scripted and predictable missions; Im not sure how much this lends itself to the chaotic/unpredictable nature of COIN missions. AI enemies cant react dynamically to players' presence and actions, which imo would be a core feature of counter insurgency.

 

Besides the lacking of autonomy, im also concerned about the general AI. For example, ground units struggle to move in a straight line often enough without getting stuck, their situational awareness is not simulated. Neither is the allimportant factor of suppression, let alone realistic fragmentation damage from bombs. COIN often relies on the inferiority of anti-air of insurgents, but there is very little limitation to AA-fire in DCS; a single AK trooper might fire with the awareness and precision of a highly trained Gepard SPAAG crew. They also are not limtied by night-time, as another issue.

 

All in all I dont see the technical basis for COIN operations in current DCS. So I wonder if (or rather hope) were gonna get some big AI upgrades with that map coming?

[f15e]

On 4/1/2024 at 11:42 PM, IMGX said:

The problem with DCS is that it is not fun, it is a frustrating game, the AI is unbeatable, it has extreme aim, an amazing reaction capacity, from the ground they kill you just by looking at you, there is no visibility either, in multiplayer you cannot see the planes Enemies, I repeat: they are not seen until they kill you.

 The only fun thing in DCS is learning how to fly airplanes and their systems.

 Then, when you learn a module and want to put it into practice by playing, it turns out that playing is not fun, because you constantly get killed without even knowing who shot you.

What are you talking about? DCS ground AI and SAM sites are exceptionally stupid, they don't even have a quarter of the capabilities they have in real life. What do you mean ground units kill you by just looking at you? It's hard to get hit by AAA on purpose let alone by accident. You can not see enemy planes in multiplayer??? 

[Tippis]

18 minutes ago, f15e said:

What do you mean ground units kill you by just looking at you? It's hard to get hit by AAA on purpose let alone by accident.

Anti-air sure. Regular ground units that have no business firing at — much less hitting — aircraft? Not so much. A bunch of them have pretty much perfect aim and no launch or firing warning — you just hear a “clunk” out of nowhere as their laser 23mm cannons rip holes in your engine. Or they launch ATGMs at you from beyond max missile flight range. They may not have all the capabilities they should have IRL, but there's also an annoying abundance of capabilities that they definitely don't have for real.

It's very likely that anti-air have actually been given a bunch of limitations and “anti-abilities” to make sure they can't aim as well and that they're restricted by their on-board sensors (or lack thereof), whereas a number of ground units (and some air units, although tat has largely been cleaned out) just have perfect knowledge. It makes some sense from an optimisation standpoint: why have them process a bunch of stuff to make them worse at a target they shouldn't even engage with the weapons they have… but then the AI doesn't know that it shouldn't, so when it does, it does so exceptionally well.

[cfrag]

15 hours ago, f15e said:

What do you mean ground units kill you by just looking at you?

I believe @IMGX has a point when we talk about infantry who can kill you (the pilot) with a single shot to the head from a mile away, using their rifle. It can be very, very frustrating. It's a bit off-putting that pilots on CAS duty (not to mention helicopter pilots) have to worry more about infantry than AAA or SAM. Hopefully we'll get an overhaul of that part of the AI. I, too, deem it problematic, but that's just IMHO.

Edited April 8, 2024 by cfrag

[SharpeXB]

On 4/1/2024 at 5:42 PM, IMGX said:

I repeat: they are not seen until they kill you.

Because in real life your enemies always make sure you can see them so it will be a fair fight?

[zerO_crash]

On 4/2/2024 at 9:41 AM, Temetre said:

Aye, Im aware Afghanistan has a lot more going on than just more recent COIN stuff. Wags was however highlighting that its the modern map and will likely be mostly about COIN missions; if people will follow that logic or not is ofc another question, but it threw up the question.

The problem with mission design is that DCS only really supports heavily scripted and predictable missions; Im not sure how much this lends itself to the chaotic/unpredictable nature of COIN missions. AI enemies cant react dynamically to players' presence and actions, which imo would be a core feature of counter insurgency.

 

Besides the lacking of autonomy, im also concerned about the general AI. For example, ground units struggle to move in a straight line often enough without getting stuck, their situational awareness is not simulated. Neither is the allimportant factor of suppression, let alone realistic fragmentation damage from bombs. COIN often relies on the inferiority of anti-air of insurgents, but there is very little limitation to AA-fire in DCS; a single AK trooper might fire with the awareness and precision of a highly trained Gepard SPAAG crew. They also are not limtied by night-time, as another issue.

 

All in all I dont see the technical basis for COIN operations in current DCS. So I wonder if (or rather hope) were gonna get some big AI upgrades with that map coming?

You didn't read closely what I wrote - Afghanistan, isn't designed by ED as a COIN environment, it is naturally a COIN environment, as there have never been any wars where Afghanistan figured as an opponent to a foreign attacked.

 

There are actually many good missions, which have a randomness-factor associated with the mission build. That said, that factor is still designed by a mission creator. In other words, it is not a run-the-slot-machine-and-see-what-comes-out type of randomness (AI). Still, some well-designed mission do offer events occuring with a possibility tag attached to them. The problem is really that it requires knowledge of scripting, to perform. It is not a default tool in the mission editor, and that is a valid complaint. On the other side, who has ever worked with professional simulation, knows that scripting is the only way to go. That's just how it works. Regardless, if it was easier to build mission, based on observations from the mess consumers enacted in the "Afghanistan"-forums, I would not hold my breath for any more content or of higher quality. Until recently, one could have held their heads high up in this community, but with the recent chain of events (people proving they won't even read what's written), I see a different reality overall. Just a shame really. 
 

I digress; often, complaints orbit around ED not delivering content fast enough. Well, if Wags didn't have enough to do already (speculation), having to additionally go back to oral/visual content-making, indeed goes against what the community complains about all the time. To put it straight - because people are illiterate/not bothering to read, more manpower has to be diverted to making the product more "consumer"-proof. That, instead of allowing ED to divert all their capacity to relevant priorities. It is just a ridiculous turn of events, and all thanks to the grand wish of community. 

 

With regards to ground/sea/etc... operations, that had not been a priority, which is why it lags behind. If you'd ask customers about better AI vs. a new module, I imagine most would chose the second, purely based on instinct. ED has confirmed that it will be improved, but as everything, ot takes time. Arguing over something recognized, is pointless in itself. It will come. Patience, is key. In the mean time, I did list a couple of ways to mitigate thosr obstacles. The tasking in missions, is often limited to the designer's capacity and vision. Again, we already have tools to do many types of rather advanced scenarios, albeit not as easily available.

Edited April 9, 2024 by zerO_crash