Why is air combat limited by region?

[PorkchopSandwiches247]

This is one of those things that has always bothered me a lot. It makes no sense that planes with sufficient range have to be limited to a specific region with predetermined and arbitrary boundaries. A lot of the time it’s just foolish looking, the actual airfield location can be a couple of miles away from another ‘air region’, and yet your planes will not participate there.

This also makes air combat a huge pain sometimes. You constantly have to play whack-a-mole to stop enemy bombers and reassign fighters to whatever new region they pick every 10 seconds, even if youve designed a plane with enough range specifically to completely cover all of these regions. Even moving your planes to cover your troops as they advance gets tiresome, since the AI will be eyeballing what region has the least planes and then send all their airforce there. Creating another whackamole situation. Yes, you can assign planes to armies/generals, which helps a bit, but its still pretty unreliable…if you could do it to for field marshals maybe it would be better.

It really causes a lot of annoying microtasking the way it’s set up, and limits the usefulness of building high range planes when that range is completely wasted when you can only assign it to one small region.

Why not just have range be represented by the radius around an airfield? At least for certain missions like interception, if nothing else.

 

[BeauNiddle]

Short version - because it's a lot faster to process and the issues are obvious to the player and so they can plan round them.

If you implement flight radiuses around airfields then what happens if two battles take place either side of the radius? Do the planes only go to one? Do they split? (One on one there is little difference, but once you get to many on many combats radiuses become horrendous to keep track off and place combats in)

By having regions you KNOW where your superiority and interceptor planes are going to be and so you can chose to contest an area or not.

Downside is the whackamole situation we have for bombers, and the issue of fighting on region boundries. (it's also probably why planes didn't suffer disruption outside the target zone - too many questions about which planes get involved when and where, although they seem to have come up with plausible approximations now)

 

[Cavalry]

You constantly have to play whack-a-mole to stop enemy bombers and reassign fighters to whatever new region they pick every 10 seconds,

It is said many times, you are supposed to use some hundreds planes as intercept for each zone. The mechanics allow 100 interceptors can fight 1000 bombers no problem. The more planes you put into an air zone, the faster AI re act by putting more of its own or switch.

 

[PorkchopSandwiches247]

It is said many times, you are supposed to use some hundreds planes as intercept for each zone. The mechanics allow 100 interceptors can fight 1000 bombers no problem. The more planes you put into an air zone, the faster AI re act by putting more of its own or switch.

I mean yeah thats what I do. Its still annoying though, Id rather try to take them head on and wipe them out than this weird proxy defense

 

[SAS]

planes also fight and take damage in the zones enroute to any designated zones.

 

[Kanitatlan]

I don't know if I'm the only one here who has been present since the very beginning of HOI games but I've seen the path taken to get where we are now with the way air power works and this is a lot better than what went before. The real issue with complaints about the way air works is the problem of how on earth should it work to be better. We can all see the weaknesses but I find it difficult to come up with any reasonable suggestions for major improvements. What we could do with some incremental improvements. The biggest weakness is major failings in the simulation of German air defence against strategic bombing. Any aircraft operating air defence over friendly territory has much more effective range than other situations since it can land at a different air base from where it took off AND interception range is strongly related to how long a raid has been detected for. Bomber raids detected at the coast line of the Benelux region should allow aircraft further east to intercept over a much wider area. However, the recent changes, including the intercept enroute, have made massive changes to how poorly this operates.

I have spent a lot of time playing the Soviets and the "follow on war" used to result in allied bombers roaming everywhere west of the Urals flying from the UK. This has effectively stopped. I think this is for a variety of reasons but the intercept on the way is definitely a possible cause. It can be difficult to be sure why the AI has decided to stop doing things.

 

[HugsAndSnuggles]

The real issue with complaints about the way air works is the problem of how on earth should it work to be better.

Scrap airzones and have planes operate based only on their range?

 

[LeanLeaf]

This is one of those things that has always bothered me a lot. It makes no sense that planes with sufficient range have to be limited to a specific region with predetermined and arbitrary boundaries. A lot of the time it’s just foolish looking, the actual airfield location can be a couple of miles away from another ‘air region’, and yet your planes will not participate there.

This also makes air combat a huge pain sometimes. You constantly have to play whack-a-mole to stop enemy bombers and reassign fighters to whatever new region they pick every 10 seconds, even if youve designed a plane with enough range specifically to completely cover all of these regions. Even moving your planes to cover your troops as they advance gets tiresome, since the AI will be eyeballing what region has the least planes and then send all their airforce there. Creating another whackamole situation. Yes, you can assign planes to armies/generals, which helps a bit, but its still pretty unreliable…if you could do it to for field marshals maybe it would be better.

It really causes a lot of annoying microtasking the way it’s set up, and limits the usefulness of building high range planes when that range is completely wasted when you can only assign it to one small region.

Why not just have range be represented by the radius around an airfield? At least for certain missions like interception, if nothing else.

I agree with you in principle, but borders had to be drawn somewhere.

Using radius would be a way too complicated process to program and would require a lot of coding done to the actual simulated battles.

While not perfect, not by far, I agree this is a good compromise.

 

[Kanitatlan]

Scrap airzones and have planes operate based only on their range?

OK, I understand the idea but what you are really thinking is that the current system would work better if only the interactions of range with air zones didn't make it so crap. The problem is that as soon as we start trying to build a system around range so that, in effect, an air zone is the circular area an air fleet can fly to, we run into insurmountable problems with completing the rule set for how it should operate. Here are a few example issues.

1. How do we appraise air superiority and administer air combat? Are fighters engaging over the entire circle that is in range? What defines an area of aerial superiority. In the current system simply having aerial superiority has friendly bonuses and enemy penalties
2. If air units are assigned to ground attack missions then which ground combat do they support. When a ground battle takes place where does it draw air support from.
3. When fighters are assigned an intercept mission then which enemy air units do they intercept. Does a fighter force split up and try to intercept every enemy unit that passes in range? How do you evaluate who that is?

Now, even at this point I can see order O(N!) algorithms and order O(N^2) algorithms asking to implemented. Any such algorithm will (pretty much guaranteed) break the game. This is the fundamental problem with air power in a game. The scope of "immediate" movement for all the air units in the games is much to great to for the game algorithms unless there is a major simplification in operation. The air zone scheme is how that is achieved in HOI4. It separates the choice of where air units are going from the choice of targets to strike and simply by existing massively reduces the computational load for making it all work.

Hence my comment that if you want to suggest a major change you have to propose a complete system. Any superficial proposal is very easy to critique by highlighting all the algorithms it asks for that are computationally unreasonable. I've tried to think of alternatives myself and haven't proposed anything because I haven't come up with anything that doesn't have fundamental problems around practicality. The real problem the game has is that short range (early aircraft) demand smaller air zones and smaller air zones have a serious impact on the computational cost of running the air war.

 

[HugsAndSnuggles]

Here are a few example issues.

Are they? There are already algorithms in place that allow planes to fight in several airzones. There also can be several land and sea battles happening within the current airzones, and those get solved, somehow. All that's required is to shrink airzones to tile size, make planes fly missions in all "airzones" in range, and balance air combat lethality accordingly. Only potential issue is CPU usage, but I doubt either of us knows enough about inner engine workings to speculate about impact of this with any degree of accuracy.

 

[Kanitatlan]

Are they? There are already algorithms in place that allow planes to fight in several airzones. There also can be several land and sea battles happening within the current airzones, and those get solved, somehow. All that's required is to shrink airzones to tile size, make planes fly missions in all "airzones" in range, and balance air combat lethality accordingly. Only potential issue is CPU usage, but I doubt either of us knows enough about inner engine workings to speculate about impact of this with any degree of accuracy.

That doesn't really work. The current air combat in multiple zones is a special for intercepting air units enroute. It has a special characteristic which is that it is an interaction between the route of one group and another group operating in a specific zone. As soon as you make it an interaction between two groups both operating in multiple zones you convert your order O(N) algorithm to an order O(N^2) algorithm and performance wise you are screwed. I have an advantage here that I am both a mathematician whose specialist subject is the theory of games and a computer programmer with a local role as performance optimiser. I can simply see how game algorithms convert into computer algorithms and how the performance will scale. Shrink airzones to a single tile and the base computation cost of aspects of the air war just went up by a factor of 100-1000. That is unsupportable right of the starting block. The key step that is absolutely required to avoid performance issues is choosing where air units operate and then choosing which air units interact with specific ground battles has to be a two step process where the two choices are made separately and independently from each other OR they have to be poor choices. Moving away from using zones as they are now would likely lead to situations where half you battles don't get air support because the air units that could support them are, for no good reason, supporting another battle. Having the game make sensible choices is very hard or very expensive in computer resources. This means there are no simple solutions to how to change the way stuff works.

Having said that, there are probably good ideas that could work if they fit in with how the game engine works but they won't be based on making the air zones a lot smaller.

 

[HugsAndSnuggles]

As soon as you make it an interaction between two groups both operating in multiple zones you convert your order O(N) algorithm to an order O(N^2) algorithm and performance wise you are screwed.

Only if those calculations can not be spread across different cores (hence engine knowledge I mentioned). Might even end up having a negligible impact otherwise, considering the current bottleneck.

 

[The Colonel]

Just think also of the micro to assign planes without zones. Remember in hoi3 you could set them to operate in a cone instead of a circular radius? Imagine if that was necessary bc otherwise your planes were too spread out and your planes always got outnumbered in dogfights, so you'd have to constantly be turning the cones of every airbase to focus over the area of combat you care about most, then swapping them off that mode or turning again to deal with some bombers showing up and so on.

I agree the current system can be frustrating at times, but the other possible solutions are nightmares...

 

[HugsAndSnuggles]

Just think also of the micro to assign planes without zones.

I'd actually say it'd be less: there's no need for cones of any kind, just make them project power from airfield they are based in - no need to remember arbitrary boundaries and when switch between zones, which airfields work and which will cost you huge chunk of efficiency; will even make assigning CAS to army work pretty well.

 

[Kanitatlan]

Only if those calculations can not be spread across different cores (hence engine knowledge I mentioned). Might even end up having a negligible impact otherwise, considering the current bottleneck.

This is where my background in optimisation really pays off. The type of algorithm we are talking about here consists of lots of relatively inexpensive operations using a reasonably compact shared dataset. Any attempt to break out the steps into other threads is going to be surprisingly inefficient due to combination of expensive thread rendezvous costs, cache contention and data sharing issues. Decision making algorithms can be extremely difficult to speed up with multi-threading due to the basic issue that decision making requires bringing data together and sharing it. The best use of threading for this sort of thing would be to push the entire "decide which air units are supporting which battle" decision making into a separate thread for each country.

Just consider, the last change I made to one of the high performance software packages that I support consisted of noticing that a search string parameter was being shared between the worker threads and changed the code so each made a local copy. That worker stage was than 6% faster. That's the consequence of one single accidental data sharing that could be avoided. Using additional cores efficiently can be very, very difficult.

 

[goodcigar]

I imagine one thing they could do is give the player automation tools to deal with it. Currently it all has to be done manually by the player or not at all.

 

[Gort11]

I'd like to see the intercept mission able to target say, three air zones. That should solve the "bombers switching target region to avoid your interceptors" problem.