Flames and Steel (official thread)

KurtGodel7

After a very interesting conversation with AI, I am now considering the following system.

Aircraft are allocated to theaters, as now. For example, the Eastern Front is the Luftwaffe against the Soviet Air Force.
There is a dogfight phase. Only air superiority aircraft participate in dogfights. This means fighters, advanced fighters, jets, and (new unit type!) heavy fighters. During the dogfight phase, all aircraft fire at their dogfight values. Destroyed aircraft are removed.
Then, there is an interception phase. Fighter-type aircraft fire at their interception values. Bombers fire back at their defensive fire values. Destroyed aircraft are removed. Damaged bombers are aborted.
Any bombers which weren't destroyed, and which weren't aborted get to bomb whatever targets they had in mind. Strafing armies, attacking navies, or strategic bombing raids.

To make this more concrete: a fighter would have a dogfight value of 3 and an interception value of 3. A heavy fighter would have a dogfight value of 2 and an interception value of 4 or maybe 5. Also it would cost a little more than a normal fighter.

I don't know for sure how this will actually work without play testing. But here is what I hope would happen. The Soviet player sees that Germany is able to allocate, say, 10 fighters and 6 dive bombers to the eastern front. The Soviets have 5 fighters, a dive bomber, and a heavy fighter they could allocate. The dive bomber hides in Siberia so it won't be shot down during the interception phase. The rest of the Soviet air are sent to the eastern front. The Soviets lose more air than the Germans during the dogfight phase. But, at least the German dive bombers didn't get to fire during the dogfight phase, so that made it a bit more even. Most of the Soviet fighters survived the dogfight, earning the chance to intercept the German dive bombers. The chance to kill or abort some of the dive bombers helps balance out this air battle. This balancing out is what induced the Soviet player to participate in the air battle in the first place, instead of hiding his entire air force in Siberia.

Vendetta

@KurtGodel7 said:

The system sounds elegant in theory, but one issue that stands out even before any play testing is that if you’re the weaker power, there is zero incentive to build anything but fighters - bombers of any sort seem useless if you don’t have air superiority. And while there was eventually a point in the historical WW2 where Germany became desperate enough to adopt that as official production policy, that wasn’t the case for most of it - most powers were still trying to field both offensive and defensive capabilities simultaneously.

The Soviets never parked their Il-2 fleet out in Siberia to wait until the VVS had won air superiority - they used it, constantly, accepting the heavy losses that entailed. So a system that encourages players to do that is rewarding unrealistic behavior.

A few ideas on ways to address that:

A) Shrink your air theaters. Instead of one for the entire Eastern Front, imagine having about three different air zones there (north, center, south). More zones means fighter coverage is spread thinner, creating opportunities for the weaker side to gain local superiority - the Germans might prevail in two of the zones, but the Soviets could concentrate their forces to win control over the third one.

B) Allow fighters on escort missions to take hits meant for the bombers they’re protecting, and allow those bombers to carry out their strafing missions if they make it through even if their side has lost the battle for air superiority overall. Systems with all or nothing results (only the side that wins the air battle gets to bomb) naturally encourage unrealistic min-maxing style gameplay.

C) Use your unique tech system to give the Soviets some advantages that encourage a mixed air strategy instead of just spamming fighters - you could make their dive bomb bombers harder to shoot down, or you could make them cheap enough to be worth using aggressively even if the losses are heavy. You would also have to do that for any other cases where the system is optimizing for less interesting play.

A is probably the easiest route to test since it doesn’t involve any other changes to the system, you just play out a few test games and redraw the air zone boundaries between each to see which results in the most interesting game dynamics. B is the next most promising, it also changes up just one element, but that element is one of your system’s core attributes. C is probably a development trap where you’d end up having to constantly fiddle around with a bunch of different bespoke elements you’ve come up with to patch different issues.

KurtGodel7

@Vendetta said:

@KurtGodel7 said:

The system sounds elegant in theory, but one issue that stands out even before any play testing is that if you’re the weaker power, there is zero incentive to build anything but fighters - bombers of any sort seem useless if you don’t have air superiority. And while there was eventually a point in the historical WW2 where Germany became desperate enough to adopt that as official production policy, that wasn’t the case for most of it - most powers were still trying to field both offensive and defensive capabilities simultaneously.

The Soviets never parked their Il-2 fleet out in Siberia to wait until the VVS had won air superiority - they used it, constantly, accepting the heavy losses that entailed. So a system that encourages players to do that is rewarding unrealistic behavior.

A few ideas on ways to address that:

A) Shrink your air theaters. Instead of one for the entire Eastern Front, imagine having about three different air zones there (north, center, south). More zones means fighter coverage is spread thinner, creating opportunities for the weaker side to gain local superiority - the Germans might prevail in two of the zones, but the Soviets could concentrate their forces to win control over the third one.

B) Allow fighters on escort missions to take hits meant for the bombers they’re protecting, and allow those bombers to carry out their strafing missions if they make it through even if their side has lost the battle for air superiority overall. Systems with all or nothing results (only the side that wins the air battle gets to bomb) naturally encourage unrealistic min-maxing style gameplay.

C) Use your unique tech system to give the Soviets some advantages that encourage a mixed air strategy instead of just spamming fighters - you could make their dive bomb bombers harder to shoot down, or you could make them cheap enough to be worth using aggressively even if the losses are heavy. You would also have to do that for any other cases where the system is optimizing for less interesting play.

A is probably the easiest route to test since it doesn’t involve any other changes to the system, you just play out a few test games and redraw the air zone boundaries between each to see which results in the most interesting game dynamics. B is the next most promising, it also changes up just one element, but that element is one of your system’s core attributes. C is probably a development trap where you’d end up having to constantly fiddle around with a bunch of different bespoke elements you’ve come up with to patch different issues.

Originally the game had more air theaters. I'm in favor of more air theaters in theory. In practice the larger number of air theaters created a GUI problem. Everything about the air phase is new, including the GUI. Without a clear, simple GUI literally nothing else matters. I think I'm up to 24 engine change requests or something, many of which are intended to make that GUI simpler to understand and easier to use. Making each individual air theater larger, so that there would be fewer theaters overall, helped with the GUI. But there's still GUI progress to be made there. One could say that the perfect GUI would allow me to have a larger number of theaters without creating a GUI problem.

As to your point, I agree that if the Soviet player is the weaker participant in the air battle on the Eastern Front, he will buy air superiority aircraft only. Fighters, heavy fighters, etc. I also agree that this is ahistorical, and that in the real war the Soviets built dive bombers or ground attack aircraft, and threw them in to the fray regardless of the losses.

I don't know if you've played Domination 1914 No Man's Land. If you haven't, I'd recommend it. The reason I bring it up is because it makes a clearer distinction between offensively oriented units and defensively oriented units than you would normally see for a TripleA map. If you're on defense your unit mix should be trenches, conscripts, infantry, and heavy guns. Plus some air for your back-and-forth battles. If you're on offense your unit mix should be conscripts, infantry, field guns, cavalry, gass, and tanks. A player who's purely on defense will never buy a tank, a player who's purely on offense will never buy a heavy gun.

I believe the problem of min-maxing game play is baked into TripleA rules sets in general. While I know that my own game is a departure from the usual TripleA game in many ways, the underlying logic which causes min-maxing still remains. I've tried pushing back against this where I can, but I still feel like I'm just nibbling at the edges of the problem.

To take your option A: under my current system bombers are allocated first, then revealed to all players, and only then are fighters allocated. If for example the Soviet player were to send dive bombers to the southern portion of the German/Soviet front, the German player would be able to see this in air allocation, before he'd allocated his fighters. He'd then adjust by moving more air to the southern portion of the Soviet front.

You could say that I could make air allocation more blind. Okay, granted. I could. The weaker party could take advantage of this by getting some strafing runs through that would otherwise have been intercepted. But the stronger party could take advantage of this as well, and could use bomber-type aircraft to do enormous damage while the opponent's fighters were out of place to defend.

I hear what you're saying about Soviet dive bombers, and I realize my simulation of air combat is in some ways less rich than the actual history deserves. But the min maxing problem is a very tough nut to crack. My map was intended to do some things well: air combat, research, special units, a distinct flavor for each nation. It was not intended to solve the min max problem. Don't get me wrong: I'd very much welcome a solution!

As for idea B: I will grant that this could gently encourage the Soviets to buy dive bombers. Assume that a Soviet dive bombing mission will generally kill an anti-tank gun, doing 3 PUs of benefit for the USSR. If the fighter takes the interception hit instead of the dive bomber, that creates a 3 PU reward for the Soviets which would not otherwise have existed. But this would work both ways. Germany would do the same thing, allowing its dive bombers to get through as well. At some point the Soviet player starts to care more about protecting his own armies from the major losses inflicted by German dive bombers, as opposed to inflicting smaller losses on German armies with Soviet dive bombers.

Idea C--use the tech system--could be a way to encourage Soviet dive bombers. Historically Soviet ground attack aircraft were heavily armored. So I could justify giving Soviet dive bombers +1 to hitpoints if they research the appropriate tech. This doesn't necessarily solve the min maxing problem, but it could help somewhat.

One idea I'd come up with to push back against min maxing was to give nations unit-specific production centers. A nation gets one free dive bomber a turn, for example, from its dive bomber production center. I have not incorporated this idea into my map, however.

Vendetta

@KurtGodel7 Yeah, players are relentless optimizers, so min-maxing is always a tough problem to try and get rid of. The one thing you really want to prevent is players feeling too railroaded, where it feels like there is only one possible strategy you can use if you want to win. If the Soviets can only survive by playing a very specific build order, people are not going to have as much fun playing them as they do with another nation that gets more choices.

One more idea... have you tried testing what happens if you reverse the order on your air turns? What if everyone does a blind allocation on their fighters first, you reveal those, and then you commit your bombers once you've seen where the fighters are? That would allow players of a weak air force to do some misdirection (send your fighters one way but your bombers another) , or take advantage of vulnerabilities where your opponent didn't send enough fighters.

Or, since you've included heavy fighters alongside normal fighters, you could further differentiate their roles by having one type get allocated in the first round, and the other type in the second round - the fighter is an aggressive unit you commit blindly, the heavy fighter is a reactive unit you send after you've seen where some of the enemy air force is going.

You could also split the different bomber types between the two rounds - medium bombers and strategic bombers go in the first round so that strategic bombing raids are telegraphed and easier to respond to, but dive bombers and torpedo bombers get to sneak in on the second round and become harder to defend against.

Not sure which particular combination would be best, but shaking up which types go in which order seems like a good approach to investigate.

Schulz

I would recommend using better unit images.

It's better if each nation's non-naval units look different. Especially NWO/WaW infantry—it doesn't even look like infantry at all.

I'd go with something like this as German infantry:

Cernel

@KurtGodel7 said:

@RogerCooper said:

@KurtGodel7 How are you going to enforce "Sticky Hits"?

I've used AI to make a few tweaks to the engine. The modified version of the engine lives only on my computer. The modified engine allows me to do the following:

Sorry if this was already asked, but I guess you are planning to make a pull request to have all new features you developed to support this map added to the official TripleA release once you have the map ready. Right? Are you confident that the main developers will accept them? Maybe you should open feature requests (assigned to you) in advance (one for each new feature which your map needs) so to try to understand if there is the willingness to accept the feature additions? I mean, what happens if you finish the map and then the feature requests needed to have it playable are refused?

RogerCooper

I am sure everyone will want to review all features separately before adding them in. Even if they are working in your mod, there can be bad interactions with other features.

malmessi

@KurtGodel7 said:

I now believe that a good German opening involves taking Leningrad on G2

This is always a good opening for GER but it is impossible in many other games

malmessi

@KurtGodel7 said:

My planned solution to this (once my weekly usage limit resets).

Just for curiosity, once your weekly limit resets, how do you interact with AI ?
You just write what you posted here on the forum in teh propmpt and AI understand everything at your 1st attempt ?

malmessi

@KurtGodel7 said:

Normal Soviet behavior varies depending on the map. If you're playing Revised, the Soviets would maybe buy fighters for back-and-forth battles.

I play Total World War (hard) recently

malmessi

@Schulz said:

I would recommend using better unit images.

It's better if each nation's non-naval units look different

100% agree !!!
I started WorldAtWar 22 nations and it is a hell !
22 powers are too much and colour palette is not enough, some powers have almost the same colour so it is hard to ditinguish a chinese inf from a russian inf (and they are also close in some part of the map).

Also I found very hard to understand if a fighter is normal or advanced, when you purchase them it is almost clear but once in teh map together with oter units it is very very hard, the thin white lines I circled in blue are almost invisible !

Schulz

Even WaW's 14 nations are too much, let alone 22.

Turns take forever to complete, and that's not even the only issue with this map:

Units too small.
Insufficient unit spaces and unit skins.
Mega stacks.
Anticlimactic endings.
Immortal air units.

The only good thing about it is Japan being split in two, which prevents them from railroading into China and/or putting heavy pressure on Russia.

This isn't specific to WaW, though—it's more my general distaste for maps with too many nations or units. I feel like the standard rules stop working when a map gets too big.

KurtGodel7

@Cernel said:

@KurtGodel7 said:

@RogerCooper said:

@KurtGodel7 How are you going to enforce "Sticky Hits"?

I've used AI to make a few tweaks to the engine. The modified version of the engine lives only on my computer. The modified engine allows me to do the following:

Sorry if this was already asked, but I guess you are planning to make a pull request to have all new features you developed to support this map added to the official TripleA release once you have the map ready. Right? Are you confident that the main developers will accept them? Maybe you should open feature requests (assigned to you) in advance (one for each new feature which your map needs) so to try to understand if there is the willingness to accept the feature additions? I mean, what happens if you finish the map and then the feature requests needed to have it playable are refused?

You ask very good questions.

I've come up with what I believe are some good ideas. But do these good ideas gel into a good game? I'd like to think the answer to that is yes, but until I get closer to a finalized version of my game I won't know that for sure.

My thought process here is as follows.

I need to develop this into a fun map to play.
Then, I need to go to the main developers and show them the finished product. Or at least the somewhat finished product.
Hopefully by that point there will be some users willing to speak up in favor of this map. "I've played this, it's really fun, and it's worth adding these new features to standard TripleA."
Is it possible for the AI to create a developer-friendly version of the engine changes? If so, that's what I'd like the AI to do before I would ask for anything. But I don't know the specifics of what the developers would want to make the AI's engine changes developer-friendly.

KurtGodel7

@malmessi said:

@KurtGodel7 said:

My planned solution to this (once my weekly usage limit resets).

Just for curiosity, once your weekly limit resets, how do you interact with AI ?
You just write what you posted here on the forum in teh propmpt and AI understand everything at your 1st attempt ?

There are two different AIs I've been using for this. Cloud Claude and desktop Claude. Normally the interaction is as follows. I will have a conversation with cloud Claude. I will tell it objectives I have, or things I believe aren't working as intended. It will give me ideas on how to solve the problem. I will take the best of Claude's ideas, the best of my own ideas, and the best of ideas I've seen elsewhere. I will tell cloud Claude to do that. Cloud Claude can change the XML file directly. There have been nearly 100 versions of the XML file.

If I want changes to the engine itself, cloud Claude creates a change document. It's written in plain English and is several pages long. I will read this document over. Then I pass it along to desktop Claude so it can change the engine. Engine changes don't always go as planned. There can be GUI problems (most common), there can be bugs. Solving that requires a follow-on engine change. Desktop Claude has implemented roughly 25 engine change requests so far.

The things pushing me towards my weekly limit are engine change requests (which can take a long time) and creating the map itself. Tweaking the map a little here and there isn't bad; building the initial map was. For three straight weeks I've gotten very close to my weekly limit.

KurtGodel7

@Schulz said:

Even WaW's 14 nations are too much, let alone 22.

Turns take forever to complete, and that's not even the only issue with this map:

Units too small.

Insufficient unit spaces and unit skins.

Mega stacks.

Anticlimactic endings.

Immortal air units.

The only good thing about it is Japan being split in two, which prevents them from railroading into China and/or putting heavy pressure on Russia.

This isn't specific to WaW, though—it's more my general distaste for maps with too many nations or units. I feel like the standard rules stop working when a map gets too big.

You're making me feel happy about some of the good things of my map.

In my map, neither Japan nor the Soviet Union want to wage a huge early game war against each other. Partly this is due to the map shape, and partly due to priorities. Early game Japan wants to focus on China and the south Pacific. The Soviet Union wants to focus on Germany. But, the USSR wants to be strong enough in the east to prevent Japan from feeling too much temptation. It achieves this by building conscripts for Vladivostok.

Japan conquers China slowly. China resists by buying large numbers of conscripts. As China loses territory its ability to buy conscripts slowly diminishes. While I haven't taken this to the late game yet, I'm guessing China can hold out at least ten rounds. Maybe more.

I've achieved all this without dividing Japan. Japan is one nation, not three or four little Japans. The Soviet Union and China are able to defend using manpower points and conscripts.

In this map, I would want the United States to divide its efforts between the Pacific and Europe. What I don't want to happen is a WaW-like situation in which the U.S. always goes kill Japan first. I am willing to contemplate extreme measures to avoid that outcome. I might even divide the United States into two nations. One of which would be anti-Germany, the other anti-Japan. But I'd prefer to avoid taking things to that point.

Schulz

The Soviets and China having cheaper cannon fodder than Japan is a nice thing that prevents the Soviets and China from being too appealing targets. It's probably enough to diversify Japanese strategies.

I would suggest keeping the United States as one nation. Make Germany at least two times stronger than Japan, and make the Soviets doomed to fall if the US doesn't put significant pressure on Germany.

Personally, I can't stand seeing Japan outproduce Germany (this must be an extremely rare case, not a regular thing). I would make sure Japan is still weaker than Germany even if they take China and the Far Eastern Soviets.