Matrix Games Forums

I’ve read a lot of threads here to learn the game and one time I heard this:

A player said that no matter how many Betty bombers the Japanese lost, it seemed like they they still had hordes, like BattleStar Galactica raiders. Just keep coming.

I’ve destroyed 1470 Japanese a/c by March 7th, 1942, in all theatres. Is there a realistic model of losses versus replacements in the game? Is there magical replacement of aircraft and crews not talked about? Have I made an impact? I want to think that every Betty and especially Nell (hate Nell’s, ship-sinking dragons, hate’em) makes an impact on the future.

Am I naive to expect that?

Which scenario are you playing?
I expect there is a big difference between scenario 1 and Ironman Nasty.

Going by the Japanese side's numbers in my last campaign in scenario one the AI had lost well over 5,000 G4Ms by January 1944. During the war total production was only about 2,400. So yes, the AI gets a lot of free planes. I seem to recall Sarduakar stating they never run out.

I believe that's programmed in to balance out the fact that the AI works off scripts and it makes stupid choices that a human player wouldn't. You would know not to keep sending your bomber at a well capped base, or at least do things to counter it, but the AI can't realy do that. It will just mindlessly throw its planes at a base until the script tells it to stop.

Consider it a great place to give your pilots on the job training.

Although the Japanese AI never runs short on planes, it will eventually run low on experienced pilots. Then the one-sided aerial butchery really skyrockets.

I’m honoured to be in a position to comment on a Gary Grigsby game. Best games ever, the whole family. Love them.

I have a question.

Can the AI actually be improved, with all the recent advances in AI?
Not criticizing here, AI works for me, I’m having fun as we speak. But I don’t write code, I enjoy the results of code. Is there a way to upgrade the AI? Could you make a human-like opponent in this game?

Asking for a friend.

Bella

Bella wrote: Sun Mar 10, 2024 1:34 am I’m honoured to be in a position to comment on a Gary Grigsby game. Best games ever, the whole family. Love them.

I have a question.

Can the AI actually be improved, with all the recent advances in AI?
Not criticizing here, AI works for me, I’m having fun as we speak. But I don’t write code, I enjoy the results of code. Is there a way to upgrade the AI? Could you make a human-like opponent in this game?

Asking for a friend.

Bella

You could write your own scripts, you could have them stop after just a short period of time so they only make one or two attacks and then go somewhere else with ships magically being transported to where they are needed. That could make things difficult. I mean, attacks in the South Pacific and then just days later the same ships are in the Gulf of Alaska? Then they go to the Bay of Bengal just days after they capture bases in Alaska?

As far as the aircraft losses go, adjust the losses down by about 33% due to FOW. (Fog Of War)

Platoonist wrote: Sun Mar 10, 2024 12:30 am Going by the Japanese side's numbers in my last campaign in scenario one the AI had lost well over 5,000 G4Ms by January 1944. During the war total production was only about 2,400. So yes, the AI gets a lot of free planes. I seem to recall Sarduakar stating they never run out.

So, I don't play the allies very often, allies are generally AI controlled for me. I can state that I have never seen the AI magic up planes, they do get built. I've got enough game data to support that. In other threads where this has come up, and the planes magically appeared, I was NOT able to re-produce that. My thought is that the player had a memory leak in a turn or two.

So, how would the AI produce 5000 G4M's in a game? Because unlike history, they never stop building them and the factories got expanded. There are 7 factories setup in Scen 1 to produce G4M planes with a potential total of 46 planes/month. Now I have seen that factories do expand, I just don't know how the AI does it. I 'm 99% sure it is not scripted. Random increase? Not sure. But, some of the AI factories do grow (assuming it is not a memory leak on my side ... lotta turns to get to '45). The real question I cannot answer is can the AI factories build when they don't actually have the resources? I cannot answer this because all of my late game saves are for allied AI, and the allies never run out of resources.

If you have a late game save of an IJ AI game, I would love to look at it. A lot of things on the AI side I would be able investigate.

So, did some testing. Ran Scen 1 in continuous mode for a while. Discovered, that it appears that the first of each month, the AI is able to increase factory sizes. These increased factory sizes are definitely using resources, so I would conjecture at this point that if they run out of resources, the aircraft builds would stop. What I still need to check is if these factory builds consumed the required resources or not. I will update more on this ...

Playing as Allies, I have set it up so that I could look at the Japanese side occasionally. The AI would not build enough engines for the aircraft that it could build. When I changed engine production, the AI would not get enough supplies to that base to actually do the expansion.

All very interesting. Thank you.

What about the AI in naval combat, in your opinions? Does it do well? I played one short game (Coral Sea) before playing the long game, as a tutorial on naval dynamics, and results were so like historical it was scary. Enterprise and Yorktown damaged, Yorktown heavily and Shikoku with 77 Flotation damage, air group destroyed and Zuikaku’s air group mauled. So, I assumed the game had that figured out.

Bella wrote: Sun Mar 10, 2024 1:07 pm All very interesting. Thank you.

What about the AI in naval combat, in your opinions? Does it do well? I played one short game (Coral Sea) before playing the long game, as a tutorial on naval dynamics, and results were so like historical it was scary. Enterprise and Yorktown damaged, Yorktown heavily and Shikoku with 77 Flotation damage, air group destroyed and Zuikaku’s air group mauled. So, I assumed the game had that figured out.

AI is a fine naval tactical opponent. What it can't do is strategic naval, meaning grouping its forces into a Deathstar that most players do. Particularly as the allies, the AI has no solution to the DeathStar once that allies create it. On the IJ side, the KB can't sail everywhere due to fuel limitations. But the Deathstar has no limits once formed.

To be fair, even IJ players struggle against the allied player Deathstar, so expecting the AI to have answer to this is kinda unreasonable ....

Bella wrote: Sun Mar 10, 2024 1:07 pm What about the AI in naval combat, in your opinions? Does it do well? I played one short game (Coral Sea) before playing the long game, as a tutorial on naval dynamics, and results were so like historical it was scary. Enterprise and Yorktown damaged, Yorktown heavily and Shikoku with 77 Flotation damage, air group destroyed and Zuikaku’s air group mauled. So, I assumed the game had that figured out.

I've noticed in the campaign game that Japanese AI surfaces force will often blunder into an area covered by Marine dive bombers. Without air cover they are usually either heavy damaged or sunk. This happens over and over again in the same areas often enough that I wonder if the AI has any way of remembering that a certain base is a threat to ships with no air cover. Especially if it's an airfield that the Allies didn't historically hold and build up like Ambon, Koepang or Dili,

PaxMondo wrote: Sun Mar 10, 2024 3:45 pm AI is a fine naval tactical opponent. What it can't do is strategic naval, meaning grouping its forces into a Deathstar that most players do. Particularly as the allies, the AI has no solution to the DeathStar once that allies create it. On the IJ side, the KB can't sail everywhere due to fuel limitations. But the Deathstar has no limits once formed.

To be fair, even IJ players struggle against the allied player Deathstar, so expecting the AI to have answer to this is kinda unreasonable ....

Off topic and just curious, but what counters are there for Deathstars that other players make? I was getting slapped around pretty hard by one in a Guadalcanal PBEM game that I as Allies couldn't really figure out how to counter until it was too late. Eventually I figured out that port bombing by 4ELB was pretty effective to pushing him back.

And to be fair, by 1943/1944 the Allies IRL were forming Deathstars. Mahan's strategic theories were proven correct.

Arkham wrote: Sun Mar 10, 2024 5:38 pm Off topic and just curious, but what counters are there for Deathstars that other players make?

There's this small thermal exhaust port below the main port about the size of a womp rat.

Oops. Wrong game.

Nami Koshino wrote: Sun Mar 10, 2024 6:37 pm

Arkham wrote: Sun Mar 10, 2024 5:38 pm Off topic and just curious, but what counters are there for Deathstars that other players make?

There's this small thermal exhaust port below the main port about the size of a womp rat.

Oops. Wrong game.

....GP

Bo Rearguard wrote: Sun Mar 10, 2024 4:40 pm

Bella wrote: Sun Mar 10, 2024 1:07 pm What about the AI in naval combat, in your opinions? Does it do well? I played one short game (Coral Sea) before playing the long game, as a tutorial on naval dynamics, and results were so like historical it was scary. Enterprise and Yorktown damaged, Yorktown heavily and Shikoku with 77 Flotation damage, air group destroyed and Zuikaku’s air group mauled. So, I assumed the game had that figured out.

I've noticed in the campaign game that Japanese AI surfaces force will often blunder into an area covered by Marine dive bombers. Without air cover they are usually either heavy damaged or sunk. This happens over and over again in the same areas often enough that I wonder if the AI has any way of remembering that a certain base is a threat to ships with no air cover. Especially if it's an airfield that the Allies didn't historically hold and build up like Ambon, Koepang or Dili,

It's about recon more than anything. If the AI has recon in the area to know about allied air activity, then yes. If it doesn't, then after a few turns the awareness dies away. Players can be guilty of the same behavior ...

Bella wrote: Sun Mar 10, 2024 1:34 am I’m honoured to be in a position to comment on a Gary Grigsby game. Best games ever, the whole family. Love them.

I have a question.

Can the AI actually be improved, with all the recent advances in AI?
Not criticizing here, AI works for me, I’m having fun as we speak. But I don’t write code, I enjoy the results of code. Is there a way to upgrade the AI? Could you make a human-like opponent in this game?

Asking for a friend.

Bella

Well a few years ago AndyMac opined that he wished some others would take up the mantle of AI Scripting for AE. To RJ's point above, even if you had a handle on how to script (which to be fair isn't rocket science) and could do it well (a bigger concern imo) I am not sure how you deal with making the scripts handle an evolving game and expansion and contraction of the Japanese and/or Allies. Is there a way to make it "revisit" locations and strengths, etc on a regular basis and then adapt to that? It makes my head hurt just contemplating what it would take to make it happen.
So, here we are in 2024 with AndyMac's series of scripts, which we are damn lucky to have. He's a busy guy in RL and I am really grateful he agreed to take on this update. Who knows? Perhaps he will find a wellspring of motivation to keep working on the scripts.
Regardless, I am really looking forward to the coming months and the new version!

As said before, Japan's problems (both AI and human player) is not planes but pilots.

Human can adjust production (depending on oil and resources and such) and AI never runs out of planes. It does run out of good pilots though, which means that as Allies, you should by end of 1942 have killed all the better ones.

Human can adjust pilot training too, even as Allies. Restricted squadrons On US West Coast are perfect to get decent replacements by using them as training units. One can produce surprising amounts of exp 50(+) air skill 70, def 50+ to 70 fighter pilots on West Coast. Brits and Commonwealth are way more difficult, since you need about every unit you have in frontline.

USN Navy pilots are more difficult, but then, you can train them with floatplanes (Kingfishers and such). And when those CVEs with reserve squadrons come, your problems are basically over.