How does AI fare in games vs mods (Dababes, RA)

Please post here for questions and discussion about scenario design, art and sound modding and the game editor for WITP Admiral's Edition.

Moderators: wdolson, MOD_War-in-the-Pacific-Admirals-Edition

medicff
Posts: 710
Joined: Fri Sep 10, 2004 10:53 pm
Location: WPB, Florida

How does AI fare in games vs mods (Dababes, RA)

Post by medicff »

Hi all,

Looking at such mods as Dababes and RA and wondering how an AI game would fare?

Probably Japan computer as read better AI.

Does it handle extended map, new locations, changes to economy, new units, new engineers setups?

BTW I usually jump sides to make minor adjustments currently to economy such as nates to oscars etc in stock AI already but not very much effort needed.

Thanks for all advice.

Looking for AI game while getting back into AE while taking full class load, but when finished hoping to have time for PBEM.

Pat
User avatar
JWE
Posts: 5039
Joined: Tue Jul 19, 2005 5:02 pm

RE: How does AI fare in games vs mods (Dababes, RA)

Post by JWE »

ORIGINAL: medicff
Hi all,

Looking at such mods as Dababes and RA and wondering how an AI game would fare?
Honestly, it sucks. If you must play the AI try BabesLite. That's what it's there for.
Probably Japan computer as read better AI.
Only because we didn't tweak the Japan side as much as the Allied. It still sucks, but not as bad. We wanted to make it work, but see the next answer.
Does it handle extended map, new locations, changes to economy, new units, new engineers setups?
Nope. Not unless some whacko is willing to port over almost 90 AI files (7 to 9 months of work).
BTW I usually jump sides to make minor adjustments currently to economy such as nates to oscars etc in stock AI already but not very much effort needed.
A good way to do it.
Thanks for all advice.

Looking for AI game while getting back into AE while taking full class load, but when finished hoping to have time for PBEM.

Pat
Ciao. John; but don't give up hope.
pharmy
Posts: 271
Joined: Sat Apr 03, 2010 2:51 pm
Location: Bangkok/Budapest

RE: How does AI fare in games vs mods (Dababes, RA)

Post by pharmy »

Damn,that's a bummer but its useful info, I was just about to try RA 4.3 playing as Allies vs Jap AI for the first time. I'll guess I'll stick to the old map and RA 3.4.2 and stock scenarios.
rockmedic109
Posts: 2440
Joined: Tue May 17, 2005 11:02 am
Location: Citrus Heights, CA

RE: How does AI fare in games vs mods (Dababes, RA)

Post by rockmedic109 »

Actually I think da babes can be enjoyable. Yes, you are going to win. But with the allies, you should. Avoiding a non-historical approach helps the AI tremendously. I took an AI babes game into 43. The only problem was two attempts by the AI to send carriers past the guns at Singapore: Nasty Brits moved all their navigation aids inland so KB thought ther were farther out than they were...unsporting. But I think this would've occured in a vanilla game. And I think it is dependant on what AI set is "drawn" when the game starts.
dwg
Posts: 319
Joined: Tue Jan 22, 2008 1:35 am

RE: How does AI fare in games vs mods (Dababes, RA)

Post by dwg »

ORIGINAL: JWE
ORIGINAL: medicff
Does it handle extended map, new locations, changes to economy, new units, new engineers setups?
Nope. Not unless some whacko is willing to port over almost 90 AI files (7 to 9 months of work).
[/quote]

I'm not sure it would be so bad with vanilla AI, most locations haven't changed, and from what I can see mostly the AI works through the location IDs and HQ structures. It won't be efficient, and it will have blind-spots where it has no knowledge of new locations and units, but there should be some functionality there.

For that matter my own scenario is a hacked Da Babes and the AI is functioning without me having touched it, no matter I'm using extended map and have a host of new units and locations. I haven't played deeply enough into it as I'm still working on the set-up, but I must have played the first 5-10 days 20 times by now, and I seem to be seeing different behaviour from the AI in different run-throughs (particularly how long the Kido Butai hangs around off Pearl), which indicates it is choosing and executing from the range of available strategies, just as it is meant to.
User avatar
JWE
Posts: 5039
Joined: Tue Jul 19, 2005 5:02 pm

RE: How does AI fare in games vs mods (Dababes, RA)

Post by JWE »

It’s really not bad when playing against a Japanese computer. And thank you all who play it and have nice things to say about it. We listen. Believe me.

But there are a lot of BigBabes things that the AI won’t recognize, and a lot of BigBabes things that were tweaked to specifically remove certain AI “advantages”.

The “moves” the AI makes (what it does) are still governed by the scripting, and will happen in an acceptable way, just like vanilla. But the “way” the AI functions (how it does it) is governed by the underlying implementation code that is found in literally every subroutine of the executable (hundreds of them). Many (if not most) of the BigBabes enhancements were created long after the implementation code, so the AI doesn’t even know they exist.

Just for example, the AI has no clue that the “Shore Party” switch exists, so that whole section of the TOEs are just empty slots as far as the AI is concerned. The AI does not recognize the differential between and among the various engineering units, so BigBabes engineers will not function correctly.

BigBabes gets rid of most base force units, tagged as Base Force, and replaces them with HQs and airbase support units, tagged as Engineers. The AI uses BF units in accord with an algorithm that lets it populate bases with any and all BF units and then grows them in accord with the base size. This is good for the airgroup movement algorithm. But is bad because BF platoons, with AvSup of 2 will quickly grow to AvSup of 90, or even 250, whenever it is required. Every base an LAX.

AI handles all hex combat as a “blop”. It is not programmed to handle phased combat. It plays stack-to-stack. BigBabes pulls out arty from some units so arty can be used as it was without rendering “other” units vulnerable to counter-battery. BigBabes has much more robust and realistic combat system, but it must be implemented manually (H2H or PBEM); AI can’t do it, don’t know how.

AI does not recognize “barge” or “LCVP” movement. If the situation is perfectly set up for a coastal barge movement and assault, the AI will preferentially teleport a passenger ship (xAP) to do the lift. AI will immediately and always resurrect any “buy from supply” class there is. It doesn’t matter how many or few Midgets are physically in the database slots, they will always be resurrected and amount to thousands, if that’s what it takes. BigBabes fiddles with that, so manual play is reasonable, but assigning dates and ports for these vessels precludes their use by the AI. It’s an all-or-nothing kind of thing.

AI does some special “everyday” TFs for a port with certain kinds of Classes in the port. Everybody bitches about the uselessness of certain small Classes, but those are the very ones the AI uses unbeknownst to the inattentive player. BigBabes maximizes (redefines) the ships available for these unsung, but highly important, missions.

And this is just a surface gloss on how BigBabes is incompatible with the AI implementation code. It would be good to have a decent Allied AI too, but there’s too much in BigBabes to ever be righteous with the AI implementation code. Best we can hope for is a decent set of movement option scripts for both sides. BigBabes will always remain a manual (H2H or PBEM) learning tool.
dwg
Posts: 319
Joined: Tue Jan 22, 2008 1:35 am

RE: How does AI fare in games vs mods (Dababes, RA)

Post by dwg »

AI handles all hex combat as a “blop”. It is not programmed to handle phased combat. It plays stack-to-stack. BigBabes pulls out arty from some units so arty can be used as it was without rendering “other” units vulnerable to counter-battery. BigBabes has much more robust and realistic combat system, but it must be implemented manually (H2H or PBEM); AI can’t do it, don’t know how.

I take your general points, but can you elaborate on the one above? Are we talking about bombardment attacks, or something different? I don't think it's a change that's necessarily apparent to players - Shore Parties and the like are easy enough to understand the role changes, they're pretty much forced onto you, but this one may have been subtle enough to slip by most people.

I suspect a lot of us who aren't able to commit to PBEM or H2H play are willing to accept the limitations on playing Big Babes against the AI. It may not be ideal, but it's the only game in town if you want the extra levels of detail that Big Babes brought us.
DD696
Posts: 976
Joined: Fri Jul 09, 2004 12:57 pm
Location: near Savannah, Ga

RE: How does AI fare in games vs mods (Dababes, RA)

Post by DD696 »

What I have done (my latest attempt) to try and make the best AI game for me playing as the Allies, is to use BigBabes as the starting point, then make the modifications I choose to make, and then to replace the Japanese land units with the land units from BabesLite. This gives the AI the units it is expecting to see and hopefully will make for a better game. Not that far enough long in it to comment on how it works, but it should work "better" against the AI.

But, as I have found throughout the years, the name of the game is "Restart in the Pacific". Someday I hope to get a version that I find satisfactory.
USMC: 1970-1977. A United States Marine.
We don't take kindly to idjits.
User avatar
JWE
Posts: 5039
Joined: Tue Jul 19, 2005 5:02 pm

RE: How does AI fare in games vs mods (Dababes, RA)

Post by JWE »

ORIGINAL: dwg
I take your general points, but can you elaborate on the one above? Are we talking about bombardment attacks, or something different? I don't think it's a change that's necessarily apparent to players - Shore Parties and the like are easy enough to understand the role changes, they're pretty much forced onto you, but this one may have been subtle enough to slip by most people.

I suspect a lot of us who aren't able to commit to PBEM or H2H play are willing to accept the limitations on playing Big Babes against the AI. It may not be ideal, but it's the only game in town if you want the extra levels of detail that Big Babes brought us.
Damn. I can't answer that question, in so far as I understand it. Would you be so kind as to send me a pm, with a bit more description as to what you are talking about. [8D]

Ciao. John
Mac Linehan
Posts: 1518
Joined: Sun Dec 19, 2004 9:08 pm
Location: Denver Colorado

RE: How does AI fare in games vs mods (Dababes, RA)

Post by Mac Linehan »

JWE -

Good information - a glimpse under the hood.

I am currently alternating between two Babes games - DaIronBabes 30v2 (Japanese AI) and DBB Scn28C v10 (H2H). I believe that my focus will change to DBB Scn28C; as I very much enjoy playing both sides. Over time, it has become apparent, that AE H2H can work - there is that which a player controls; but so much more (i.e. Combat Resolution!) that is out of a Player's hands. And when you throw in "advanced weather" - lots of bad weather, and thus even more uncertainty, it is really quite enjoyable. I am gradually putting together notes on how to balance solitaire play - giving the Japanese better radio intell, for instance. I also attempt to stick with historical goals; and try to always apply the "rule of reason" - would a historical commander / leadership do this"?

dwg -

please know that the AI does a good job, my DaIronBabes game has been very enjoyable, the Japanese AI has given me a hard time. Obviously even the most thoroughly scripted AI - equating to many months of work by Andy Mac or the Henderson Group (and the AE AI does very well!) can not match the flexibility and planning in a human player. In my modded IronBabes game, I gave the Japanese more of just about everything - in large quantities - to allow room for error and losses. I have not yet switched sides to tweak; but may do so just to help the AI out.

So - there are many ways to work around various limitations. I choose Babes because of what you have already mentioned - all the changes, tweaks and different approaches that makes Babes AE a new game. Please stick with it, it will be well worth the effort.

Just some thoughts -

Mac
LAV-25 2147
User avatar
JWE
Posts: 5039
Joined: Tue Jul 19, 2005 5:02 pm

RE: How does AI fare in games vs mods (Dababes, RA)

Post by JWE »

ORIGINAL: dwg
AI handles all hex combat as a “blop”. It is not programmed to handle phased combat. It plays stack-to-stack. BigBabes pulls out arty from some units so arty can be used as it was without rendering “other” units vulnerable to counter-battery. BigBabes has much more robust and realistic combat system, but it must be implemented manually (H2H or PBEM); AI can’t do it, don’t know how.
I take your general points, but can you elaborate on the one above? Are we talking about bombardment attacks, or something different? I don't think it's a change that's necessarily apparent to players - Shore Parties and the like are easy enough to understand the role changes, they're pretty much forced onto you, but this one may have been subtle enough to slip by most people.

I suspect a lot of us who aren't able to commit to PBEM or H2H play are willing to accept the limitations on playing Big Babes against the AI. It may not be ideal, but it's the only game in town if you want the extra levels of detail that Big Babes brought us.
Finally figured out what you were asking; must be getting old.

Phased combat includes bombardment, but it also includes the other stuff. Couple caveats, we mostly configure for the Pacific part of War in the Pacific. Don't pay much attention to collision of mass armys on a land mass like China. Engine is not at it's best for that scale. Also, we've been tweaking heavily for our small map scenarios (DEI, PI, Guad, OP FS). The BigBabes TO&Es are more "stock-like" than the stuff for our CPX games. Having said that:

Our division is an HQ unit. It has 3 Rgts (duh) and the arty split out as Bns; no recombine. Japan is the same except where it has a Cav or Recce unit; that is also split out as a separate Bn. This works well for Pacific island, DEI, PI type operations. In a game stack, of division scale, you might have 1 Rgt set to attack, 1 Rgt set to pursue, 1 Rgt resting, 2x 105 and 1x 155 Bns bombarding, and the last 105 Bn either set to pursue or rest.

Bombardment comes first, and counter-battery is limited; it doesn't whack the whole stack and cause those wierd imbalances. Attack comes next, but the Rgts still have their TOE regimental arty, so they get the firepower push. Resolve the attack, see wha'hoppen, if ya win, a kinda-sorta reserve (the pursuit) unit takes over, and the process repeats (think self-sharpening chain saw) [:D]

AI isn't programmed to think in these terms. If ya think on it, the engine works according to GG's original concept. In that, divisions are broken into 3 equal pieces (A, B, C) each with a proportional allocation of DivArty. So bombardment is by whole divisions or regiments ('cause that's where the guns are), not by arty units. So counter-battery effects whole divisions or regiments. The AI is not a separate implementation routine. The different bits of it are embedded in every single one of the hundreds of sub-routines and algorithms. That's what makes code tweaking so heartbreaking; try to spray for a couple dandelions and you end up killing the azaleas too.

Think so far as BigBabes is concerned the AI, doing stack attacks, is kinda ok. I mean, it has to be ok, 'cause that's all it knows. But whatever manual side you play, you might want to try phasing.
User avatar
Sardaukar
Posts: 12731
Joined: Wed Nov 28, 2001 10:00 am
Location: Finland/Israel

RE: How does AI fare in games vs mods (Dababes, RA)

Post by Sardaukar »

I am getting quite decent game as Allies vs. IJ AI, DaBigBabes A. I am limiting myself to mostly defensive fights until 1/1/1943 and go to offensive only after that. AI gives quite decent game for first year, if you limit yourself to playing mostly historically first. But, for example, I reinforced Port Moresby quite heavily and was really surprised that I almost lost it to AI. Was very close thing and I have to rush reinforcements in quickly.

AI can still give you some surprises.
"To meaningless French Idealism, Liberty, Fraternity and Equality...we answer with German Realism, Infantry, Cavalry and Artillery" -Prince von Bülov, 1870-

Image
User avatar
moonraker65
Posts: 565
Joined: Wed Jul 14, 2004 3:11 pm
Location: Swindon,Wilts. UK

RE: How does AI fare in games vs mods (Dababes, RA)

Post by moonraker65 »

I've just started a BigBabes RA 4.3 game against the Japanese AI and you certainly notice a difference in the speed at which he can make his moves. Not nearly as quick as a stock game. Although as mentioned above he can still throw some nasty surprises your way as I lost all but one of the ships evacuated from Hong Kong near Borneo after a surprise encounter with one of his Surface TF's [8|]
intel i9 13900k 128 GB RAM, RTX 4070 ti GFx card
User avatar
JWE
Posts: 5039
Joined: Tue Jul 19, 2005 5:02 pm

RE: How does AI fare in games vs mods (Dababes, RA)

Post by JWE »

@ Sardaukar
@ moonraker

Yes, the AI can give you a host of surprises. It's important to understand that there's 2 parts to the AI, though; "what" it does and "how" it does it.

The "how" is governed by the code and affects the combat methodology and its results (among many other things).

The "what" is governed by the scripts. Those are pretty much defined by an apocalyptically evil, blue-painted, Caledonian barbarian, who wears skirts with nothing underneath for gosh sakes. Grendel has nothing on this guy.

Always wondered what kind of people could produce a prince of darkness like Andy McPhie and a queen of love like Christine McVie, till I got to the skirt with no underwear part [:D]

(Andy, if this is offensive in any way, I will stop at once. Just say so. No offense is meant, it's all in jest. Sometimes humor pushes limits. If so, profound apologies. Have too much respect for you to do otherwise. J)
User avatar
Don Bowen
Posts: 5190
Joined: Thu Jul 13, 2000 8:00 am
Location: Georgetown, Texas, USA

RE: How does AI fare in games vs mods (Dababes, RA)

Post by Don Bowen »

who wears skirts with nothing underneath for gosh sakes

Watch that, you lowlander.



Image
Attachments
IMG_4387.jpg
IMG_4387.jpg (206.78 KiB) Viewed 692 times
User avatar
JWE
Posts: 5039
Joined: Tue Jul 19, 2005 5:02 pm

RE: How does AI fare in games vs mods (Dababes, RA)

Post by JWE »

ORIGINAL: Don Bowen
Watch that, you lowlander.
Ahh, but what's behind the Sporan, Brother Bowen. Are ya willing to lift that little kiltie up and show us what's beneath??

Aaagrrhh Fruit of the Loom ??? [8D] Dude !! [:D]
User avatar
Don Bowen
Posts: 5190
Joined: Thu Jul 13, 2000 8:00 am
Location: Georgetown, Texas, USA

RE: How does AI fare in games vs mods (Dababes, RA)

Post by Don Bowen »

ORIGINAL: JWE

ORIGINAL: Don Bowen
Watch that, you lowlander.
Ahh, but what's behind the Sporan, Brother Bowen. Are ya willing to lift that little kiltie up and show us what's beneath??

Aaagrrhh Fruit of the Loom ??? [8D] Dude !! [:D]

Nae laddie, my wife is the only one allowed up under the kilt.
User avatar
Falken
Posts: 289
Joined: Wed Aug 08, 2007 4:08 pm
Location: ON, Canada

RE: How does AI fare in games vs mods (Dababes, RA)

Post by Falken »

Mac,

Is your version of DaIronBabes-B available anywhere? Due to work schedule, I have to stick to AI (as I would seriously upset my opponent in not producing turns every day) and I'm still debating between DBB, DIB, and RA 4.3 for solo play, but I am interested in your comments on your version of DIB.

I'm having a hard time committing to a version for long term as i'm still undecided between the 3, but I do like the fact that RA has reduced supplies on the West Coast.
User avatar
drw61
Posts: 842
Joined: Wed Jun 30, 2004 12:58 pm
Location: South Carolina

RE: How does AI fare in games vs mods (Dababes, RA)

Post by drw61 »

ORIGINAL: Falken

Mac,

Is your version of DaIronBabes-B available anywhere? Due to work schedule, I have to stick to AI (as I would seriously upset my opponent in not producing turns every day) and I'm still debating between DBB, DIB, and RA 4.3 for solo play, but I am interested in your comments on your version of DIB.

I'm having a hard time committing to a version for long term as i'm still undecided between the 3, but I do like the fact that RA has reduced supplies on the West Coast.

Here are the links

DaBigBabes - Scenario 2 Hakko Ichiu
https://sites.google.com/site/dababeswitpae/ (DaIronBabes)
or
tm.asp?m=3075475

DaBigBabes - Scenario 10 Ironman
tm.asp?m=3062873

Reluctant Admiral - by John 3rd
https://sites.google.com/site/reluctantadmiral/
or
tm.asp?m=3071237
Mac Linehan
Posts: 1518
Joined: Sun Dec 19, 2004 9:08 pm
Location: Denver Colorado

RE: How does AI fare in games vs mods (Dababes, RA)

Post by Mac Linehan »

Falken -

drw61 has kindly linked the IronBabes website. He is also responsible for converting stock Ironman to DaIronBabes - and did an outstanding job; it is good.

As for modding - I am at a very basic level; and still make mistakes - but that doesn't affect my enjoyment of the game at all. It is a constant learning process.

In my personal IronBabes Mod, I added more Japanese aircraft to the AC pools; gave the Home Islands more Oil, Fuel and Supply. Supply was also added to key bases - Truk and Kwaj, for example.

The AI (to the best of my knowledge) does not do conversions; so I added about 60 or so Patrol Boats - divided between the mighty To'sus and his bigger brothers - the Ansyu's and Kiso's.

I also gave the IJN additional Destroyers - repeats of the last three modern classes.

Please know that all of this is strictly non historical, and way beyond real life Japanese Industrial capability. I wanted to allow room for wastage in aircraft and ships while under AI control - and to enhance the AI's staying power.

All of this is just my amateur effort to make it a tougher game - in addition to Andy Mac's devilish AI and additional Japanese units. I am not deriding my own experience or effort; there is so much to learn in AE - perhaps one of the most detailed and complicated games on the market.

Very recently I received excellent guidance on the Babes shore party concept; greatly enhancing my understanding of how it is was really supposed to work. Thus I spent most of last night reworking my personal mod to align with the intended concept. And this is only one small part of the puzzle; I am sure that this process will be ongoing.

I would like to emphasize that the Babes mods are not "what if" scenarios - quite the opposite, and are meant for players who want max realism.

And: If I can (at least) begin to grasp how Babes works, any one can.

Go for it, Sir!

Mac
LAV-25 2147
Post Reply

Return to “Scenario Design and Modding”