single player/AI

[Shark7]

Honestly, I'm not above helping the AI either by setting the game to head to head for a turn or two and upgrading its airgroups to something better. Just makes the game a bit more challenging. Is that gamey? Cheating for the AI? [:'(]

[Feltan]

ORIGINAL: Bullwinkle58

ORIGINAL: Feltan

I am in mid-November '42 as the Allies against the Japanese AI.

The AI is a wonderful training tool for future PBEM, and it is a lot of fun in its own right. As others have stated, it is an order of magnitude more sophisticated than the original WITP AI.

The only issues I have seen with the AI is that the digital Japanese commander doesn't know when to call it quits. I have followed a fairly standard Allied strategy, and the Japanese AI has focused laser-like on Canton Is and especially Port Blair (which I strongly reinforced before knowing the AI would be interested in it).

Over the last few game-months, I'd estimate the Japanese have lost three+ divisions with all transports on the beaches of Port Blair -- it would really be a grand artifical reef for decades to come!

Other than that, the AI has done a good job of offering a challenge and attacking/defending in a reasonable sort of way.

Regards,
Feltan

In my first game I got these scripts re Baker and Canton, plus Midway Island raids. In this game, I have not. OTOH, I've gotten action around Oz I didn't see in the first run through, as well as very, very aggressive IJN carrier usage around the Solomons.

Port Blair has been in contention both times. The AI REALLY wants it, and I can see why, given its location and huge airfield. In both games I garrissoned it well from the get-go, but it's costly to get supplies in, as TFs have to take an extra day to negotiate the reefs, and the LBA from Burma and Thailand is murderous. I've used the RN carriers when I really had to get supplies in, but Hermes is in drydock for 155 days right now as a result. The IJA has two full divisions and a Guard Force ashore, and I have one British Division and about seven support units behind Forts 5, headed for 6, and 8000 supply. If it falls it will be a big hit to the region. OTOH, 8000 supply isn't going to get the job done, and I've lost a lot of xAKs keeping the lights on.

I know a lot of PBEM players sneer at the AI, but it gives a good game which forces hard choices.

Mr. Moose,

I pulled everything I could out of Singapore (II Indian Corps, ground combat units and attachments) and sent it there, and stockpiled around a 100K supply. Right now it is as close to invulnerable as any place on the map.

One technique I have had some success with is using AKL's in single ship task force -- forming a bucket brigade from both Columbo and Calcutta -- dispatching a single ship from each port every turn or other turn. I've lost maybe a dozen or two over the last game year, mostly to surface action. For some reason the AI isn't keen on sending bomber formations against a lone AKL.

Regards,
Feltan

[Bullwinkle58]

ORIGINAL: Feltan

One technique I have had some success with is using AKL's in single ship task force -- forming a bucket brigade from both Columbo and Calcutta -- dispatching a single ship from each port every turn or other turn. I've lost maybe a dozen or two over the last game year, mostly to surface action. For some reason the AI isn't keen on sending bomber formations against a lone AKL.

Regards,
Feltan

I've tried this only from Colombo, usually with xAKs, and have gotten a few through. Not once a week though. I moved too many of my small merchants to Perth at the start, hoping to use them for coastwise resupply of Oz, but the new logistics model since I played the first time has made keeping a CAP up over northern Oz impossible for me. I've taken to hitchhiking the xAKLs into CT convoys with high-endurance ships to provide underway fueling, and hauling mostly fuel with them.

A number of the merchants I did leave at Colombo have fallen to subs, many east of Tricomalee. In late Dec 1942 I've only managed to sink four IJN non-mini subs, although I have done penetrating damage several dozen times. Columbo is down to only a handful of merchants, and I need to send some from CT and do the best I can on fuel. Fuel needs are eating me alive, as any Allied player is well aware will happen in 1942.

[Thayne]

As the author of that AAR called "Thayne Reports", I can say that I am finding the AI to be quite entertaining so far.

And it has some advantages over a PBEM opponent - such as the freedom to adopt personal home rules that . . . well, that make for a better story, I think.

Still, I like the idea of having somebody in the wings to look over the AI's shoulder from time to time. I may take advantage of that option. I keep worrying that the AI is going to goof up in ways that will ruin the story.

Thayne

[Banzan]

ORIGINAL: Bullwinkle58

A number of the merchants I did leave at Colombo have fallen to subs, many east of Tricomalee. In late Dec 1942 I've only managed to sink four IJN non-mini subs, although I have done penetrating damage several dozen times.

I had a lot of trouble with anti-submarine-warfare, till i started to build "kill groups". 4 ships with ASW weapons (even if only 1 or 2, but choose ships with enough ammo) and put a commander with high (above 60) naval rating in each of the ships. Set reacting range high (depend a bit on ship and patrol area)and have planes on ASW-patrol to "find" the subs. Suddenly even my poor AMs in australia started to sink/seriously damage some subs.

You can't build too many of these "kill groups", because you don't have enough commanders and will need them also to protect your carriers (more important). I only run 4 of these groups in mid. '42, but i think they have sunk 4-6 subs already. With FoW on you can't be sure, but i guess "forced to surface" and shooting till sunk as "explosion below the water with large oil slicks" can be verified as kills (at least i hope so).

And yes, it was a bit off-topic

On-topic:

I really like the AI, not perfect but very good

[steamboateng]

That's the way I've learned to do it, playing Guadalcanal Scenario. I found these ASW 'hunter-killer' TF's to be very effective. SC's and AM's fill this role well close to ports; Jap subs either miss (shallow draft?) or won't waste a torp on smaller vessels. I use minimum 2 ship TF's with a reaction range of 1. If one ship gets hit, the other(s) will seek out the offending scalliwag! 3 ship TF's are sooo much better; 'tis a wonderfull sight to watch an ASW 'gang bang' on a lurking shark. Your right, a TF commander w/ good stats help a lot. Over time, the crews get to be well trained and quite adept at their mission.
Regards

[Bullwinkle58]

ORIGINAL: Thayne

And it has some advantages over a PBEM opponent - such as the freedom to adopt personal home rules that . . . well, that make for a better story, I think.

While the AI will never match a human for original thought, it has an advantage not often commented upon which makes it difficult to play against.

It never gets tired.

It is programmed, but within that programming it never makes a mistake. It never forgets to adjust CAP ratios. It never forgets--for two weeks--to send a resupply convoy while the troops decline in strength. It never forgets to do upgrades. Or to set a better sub patrol area. It doesn't get tired, bored, lazy, or distracted by the game on TV. It makes the human opponent pay for mistakes when all of these things happen to him.

[JocMeister]

Having played only vs. the AI (so far) I agree its a very good learning tool. Unfortunantly it has its limits. I´m now in early 43 and I´m inclined to give up. I´m playing against the Jap AI which did very well in the opening stages. But in late 42 the japs lost all its CVs. It lost all of them in pairs in range of LBA. My feel is that the AI cannot properly project power. All its CVs were lost in useless raiding againt Nomea and Suva. It also have problems with moving its airgroups where its needed and wasting Betty pilots on useless targets. (Bombing a lone baseforce that is covered by heavy LRCAP again and again). I did learn ALOT though, but for me atleast I have reach a point where I feel its lacking...

I´m thinking of beginning a new game as the Japs vs the Allied AI instead. How is it? Will it do something or will it just sit and do nothing until 46? Or is it time to move on to PBEM?

[mike scholl 1]

ORIGINAL: JocMeister

I´m thinking of beginning a new game as the Japs vs the Allied AI instead. How is it? Will it do something or will it just sit and do nothing until 46? Or is it time to move on to PBEM?

It will prove even less satisfying, unfortunately. "Scripting" works best for the side with the initative. You yourself found that the AI played well in the early going in your game..., but then came "unglued" as the situation equalized. As the Allies, it begins on the defensive, in "reaction" mode..., which is already it's weak suit. Not good, unless you really enjoy kicking an opponent who's already down.

The opening 6-7 months as the Japanese is what the AI does best. Otherwise it kind of devolves into "Animated Idiocy"..., where it keeps trying to "do something", but does it quite poorly. Andy's scripts can do a lot..., but once the game really gets going, there are just too many possibilities for anyone to script for.

[Andy Mac]

I just wonder how we know that ? I have seen no Jap v Allied AI games AAR's or saves so I cannot actually say how good it is because I havent seen any games
 
I think a few players tried the Allied Ironman mod but no feedback as yet

[RUDOLF]

ORIGINAL: hawkeye_de

I'm considering to buy this game for the upcoming holiday season.

How is the AI...I mean is it at least somehow dynamic...so if you replay a scenario does it change its tactic etc?

Thanks.

The AI is poor, it cheats alot.
Better is to play a Human, more "realistic" AND more challanging.

[Andy Mac]

For a new player I dont think the Ai is 'poor' it may not challenge an experienced player long term but I have had enough feedback from players and saves to say it does ok
 
Its hard for me to judge as I wrote most of it and therefore it def cannot suprise me

[mike scholl 1]

ORIGINAL: RUDOLF
The AI is poor, it cheats alot.
Better is to play a Human, more "realistic" AND more challanging.

I would disagree, RUDOLF. For what it is (a learning tool), the AI is very good. Especially considering the enormously complex nature of WITP-AE as compared to other games. The fact that it coordinates air, sea, and land units (with many varying capabilities) as well as it does is a terrific accomplishment.

But you are right the once you have learned how to play, the AI is a poor substitute for a human opponent. It will never be able to think, scheme, plot, and "think outside the box" the way a human can.

[Feltan]

ORIGINAL: RUDOLF
The AI is poor, it cheats alot.
Better is to play a Human, more "realistic" AND more challanging.

Human opponents and PBEM allows for a far more challenging game -- agreed.

However, I think your statement about the AI is somewhat less than kind and generous. Compared to the original WITP, the AI is vastly improved. On its own merits it ranks high versus any AI I've played against in any computer game -- and that is amazing due to the complexity of WITP-AE.

I am not sure what your expectation are for an AI, but the WITP-AE AI exceeded my expectations for a computer based opponent.

Regards,
Feltan

[Bullwinkle58]

ORIGINAL: mike scholl 1

But you are right the once you have learned how to play, the AI is a poor substitute for a human opponent. It will never be able to think, scheme, plot, and "think outside the box" the way a human can.

All true.

But, and a big but for many/most AI players, is that a human opponent wouldn't satisfy the way I/we play the game. Too slow. I play a week or more of turns at a sitting, then sometimes don't play for 4-5 days. The AI is always ready for multiple turns. PBEM players either don't care about that, or compromise with it in order to get that human opponnet experience. PBEM players seem to downplay this aspect of the AI player's world, when in reality it is the BEST feature of the AI to me. I might play a PBEM game some day when life allows, but I would still have an AI game going too. One turn a day would not be enough.

[janh]

ORIGINAL: Feltan
I am not sure what your expectation are for an AI, but the WITP-AE AI exceeded my expectations for a computer based opponent.

I suppose realistic expectations for an AI that has to "deal" with a problem of the extreme complexity such as of WiTP-AE is that it can and will make mistakes. In general, for any game, I would say there are two classes of AI mistakes: small ones, and major ones that cost the motivation. Small ones, like a few unnecessary air raids , stepping in player laid (CAP...) traps once in a while, or unnecessary bloody use of its LCU's are not seriously interfering with fun. Where I would draw the line is use of key assets (like CVs, BBs, CA/CL and failing with major amphib operations in case of WITP). If it keeps sending units in after having gotten beaten there badly already, the AI scripts ought to be canceled, or delayed until reinforced. Maybe it cheats too little and should be allowed to see more through FOW and use that recon info before calling such scripts -- like a human make the decision based on expected opposition, but have the real recon info rather than having it tracked. Suppose that would be one idea for WITP2, or a major addon/patch, if there would be one in the planning.

[Feltan]

ORIGINAL: janh

ORIGINAL: Feltan
I am not sure what your expectation are for an AI, but the WITP-AE AI exceeded my expectations for a computer based opponent.

I suppose realistic expectations for an AI that has to "deal" with a problem of the extreme complexity such as of WiTP-AE is that it can and will make mistakes. In general, for any game, I would say there are two classes of AI mistakes: small ones, and game breakers. Small ones, like a few unnecessary air raids , stepping in player laid (CAP...) traps once in a while, or unnecessary bloody use of its LCU's are not seriously interfering with fun. Where I would draw the line is use of key assets (like CVs, BBs, CA/CL and failing with major amphib operations in case of WITP). If it keeps sending units in after having gotten beaten there badly already, the AI scripts ought to be canceled, or delayed until reinforced. Maybe it cheats too little and should be allowed to see more through FOW and use that recon info before calling such scripts -- like a human make the decision based on expected opposition, but have the real recon info rather than having it tracked. Suppose that would be one idea for WITP2, or a major addon/patch, if there would be one in the planning.

I don't disagree with you. There is room for improvement -- and in my experience to date it would be with AI amphibious assaults & the issue of repeat assaults.

Game breaker? Not for me. I've played PBEM against some human opponents who probably wouldn't do the repeat amphibious assault mistake, but made up for it with other questionable actions.

Regards,
Feltan

[Bullwinkle58]

ORIGINAL: Feltan

Game breaker? Not for me. I've played PBEM against some human opponents who probably wouldn't do the repeat amphibious assault mistake, but made up for it with other questionable actions.

Regards,
Feltan

Yeah, I've never had the AI quit on me and refuse to answer e-mails. [:)]

[janh]

Right, game breaker is too harsh in most cases. Edited. But I can understand people that switch to playing the weaker side, or the one that has to engage in offensive actions in many games. Anyway, if you think back to Japanese AI in PacWar, or other AIs like in TF1942 or Civilization, I think we ought to be quite satisfied with most games today...

[berto]

ORIGINAL: Bullwinkle58
One turn a day would not be enough.

agreed++

When I play, I tend to binge on marathon sessions, then maybe won't touch a game again until a week or two (or ...) later. Much like I prefer watching a full-length movie (better: a movie marathon) than watching a weekly TV series (in fact I never watch TV). Computer opponents are totally adaptable and acquiescent to my weird on again off again playing schedules.