Japanese Strategy?

[usersatch]

I am at July of 1942 and have noticed that everything that happened in the real war (from the Jap point of view) hasnt happened here. For example, no Japanese troops in Guadalcanal, no Coral Sea battle, no Midway battle. I play to win and try not to play "scripted", but I was wondering if the AI trys to follow a script of some sort, or does it react to what it knows I am doing?

I noticed that the Gilbert Islands are fairly unoccupied at this point as well and am thinking about starting that campaign earlier than it did in real life.

Is the AI sucking me into some great void ready to ambush me or is it not that aggressive/smart?

[ckk]

The more aggressive or non historic you play the AI becomes more confused and finally stops being aggressive and then just keeps trying to supply and keep bases like Truk and Rabaul which reults in convoy slaughters[:(] Play on very hard and use house rules like no allies escape PI and DEI and play on the same timeline as IRL and you'll keep the IJ AI in the game longer[:)]

[dpstafford]

ORIGINAL: usersatch
Is the AI sucking me into some great void ready to ambush me ......

You're half right. The AI is sucking........

[bradfordkay]

The AI has certain bases rated more highly than others when it comes to what it wants to capture. It first tends to go after the big bases in the Philippines, Malaya and the DEI. What order it goes after them is somewhat random. How well you hold off the AI invasions is going to determine just how far the AI will continue to invade. My feeling is that a Battle of Midway occuring at Midway is unlikely unless you've already lost Wake, Rabaul, Lae, Buna, Port Moresby, Gili Gili (Milne Bay), Lunga, Luganville and Noumea. Midway just isn't worth that much.

Now some form of the Battle of Coral Sea seems to occur in every game I've played against a Japanese AI. At some time it is going to send its carriers (or some of them - the last time I got Shokaku, Zuikaku and Zuiho) to support some form of offensive operation against PM/GG. If you've the carrier forces to stand up to them, you'll get your Battle of the Coral Sea.

It will react to your advances, with vigor if you grab a highly regarded base. One trait I think that the AI does seem to have is to react weakly everywhere if you attack too many places at once.If you're looking for a decisive battle, you might want to concentrate on only a few major campaigns.

[LargeSlowTarget]

I've noticed that the Japanese AI tends to overstack ground units at some places, effectively making them 'un-assaultable', but neglects to adequately defend other locations nearby. My recon has counted over thirty LCUs on Saipan, but only two each at Tinian and Guam - resulting in two easy conquests for me and a lot of isolated and starving troops left on Saipan to wither on the vine and yielding easy VPs... Not very smart AI. Too bad I don't have the time for PBEM.

[Lrfss]

[font="Arial"][/font]
Hey LST!

I agree with your remark! Should you find the time for a PBEM WITP, look me up!

Later,

Lrfss...

[Sharkosaurus rex]

If you are playing against the AI and want it to be a little more interesting then why not attack the heavily defended bases instead of avoiding them? Take out Rabaul, Truk, Saipan etc where there are heaps of Japs then you might enjoy the battle. Sure island hopping was a strategy but in the AI game you might want to attack head-on?

[Graymane]

ORIGINAL: usersatch

I am at July of 1942 and have noticed that everything that happened in the real war (from the Jap point of view) hasnt happened here. For example, no Japanese troops in Guadalcanal, no Coral Sea battle, no Midway battle. I play to win and try not to play "scripted", but I was wondering if the AI trys to follow a script of some sort, or does it react to what it knows I am doing?

Well, I'm not sure at what level it is doing "strategic" kinds of thinking. Most of those strategies IRL were strategically suspect ones to begin with. Why would an AI want to repeat them? The 3 battles/campaigns you mention were all cause/effect kinds of things. I think it makes at least a few better decisions than the Japanese did historically, for instance it seems to keep the 1st air fleet together. It doesn't seem to go wandering off to the Indian ocean for no apparent reason allowing the allies to do raids in the central pacific to increase experience.

I noticed that the Gilbert Islands are fairly unoccupied at this point as well and am thinking about starting that campaign earlier than it did in real life.

The AI seems to try for the gilberts in the first few weeks of the war in my game but I beat him to it anyway.

Is the AI sucking me into some great void ready to ambush me or is it not that aggressive/smart?

Well, I would be much more concerned with the tactical things the AI is doing rather than whether or not it is following some script from the past. It is rather bothersome that the pathing AI doesn't seem to understand the notion of air power. This seems to work on both the comp AI and the AI that runs your own ships if you let them return to port on their own. If I try to send a convoy to the south pacific from S.F., for example, it will path right through Enwietok area.

The other problem is that it doesn't seem to want to cover its TF with LCAP. I still hold the Andaman Islands and it sends convoy after convoy right up and down the burma coast in range of its own air bases while my bombers kills its ships. Same thing is happening in N.G./Solomons where I hold P.M., Gili and Guadalcanal. It also seems to send lots of capital ships sailing aimlessly around those areas.

[LargeSlowTarget]

ORIGINAL: Sharkosaurus rex
If you are playing against the AI and want it to be a little more interesting then why not attack the heavily defended bases instead of avoiding them? Take out Rabaul, Truk, Saipan etc where there are heaps of Japs then you might enjoy the battle. Sure island hopping was a strategy but in the AI game you might want to attack head-on?

Okay, I'll start a new campaign right away and will invade Tokyo in February 42 just to make it a little more interesting. I've always been more interested in mind- and senseless butchery than in pursueing a viable strategy, that's why I'm playing strategy games and not shooters... [8|]

[bradfordkay]

I some respects wasn't this what happened IRL? The Japanese held certain bases with great strength and Nimitz decided not to go after those (Truk, Rabaul, Kwajeilin?). Yes, we went after Saipan because we needed the air bases for B29s, but there were a lot of Japanese strongholds that we left to wither on the vine.

[Bradley7735]

Tarawa, Kwajelein, Palau, Saipan, Iwo Jima, Okinawa, ...... None of these fortresses were bypassed.

Most of what we bypassed was in the solomons and New Guinea.

I have no idea why Palau was not bypassed. At least there was some logic to the others.

[rtrapasso]

ORIGINAL: Bradley7735

Tarawa, Kwajelein, Palau, Saipan, Iwo Jima, Okinawa, ...... None of these fortresses were bypassed.

Most of what we bypassed was in the solomons and New Guinea.

I have no idea why Palau was not bypassed. At least there was some logic to the others.

The big bases at Rabaul (New Ireland?) and Truk were also bypassed.

[ckk]

IIRC The Gilberts and Marshall's were taken as forward air and supply bases. The Mariana's were taken as B-29 strategic bombing bases Iwo as fighter airfield/B29 emergency landing field. Also eliminated an IJ early warning station and Okinawa in preparation for the Home Islands Invasions. Pelelieu was a mistake and the PI was for Politics? Formosa would have been better but Chiang was making the price way too costly. In other words there were usually good reasons for taking on Imperial fortresses.

[Hornblower]

Don't forget Marcus and Wake being bypassed. And i would think the Marines would have loved to Hit Wake and take it back. Not to mention the Crew of the saratoga and the rest of her dec '41 taskforce- well whats left anyway.

[Bradley7735]

ORIGINAL: rtrapasso

ORIGINAL: Bradley7735

Tarawa, Kwajelein, Palau, Saipan, Iwo Jima, Okinawa, ...... None of these fortresses were bypassed.

Most of what we bypassed was in the solomons and New Guinea.

I have no idea why Palau was not bypassed. At least there was some logic to the others.

The big bases at Rabaul (New Ireland?) and Truk were also bypassed.

I think of them as the Solomons and New Guinea. I know they aren't, but they are the main bases involved in the 1942 and 1943 battles in the Solomons and New Guinea. I kind of lumped them in with Lae, Wewak, Bougainville and Shortlands.

[Graymane]

ORIGINAL: Bradley7735

Tarawa, Kwajelein, Palau, Saipan, Iwo Jima, Okinawa, ...... None of these fortresses were bypassed.

Most of what we bypassed was in the solomons and New Guinea.

I have no idea why Palau was not bypassed. At least there was some logic to the others.

Probably because hindsight is 20/20 and the threat estimate from Truk was off. It was meant to complete the "encirclement" of Truk and as a support area for the upcoming P.I. invasion.

[Mike Scholl]

ORIGINAL: usersatch

I am at July of 1942 and have noticed that everything that happened in the real war (from the Jap point of view) hasnt happened here. For example, no Japanese troops in Guadalcanal, no Coral Sea battle, no Midway battle. I play to win and try not to play "scripted", but I was wondering if the AI trys to follow a script of some sort, or does it react to what it knows I am doing?

Is the AI sucking me into some great void ready to ambush me or is it not that aggressive/smart?

You are trying to think of the term "AI" as "automated intelligence"...., it will help if you keep the phrase "animated ignorance" in your head instead. After a great deal of work (and a lot of bizarre tweeks and adjustments), the designers provided an AI that could conduct a moderately successfull opening Japanese Offensive (provided the Allied Player followed the "historical script' in a general manner as well. Given the massive interaction of the hundreds of units of varying types, capabilities, and sizes involved, this was no mean accomplishment. But to expect the AI to actually plan strategically is a pipe-dream. Just to get it to play the opening 150 turns in a reasonably successful manner caused all sorts of problems when you shift to head-to-head play.

The game would have been a lot getter if they had only promised an AI that could successfully handle several shorter scenarios (which COULD BE heavily "scripted") and left the real game to PBEM. What we have now is an abortion. The AI CAN'T (nor can it reasonably be expected to) play a 1600 turn campaign game. And because of the changes and adjustments made to enable it to "try", players are having a tough time as well (resorting to all kinds of "house rules" to try and restore some competative reality).

[hawk66]

ORIGINAL: Mike Scholl


The game would have been a lot getter if they had only promised an AI that could successfully handle several shorter scenarios (which COULD BE heavily "scripted") and left the real game to PBEM. What we have now is an abortion. The AI CAN'T (nor can it reasonably be expected to) play a 1600 turn campaign game. And because of the changes and adjustments made to enable it to "try", players are having a tough time as well (resorting to all kinds of "house rules" to try and restore some competative reality).

I agree, that it is very difficult to implement an AI for such a game. But I disagree that only an AI should exist for the smaller scenarios of the game. Personally I would never buy such a game.
Furthermore I think it is not a question if it's possible but how many ressources are available for implementing one. AI research has done some progresses in the last year and I think such a game AI has to be a mixture of learning algorithms and having expert knowledge...

I have not bought the game already but I plan to do this during the next few weeks when I've more time and looking forward to it and let me suprise by the quality of the AI [:)]

[Mike Scholl]

ORIGINAL: hawk66


I agree, that it is very difficult to implement an AI for such a game. But I disagree that only an AI should exist for the smaller scenarios of the game. Personally I would never buy such a game.
Furthermore I think it is not a question if it's possible but how many ressources are available for implementing one. AI research has done some progresses in the last year and I think such a game AI has to be a mixture of learning algorithms and having expert knowledge...

I have not bought the game already but I plan to do this during the next few weeks when I've more time and looking forward to it and let me suprise by the quality of the AI [:)]

So you WILL pay for a game with an AI that can't play the Major Scenario worth a damn?
I'm not blaiming the designers for this..., implementing an AI that can deal with Grand Strategy, Operations, Tactics, and Logistics in a four year, daily turn tactical game is probably beyond any programing technique available to them. I blame them for the screwy choices thay made to TRY and make the AI workable in such a scenario...which made for all kinds of strange occurances when playing head-to-head games.

If all you want to do is play the computer, it would give you a better game playing a series of smaller scenarios in succession. Those COULD be effectively "scripted" for the AI to play as the choices and reactions are much more limited. The need to make special rules for the AI would have been much less, and the play more realistic. Then if you ever graduated to playing head-to-head, you would have the "Grand Scenario" fresh for your enjoyment. Or are you just a fan of "beating up on the computer"?

[Graymane]

A scripted AI has it owns problems, but probably better for a game like this. The Paradox folks have been doing scripted AIs for years and they work pretty well given the various scope and time periods they are geared for. A scripted AI like they have can do exactly what you are talking about without resorting to separate scenarios via switching the AI files dynamically according to events in the game. It is sort of like those books that kids read that have different outcomes based on choices you take while reading

But before a scripted AI can be even sort of useful, it has to solve the tactically easy stuff first or it will still act stupid no matter how much it decides on the "right" island to take. Use LCAP over your convoys when you can, don't path convoys right through a concentration of heavy air power just cause it is the shortest distance, etc.

One of things that really turned me off from HOI 2 was that while you could script almost anything in the AI files, the AI couldn't build a correct naval TF or use air units in any meaningful fashion (but it does path convoys a hell of a lot better than WiTP).

They do have the same kinds of problems with island areas as WiTP though. It would seem that doing an island AI is much harder than a land AI.