Is the Japanese AI really this weak?

Post descriptions of your brilliant successes and unfortunate demises.

Moderators: wdolson, Don Bowen, mogami

Post Reply
BKnowles
Posts: 10
Joined: Fri Nov 26, 2004 10:09 am

Is the Japanese AI really this weak?

Post by BKnowles »

Hi,

I am playing my first serious game of WITP as the allies using the CHS mods and Mike Browns' Extended Map.

Based on the comments from other in this forum I will gladly admit that I did not really have a proper sense of the right and wrong things to do in this game, nor the full knowledge of the history of this theater of operations. However, over the years I have played a number of board games from other manufacturers in this theater, thus I did have some idea on the need to avoid the Japanese carriers in the early part of this war.

After a couple of experiments with the game using the stock map, I had gathered enough understanding of the game interface to settle down and start a real campaign. The one thing I had noticed during these experiments was the similarity of the basic Japanese initial battle plan (this as I played the Japanese in each experiment).

I thus decided to play as the allies and configure the game with all realism switches in the on state and reinforcements for both sides set at fixed.

At the outset I had settled myself on striving to interfere with Japanese merchant shipping in every way possible. I also intended to use the Chinese forces aggressively to engage as many Japanese forces as possible in the hope of keeping them from being used elsewhere. I had also concluded that both the Phillipines and Malaya were lost causes and the best think I could do in both locations was to execute a fighting retreat towards Singapore and Manila respectively.

Singapore fell mid January '42 and Manila eventually fell May 20th.

Before the fall of Singapore I had been able to extract most of the air force and a number of infantry, engineer and aviation units, most of which were deployed in Panembang or Batavia. A few were routed to Port Moresby and Rabaul.

I then set up a kill zone in the straits between Singkep, Singkawang, Palembang triangle and then spent my time bombing everything that passed into that region from Palembang, Batavia and a tiny carrier force. Eventually I was forced out of Palembang and retreated the ground troops in good order to Teloekbetoeng and Batavia, and move the airforce to Batavia.
I also placed some Catalinas in Teloekbetoeng.

And thus began the battle of the northern Java Sea.

It is now the start of July 7th 1942, and I am still bombing Palembang and the triangle in force and the current Intelligence screen below details the state of the game I guess.

Although this is my first game, and it is only July, I am not impressed with the performance of the AI. However, I would nevertheless ask an opinion in this regard from the more experienced players, and also whether it is also worthwhile continuing this particular game.

Archangel






Image
Attachments
SouthWest..5th1942.jpg
SouthWest..5th1942.jpg (35.46 KiB) Viewed 188 times
BKnowles
Posts: 10
Joined: Fri Nov 26, 2004 10:09 am

RE: Is the Japanese AI really this weak?

Post by BKnowles »

Oops, wrong graphic supplied the first time.



Image
Attachments
IntelRepo..5th1942.jpg
IntelRepo..5th1942.jpg (63.09 KiB) Viewed 188 times
princep01
Posts: 945
Joined: Sun Aug 06, 2006 10:02 pm
Location: Texas

RE: Is the Japanese AI really this weak?

Post by princep01 »

You should await the comments from those more experienced than me (currently in my first PBEM CG and having played two, partial, CG as the Allies thru the first couple of months).  However, I suspect the answer will be the same....the game is over.  400+ sunken Japanese ships at this stage of the war!!??  Allied air losses 1/2 of the Japanese air losses!!?? 
 
Yes, I think the AI is weak.  But, it is useful as a tool for learning the game mechanics and unit capabilities  As a total novice, I had a very easy time beating up on the Japanese AI.  It clearly appears you have had the same experience.
 
Now...let's talk about playing the game as the Allies against a human opponent.  I think you will find that the IJN/IJA are very formitable, indeed.  This is especially true if those forces are in the hands of an experienced Japanese commander.  In my game, a more experienced Japanese commander has just landed two full divisions, a Base Force and construction unit (successfully) on the NW end of New Caledonia. I am in Noumea in force, but this is scaring the bejabbers out of me because it is only mid-March 1942.
 
A wholly different experience awaits you.  Find a human opponent in the game wanted sub-forum and let the fun begin.  Quite stimulating.
 
Hope this helps.
AmiralLaurent
Posts: 3351
Joined: Tue Mar 11, 2003 8:53 pm
Location: Near Paris, France

RE: Is the Japanese AI really this weak?

Post by AmiralLaurent »

Archangel, first you will have more returns if you post this on the main forum, this post and your former about captured aircraft should not be in the AAR section.

Then you are fully right that the AI is weak. WITP is a complex game, and not sold at a scale large enough to enable Matrix to spend money to have a real AI. At least that is my opinion. So the AI, rather than evaluating the situation each turn, has more or less fixed objectives (Manila, Singapore, etc...) and if you do something unhistorical, it will be easily confused and you can slaugther it piecemeal easily.

By the way the AI is better on the stock map than on any mod, for the same reason (fixed objectives).

Anyway the AI is good enough to learn how the game works by playing against it. Then the true interest of the game is to play PBEM against an human opponent.
User avatar
saj42
Posts: 1132
Joined: Tue Apr 19, 2005 12:02 pm
Location: Somerset, England

RE: Is the Japanese AI really this weak?

Post by saj42 »

To give the AI a bit more of a helping hand set the difficulty level to at least 'HARD'. This gives the poor AI some help - like ignoring 'out-of-supply' and having a peek at some of your TFs and bases. It too easy to beat up the AI when set to 'historical' difficulty level.
Good learning tool. Oh and yes this game is over.
If you have had enough practice playing the first 6 months/year then consider playing a '43 or '44 scenario - as Allied this may help you develop planning and organising skills for launching large amphibious combined ops.
Image
Banner by rogueusmc
modrow
Posts: 1100
Joined: Sun Aug 27, 2006 10:02 am

RE: Is the Japanese AI really this weak?

Post by modrow »

I, too, have limited experience... but I want to encourage you to go for PBEM.
 
I did play for quite a while against AI, which partly does really dumb things, especially if you can throw it "off its script". In a Gualdacanal scenario as Allied player against AI I once took Gasmata (sp?) - and AI was playing on as if I had not done that.
 
Only recently I started PBEMs, mostly small scenarios, though I have one full campaign running as Allied player as well. When I started my first I thought WOW - what a difference. Got spanked, of course, but I learned a lot of lessons AI would never have taught me.
 
When I started playing against a second opponent, he had a completely different style of play - again a few lessons to learn, but most importantly it was like a completely different game ! Third opponent - same story, new game again !
 
So it's LOTS of fun to do PBEM - dare to do the jump.
 
Just my 2cts...
   Hartwig
BKnowles
Posts: 10
Joined: Fri Nov 26, 2004 10:09 am

RE: Is the Japanese AI really this weak?

Post by BKnowles »

To everyone who offered guidance,

Thank you very much.

I will also try and post into the relavant areas in future. Please accept my apologies.

Archangel
herwin
Posts: 6047
Joined: Thu May 27, 2004 9:20 pm
Location: Sunderland, UK
Contact:

RE: Is the Japanese AI really this weak?

Post by herwin »

ORIGINAL: AmiralLaurent

Archangel, first you will have more returns if you post this on the main forum, this post and your former about captured aircraft should not be in the AAR section.

Then you are fully right that the AI is weak. WITP is a complex game, and not sold at a scale large enough to enable Matrix to spend money to have a real AI. At least that is my opinion. So the AI, rather than evaluating the situation each turn, has more or less fixed objectives (Manila, Singapore, etc...) and if you do something unhistorical, it will be easily confused and you can slaugther it piecemeal easily.

By the way the AI is better on the stock map than on any mod, for the same reason (fixed objectives).

Anyway the AI is good enough to learn how the game works by playing against it. Then the true interest of the game is to play PBEM against an human opponent.

Doing AI is *Hard*. I know--I'm the house neuroscientist in an AI group.
Harry Erwin
"For a number to make sense in the game, someone has to calibrate it and program code. There are too many significant numbers that behave non-linearly to expect that. It's just a game. Enjoy it." herwin@btinternet.com
User avatar
wdolson
Posts: 7689
Joined: Tue Jun 27, 2006 9:56 pm
Location: Near Portland, OR

RE: Is the Japanese AI really this weak?

Post by wdolson »

ORIGINAL: herwin
Doing AI is *Hard*. I know--I'm the house neuroscientist in an AI group.

I have some limited experience writing a rudimentary AI too. I was also an embedded programmer for 15 years. I learned how to squeeze everything out of hardware.

The more an AI has to consider, the more complex is gets. And the progression is geometric or expodential, not linear. For example an AI that has to consider 8 variables is going to be much more than twice as complex than an AI that considers 4.

The AI in the game could be better, but I don't think it's possible to make it a really tough opponent. It has to evaluate and consider a map that covers over half the Earth's surface and has to manage it all down to pretty small detail.

I still play against the AI. My time is too variable to play PBEM right now. I can often play a few turns against the AI in the evening, but then I might end up having to go a month without playing at all.

I'm practicing with Japanese industry and trying to help out the AI. I've noticed Japanese air power is a bit better than the last campaign I played. Last time many Japanese air units were out of aircraft without enough replacements by late 1942.

I also sort of play against myself to see how quickly I get go over onto the offense and push the Japanese back to the Home Islands. It does have the advantage of keeping the campaigns shorter. Still takes a while to completely crush Japan as the Allies.

Bill
WIS Development Team
Post Reply

Return to “After Action Reports”