CHS versus Stock

[herwin]

I'm currently running a CHS scenario using the extended map, and the Japanese AI appears to be a good deal more careful than its style in the stock game. Is this consistent with the experience of others?

[keeferon01]

Always seems to me when I play CHS that the Japanese AI is very very careful with the aircraft, almost to the point where I think the air routines using the mod are goofy.

[herwin]

ORIGINAL: Ron James

Always seems to me when I play CHS that the Japanese AI is very very careful with the aircraft, almost to the point where I think the air routines using the mod are goofy.

In my game during December 1941, the AI playing Japan had 10% more air losses than the Allies. On the other hand, the merchant vessel losses for the Allies were phenomenal. Currently (1 Jan 1942), the Allies still hold Hong Kong, Singapore, all of Borneo except Kuching, a patch of hexes around Clark, part of Mindanao, miscellaneous hexes in the Philippines, Burma (except for Tavoy), Rabaul, and Wake. Most of the northern Malaya forces were evacuated to Sabang after Kota Bharu was lost and the Japanese started rolling down the coasts. The AI just landed forces at Kavieng and Shortlands--in both cases, a squadron based at Rabaul descended on the invasion fleet and sent it home before it could land supplies.

[herwin]

Singapore fell on 1 Jan, but everything else is still holding out on 11 Jan. The Japanese move into Burma seems to be delayed, since I've already evacuated the southern half of the country. Based on SIGINT, there's something moving around in the western Marshalls, but I can't figure out what. I'm also wondering where the Death Star is. The Japanese attack on Kavieng died and went away, and the Shortlands force is apparently too weak to attack my ungarrisoned base. I'll be putting the 2nd USMC para bn in shortly if the IJN doesn't walk in.

Amazing.

[keeferon01]

harry if you had to improve the AI, what area would you work on first

[herwin]

ORIGINAL: Ron James

harry if you had to improve the AI, what area would you work on first

Good question. An effective AI has to be able to do forward planning (medium to long-term) and at the same time has to have short-term judgement about what is likely to pay off or hurt immediately. The WiTP AI appears to have weak short-term judgment. I've evacuated the southern half of Burma, and any human player would advance cautiously to take what I've offered him. The AI watches from Tavoy. Meanwhile it grides its forces into dust against Hong Kong, Manila, and Clark Field. I have a surface fleet within dash speed of Tarakan, Western Borneo, Kavieng, and Shortlands, and every time the AI sends a landing force in those directions, it gets jumped on. I have a barrier patrol force based on Sabang closing the northern end of Malacca Straits, and the AI persists in sending transports into the meat grinder. So it is clear the WiTP AI needs some basic common sense. My strategic plan is to hold on where I am strong and force the AI to use the KB, and similar assets to make progress. I retreat from that and take advantage of the AI's focus to hold or advance elsewhere. So far, it's working.

To improve the AI, I would survey all the game reports and come up with some short-term common sense about what works and what doesn't. The Japanese didn't need to pull off grand strategic advances to reach their goal of a stalemate--they just needed to be very hard to defeat once they captured their defense perimeter. An AI with better common sense and a few carefully-designed plans covering the key points of the defense perimeter and internal stop lines that could be adapted to the local strength of the Allies and normal variation in combat outcomes would do fine.

[herwin]

ORIGINAL: herwin

ORIGINAL: Ron James

harry if you had to improve the AI, what area would you work on first

Good question. An effective AI has to be able to do forward planning (medium to long-term) and at the same time has to have short-term judgement about what is likely to pay off or hurt immediately. The WiTP AI appears to have weak short-term judgment. I've evacuated the southern half of Burma, and any human player would advance cautiously to take what I've offered him. The AI watches from Tavoy. Meanwhile it grides its forces into dust against Hong Kong, Manila, and Clark Field. I have a surface fleet within dash speed of Tarakan, Western Borneo, Kavieng, and Shortlands, and every time the AI sends a landing force in those directions, it gets jumped on. I have a barrier patrol force based on Sabang closing the northern end of Malacca Straits, and the AI persists in sending transports into the meat grinder. So it is clear the WiTP AI needs some basic common sense. My strategic plan is to hold on where I am strong and force the AI to use the KB, and similar assets to make progress. I retreat from that and take advantage of the AI's focus to hold or advance elsewhere. So far, it's working.

To improve the AI, I would survey all the game reports and come up with some short-term common sense about what works and what doesn't. The Japanese didn't need to pull off grand strategic advances to reach their goal of a stalemate--they just needed to be very hard to defeat once they captured their defense perimeter. An AI with better common sense and a few carefully-designed plans covering the key points of the defense perimeter and internal stop lines that could be adapted to the local strength of the Allies and normal variation in combat outcomes would do fine.

By the way, the barrier patrol just got hit by the Eye of Sauron. Two CLs torpedoed by G3Ms/G4Ms. The aircraft based in Sabang took up the slack. Also the KB may have found me. I was refueling the Sara off the north end of Buka and 19 Kates jumped her, followed by a second, smaller strike. She had CAP up and shot 13 of them down with no damage. I'm trying to decide whether to cut and run or bet it's a CVL. Given that my patrol aircraft missed the fleet, I suspect it's not the full KB. I should probably pull back six to twelve hexes, bring the Enterprise over to help (she was nosing into Makin to find out what was happening there), and let the situation resolve itself. Historically, IJN carrier forces could only steam at full speed for a couple of days, so whatever it is, it should be hurting for fuel if it chases me.

[herwin]

Getting hit by the KB (or KB jr) raises a question. When an American fleet carrier had to lift her skirts and run in WWII, what did she take with her as escorts? How far did she run? And then, what should she be ordered do in WiTP? I form my air TFs using a carrier, three heavy (or large light) cruisers, and six destroyers. There are IJN subs in the area. If the game were realistic, they would be able to do much of anything about a carrier passing through at 30+ knots, but you know what they say--it's only a game.

Oh, the other carrier is the Lexington, not the Enterprise. The Enterprise got punched in the nose when she checked out Kwaj after visiting Wake, and needed some repair work in Pearl.

[herwin]

Here's the end of the story. My carrier TFs were cruising about 200 miles behind my reconnaisance line (a line of VP bases), which were about 600-1200 miles back from the AI's line of forward bases. I run a day and check--WTF? The KB is now sitting next to my carriers and has just hit them with a major strike. Not even a clue the previous turn it was in the area, and the carriers apparently were caught completely by surprise. Bubble, bubble...

The only thing I can figure is that the IJN has teleportation. Of course, Halsey's run across the Marshalls in the first game I tried had that flavour. One day, the Enterprise was off Tulagi, the next day she was next to Kwaj. Does this happen often?

[keeferon01]

I think I am the only one reading your lil experiment with the ai, dont think anyone really cares about chs to be honest on here, what with the new thing thats being worked on.

[herwin]

ORIGINAL: Ron James

I think I am the only one reading your lil experiment with the ai, dont think anyone really cares about chs to be honest on here, what with the new thing thats being worked on.

I'm not sure the new thing will be that much better.