Wow! - Make that WOW!

[Don Bowen]

ORIGINAL: Tankerace

Why AI historical? I have noticed you have to set AI to hard or very hard to get decent results. Granted, nothing to do with this test, but would explain relatively slow game progress.

I have a very good explanation for that - I forgot to change it after the V1.5 update reset everything.

[Don Bowen]

ORIGINAL: bstarr

I'd bet my bottom dollar that most of those kills are subs hit while bombers are attacking an allied port. Particularly Manila.

Sunk locations are mostly at sea in the choke points. Hint - keep your subs away from Amoy!

[bstarr]

ORIGINAL: Don Bowen

ORIGINAL: bstarr

I'd bet my bottom dollar that most of those kills are subs hit while bombers are attacking an allied port. Particularly Manila.

Sunk locations are mostly at sea in the choke points. Hint - keep your subs away from Amoy!

Ouch! Not good.

[Andrew Brown]

ORIGINAL: Bradley7735
The issue with subs has me concerned. But not as much as the AI's issues. Do you think CHS will be playable vs the AI? I don't do PBEM and I'm looking forward to CHS. But not if the AI acts like a wet noodle.

I'm hoping the AI does a bit better when it's 1 human vs ai instead of ai vs ai.

Thanks!!

I agree, the AI performance is much more of a concern. I ran a short test last night and it was slow to advance in the Philippines and Malaya as well. I didn't have time to see why.

Perhaps a test should be run as follows: Run the CHS scenario AI vs AI for a few game months. Stop it and look at what LCUs the Japanese are using in the Philippines and Malaya. Then run an AI vs AI game of the stock map and scenario 15, then look at the difference between the forces committed by the Japanese in the two theatres. This should provide clues as to what is going on. I may have time to do that in the next few days but if someone else can try this it would be most helpful.

Just as an aside, I have run several AI vs AI games to test my scenarios (e.g. Scen 115) and they seem to work OK.

[m10bob]

UH UHHH Andrew..Too many people reporting similar ASW problems who are not on CHS..
It's *possibly* linked to 1.5 (looking for the thunderbolt from the sky)..

[pad152]

The problems gotta be the Hysterical first turn[:D]

Hysterical First Turn: ON

Somebody needs to run scenario 15 AI vs AI and see what happenes.

Is witpchk1.2 showing anything.

[jwilkerson]

ORIGINAL: Don Bowen


Allied Sub Doctrine: ON
Japanese Sub Doctrine: ON
Fog of War: ON
Advanced Weather Effects: ON
Allied Damage Control: ON
Hysterical First Turn: ON
Player Defined Upgrades: ON
Vary Setup: OFF
December 7th Surprise: ON
Reinforcements: Allied and Japanese both FIXED
Auto Sub Ops: ON
AI Difficulty: Historical
All three of the auto upgrades/replacements/expands are set to off but I BELIEVE these are ignored in an AI-AI game.
Combat Animations off, reporting off, clouds off, all timing controls set to minimum delay (not that these matter).

Don

Ok, here are my test results, from Stock Scenario 15 through 5/29/42 AI vs AI ( above settings ). I'll first post without comment. But as we can see many less subs sunk and all ( known causes ) were by DC.

[Don Bowen]

(Sigh)

Well, it is obvious that CHS has confused the AI.

In last night's test I slept very well and the game ran until September (1942). The Japanese completed the initial landings in the Philippines and then just sat there. Supply seems to be a concern. Almost all of Malaya was taken and the Japanese were on the doorstep of Singapore, and they had advanced to the ourskirts of Rangoon. None of the follow-on invasions took place: Lingayen, Davao, Kuching, Rabaul. Also, there appears to have been no use of heavy units after the pre-planned actions. ALL Carriers, Battleships, and Cruisers were sitting in port except for the Kota Bharu bombardment ships which were still at Kota Bharu. Submarine results were about the same as the previous test - dozens of allied subs lost to aircraft, no Japanese subs lost.

I'm going to mull over options but it appears that a "Not suitable for AI's under 13 Years Old" tag may be appropriate to CHS. I am depressed.

[Ron Saueracker]

Hey, Don. Try not to be too depressed. You did not program the AI. Perhaps when/if the code is released the ability to script the AI will be too.

Think of it this way. This is one hell of a PBEM only scenario campaign. Now to get any kind of entertainment out of WITP you will be forced to play PBEM. You might just catch the bug.[;)]

Those sub losses are obviously attributable to the whacked out ASW routines which are working as designed. That at least I suspect has nothing to do with the AI "breaking".

[Ron Saueracker]

ORIGINAL: bstarr

I'd bet my bottom dollar that most of those kills are subs hit while bombers are attacking an allied port. Particularly Manila.

No way, Jose. [:)] I bet these are subs hit in open water by new ASW routines.

[Ron Saueracker]

In last night's test I slept very well and the game ran until September (1942). The Japanese completed the initial landings in the Philippines and then just sat there. Supply seems to be a concern.

I'm just wondering. Seeing as the game was designed with all this extra supply all over the place to "aid the AI", maybe it does more than simply aid. Maybe it was a bandaid added because these problems you are seeing may have occurred early on in development. With all the reductions to supply at bases and with capacities, perhaps the AI is now incapable of functioning at all.

What if for Player vs AI games, the side which the AI is to "command"([8|]) retains all original specs (supply totals at bases, original load capacites etc) while the side which is to be played by human retains the modified CHS values?

[Bradley7735]

Yeah Ron. I tried to say essentially the same thing in Don's other thread, but my wording wasn't as good as your message above.

[Don Bowen]

ORIGINAL: Ron Saueracker

In last night's test I slept very well and the game ran until September (1942). The Japanese completed the initial landings in the Philippines and then just sat there. Supply seems to be a concern.

I'm just wondering. Seeing as the game was designed with all this extra supply all over the place to "aid the AI", maybe it does more than simply aid. Maybe it was a bandaid added because these problems you are seeing may have occurred early on in development. With all the reductions to supply at bases and with capacities, perhaps the AI is now incapable of functioning at all.

What if for Player vs AI games, the side which the AI is to "command"([8|]) retains all original specs (supply totals at bases, original load capacites etc) while the side which is to be played by human retains the modified CHS values?

This is a pretty good idea. I see that I have lots of work to do.

Should I release CHS as a Head-to-Head right away?? These other tests could take some time!!!!

[Ron Saueracker]

Should I release CHS as a Head-to-Head right away??

You would make many people very happy. More time out on the street more time for any booboos to come to light as well while the AI issue is investigated.

[bstarr]

ORIGINAL: Ron Saueracker

Should I release CHS as a Head-to-Head right away??

You would make many people very happy. More time out on the street more time for any booboos to come to light as well while the AI issue is investigated.

Ron,
Oh . . . uh . . . sounds like we're on the same track. I just replied with almost the exact same suggestion above. It sounded like you were throwing in the towel earlier, but I knew damn good and well you weren't no quiter. [;)]
bs

[Ron Saueracker]

ORIGINAL: bstarr

ORIGINAL: Ron Saueracker

Should I release CHS as a Head-to-Head right away??

You would make many people very happy. More time out on the street more time for any booboos to come to light as well while the AI issue is investigated.

Ron,
Oh . . . uh . . . sounds like we're on the same track. I just replied with almost the exact same suggestion above. It sounded like you were throwing in the towel earlier, but I knew damn good and well you weren't no quiter. [;)]
bs

Well, I can't really do alot not having the game with me and no ISP at the moment but I still can't keep my mouth shut. Foghorn Leghorn Syndrome I guess.[;)]

May look like we'll need a CHS version for AI and PBEM both.[X(]

[Don Bowen]

I'm continuing to play with this and I am beginning to form the opinion that the problem is not any of the settings that we have made (bases, capacities, etc.) but rather some form of invalid entry somewhere. Now, if I could just find it...


Actually, I used to do this sort of thing for a living and I believe I can isolate the problem. Will take a while. I'm going to release CHS to Spooky tomorrow and then get right to it.

Don

[Bradley7735]

Don, I remember hearing that the reason a lot of bases started out with a port of 1 and 1,000 supply was to help the AI. You might want to put back the supply and make a minimum of 1 port (0 airfield is ok, I think) to all the bases the AI takes up to May 42.

I haven't seen the CHS mod, but I bet a lot of the bases in PI and Solomons were reduced to beaches (historically correct, but not if it kills the AI)

Anyway, something to look at that might solve the AI thing.

bc

[Oleg Mastruko]

Two notes:

a) Subs were not sunk in harbor (except few). When sunk in harbor ship is listed as "sunk at Manila", when sunk on open sea ship is listed as "sunk NEAR Manila". Note that most subs are sunk NEAR some place or other.

b) I have a pretty strong opinon that every AI vs AI test should be done on "historical" AI. Hard and very hard do not improve AI "per se", they just give additional die roll modifiers (is that the correct term?) to the AI, and are aimed primarily for play vs. human.

If you test AI vs AI, with both sides having significant help in die roll percentages, you'll get "loaded dice" for both sides, and thus whacky results.

BTW WAW works similarly - higher AI levels do not mean "smarter AI", just more supply/combat/transport/die roll/you name it benefits for the AI. It's OK to test vs human, but in AI vs AI games all sides should be made equal (no benefits, no loaded dice, no supply help etc.)

O.

[Tanaka]

ORIGINAL: Don Bowen


(Sigh)

Well, it is obvious that CHS has confused the AI.

In last night's test I slept very well and the game ran until September (1942). The Japanese completed the initial landings in the Philippines and then just sat there. Supply seems to be a concern. Almost all of Malaya was taken and the Japanese were on the doorstep of Singapore, and they had advanced to the ourskirts of Rangoon. None of the follow-on invasions took place: Lingayen, Davao, Kuching, Rabaul. Also, there appears to have been no use of heavy units after the pre-planned actions. ALL Carriers, Battleships, and Cruisers were sitting in port except for the Kota Bharu bombardment ships which were still at Kota Bharu. Submarine results were about the same as the previous test - dozens of allied subs lost to aircraft, no Japanese subs lost.

I'm going to mull over options but it appears that a "Not suitable for AI's under 13 Years Old" tag may be appropriate to CHS. I am depressed.

Don I am running an AI vs AI with the new CHS u just put out with 1.5 and the Japanese AI is doing well...

The AI has captured many of the bases it did not in your scenario and is attacking Kutching and Rabaul but failed to take Wake. it is only Dec 16...

The allies have only lost 2 subs and pearl is still well supplied....

Just to alleviate some of your worries: