I need Help - AI Testing (Full Version)

All Forums >> [New Releases from Matrix Games] >> War in the Pacific: Admiral's Edition >> Scenario Design and Modding



Message

Andy Mac -> I need Help - AI Testing (3/5/2016 9:34:22 PM)

I think I have articulated a few times that I am working on a new revised base Ai to try and reduce the waste in the AI or at least minimise it.

The background was the old WITP AI was much maligned as being too easy too predictable not worth playing too passive etc etc.

I don't actually agree with that what they achieved with what they had for a game of this compeity was spectacular

When we went into developing the AI for AE we tried to make it more dynamic but in essence because we couldn't get it to work
ended up reverting to a scripted AI which in essence has meant everyone has been playing me via proxy of the Ai for a number of years !!!

To try to get a bit of shock and awe in the AE AI I realised quite quickly that given time a player can set up roadblocks in bases that will stop the AI
taking "must take" bases

So as we got closer and closer to Gold the AI kept getting more and more aggressive as I ramped up the pressure and raided HI and other theatres for resources to
try and keep up the skeer while the Allies were weak.

For about 6 months after release all those years ago this worked and worked well - we are all used to the AI now and probably don't remember the initial shock when the AI
went back to PH for a 2nd strike - or hit PM in Dec 41 - or tried the Mersing Gambit or or or

However there was a price to the aggression

As players learned the game some or most of those early AI "lunges" and not fully supported invasions or the AI's tendency not to provide sufficient escorts made them horribly vulnerable

So after 6 months the Ai became a punchbag in Scen 1 and Scen 2

So the began my Ironman series where I give the AI enough resources to kaje a hit and keep swinging.

Anyway the reason for going through all this is that I have been working on a new baseline AI script I only have one variant so far and I need some folks to try it out while I keep testing

So I am attaching three scenarios to this thread

A New Scen 1
A new Scen 2

and a
new Tier 1 Ironman (i.e. Scen 10 equivalent where the AI does not get out right free stuff ie hidden supply but it does get large amounts of extra on map stuff and a much stronger starting
economy.

They all have the same AI script at present so I don't mind which scen people play but I need feedbhack to improve the script and KI simply don't have the time anymore to do it all myself.

So this is a plea for help

I will post the scenarios next

I need help on either assessing the scens so PBEM players telling whether especially Scen 2 is out of wack and the impact of the new bases

or

AI player telling me how the Ai plays out

Scen 1 (set to Scen 100) is in this post




Andy Mac -> RE: I need Help - AI Testing (3/5/2016 9:34:57 PM)

Scen 2 update (Scen 101)




Andy Mac -> RE: I need Help - AI Testing (3/5/2016 9:35:33 PM)

Scen 10 (Scen 102)




Andy Mac -> RE: I need Help - AI Testing (3/5/2016 9:37:25 PM)

Thanks to everyone who helps with this



Andy Mac -> RE: I need Help - AI Testing (3/5/2016 9:44:29 PM)

ps not Extended Map standard map only at present



JeffroK -> RE: I need Help - AI Testing (3/5/2016 9:56:18 PM)

Gday Andy.

I'll give them a bash, sick of being chased around the Pacific in Ironman 3.

My approach will be a somewhat historical one, defending on the PI, Malaya, Burma & China (and defending Tahiti!) (if thats where the AI attacks)

I'll try and keep notes.



PaxMondo -> RE: I need Help - AI Testing (3/5/2016 10:08:51 PM)


quote:

ORIGINAL: Andy Mac

Scen 10 (Scen 102)



This is the straight Ironman scenario (IJ AI, player = Allies) correct?



Andy Mac -> RE: I need Help - AI Testing (3/5/2016 10:17:53 PM)

Thanks guys yes scen 10 is 1st tier ironman v Allied



JeffroK -> RE: I need Help - AI Testing (3/5/2016 10:30:26 PM)

I've powered up scen 100.

AI=Hard, Historical first turn.

2 IJN minisubs penetrate Pearl and make hits on BB, its very rare to get any minisub hits.
KB manages hits on some 30 ships at PH, 3 DD are sunk but the damage is spread around.

2BB have over 70flot, 1 other 51flot
3 good CL have around 50flot and a number of other surface combat ships have 10-20flot.
Really a good strike, the BB fleet is heavily damaged and 3/4 of the modern CL have been damaged enough to spend a lot of time in port.

Force Z copped 9 torp hits, both sunk.

Most of the remaining airstrikes seem normal, only possible oddity was the LBA flying from the Saigon area attacked Mersing rather than Singapore on their Day 1 strikes?



btd64 -> RE: I need Help - AI Testing (3/6/2016 12:10:03 AM)

Trying 102. I'll let you know....GP



Andy Mac -> RE: I need Help - AI Testing (3/6/2016 11:22:57 AM)

Thanks guys I will check out Saigon issue



Andy Mac -> RE: I need Help - AI Testing (3/6/2016 3:47:19 PM)

Jeffk also check out the revised Australian locations look better ?



Andy Mac -> RE: I need Help - AI Testing (3/6/2016 5:40:22 PM)

Just remember even 102 is Ironman |Tier 1 ie not nasty or nasty nasty - I am trying build up and get the Ai working better 1st



JeffroK -> RE: I need Help - AI Testing (3/6/2016 7:42:12 PM)


quote:

ORIGINAL: Andy Mac

Jeffk also check out the revised Australian locations look better ?

Got stuck pretending to be a plumber, I'll have a look tonight.



PaxMondo -> RE: I need Help - AI Testing (3/7/2016 2:10:34 AM)


quote:

ORIGINAL: Andy Mac

Just remember even 102 is Ironman |Tier 1 ie not nasty or nasty nasty - I am trying build up and get the Ai working better 1st

Playing 102 ... up to Dec 20th or so. AI is doing fine so far. All advances are well supported. nothing stupid so far.



JeffroK -> RE: I need Help - AI Testing (3/7/2016 6:54:39 AM)

Andy

1st Motor Bde is shown at Cairns, its units were at Gatton, a western suburb of Brisbane.

4, 10 & 15 Bdes are at Puckapunyal, they should be at Bonegilla which is just outside of Wodonga (base 1657)

Units of 13 Bde in Western Australia should be in Perth, 10 LH is OK at Geraldton.

Except for 1 Mot Bde only a few hexes.




JeffroK -> RE: I need Help - AI Testing (3/7/2016 9:08:16 AM)

Dec 11

3 days of PH attacks, must be an AI record[&o]

Not a lot of extra damage but my fighters took more beatings, everything else normal.



Andy Mac -> RE: I need Help - AI Testing (3/7/2016 12:47:19 PM)

Good thanks Pax

quote:

ORIGINAL: PaxMondo


quote:

ORIGINAL: Andy Mac

Just remember even 102 is Ironman |Tier 1 ie not nasty or nasty nasty - I am trying build up and get the Ai working better 1st

Playing 102 ... up to Dec 20th or so. AI is doing fine so far. All advances are well supported. nothing stupid so far.




Andy Mac -> RE: I need Help - AI Testing (3/7/2016 12:47:41 PM)

OK will update are new bases ok ??

quote:

ORIGINAL: JeffK

Andy

1st Motor Bde is shown at Cairns, its units were at Gatton, a western suburb of Brisbane.

4, 10 & 15 Bdes are at Puckapunyal, they should be at Bonegilla which is just outside of Wodonga (base 1657)

Units of 13 Bde in Western Australia should be in Perth, 10 LH is OK at Geraldton.

Except for 1 Mot Bde only a few hexes.






JeffroK -> RE: I need Help - AI Testing (3/7/2016 7:36:29 PM)

Yes, both Puckapunyal & Greta were major training bases, you included Wodonga )Bandiana & Bonegilla camps) but 3rd Div (4,10 & 15 Bdes) were based there.



PaxMondo -> RE: I need Help - AI Testing (3/8/2016 3:12:52 AM)


quote:

ORIGINAL: Andy Mac

Good thanks Pax

quote:

ORIGINAL: PaxMondo


quote:

ORIGINAL: Andy Mac

Just remember even 102 is Ironman |Tier 1 ie not nasty or nasty nasty - I am trying build up and get the Ai working better 1st

Playing 102 ... up to Dec 20th or so. AI is doing fine so far. All advances are well supported. nothing stupid so far.



About Jan 1 ... AI still ok. I think biggest issue so far is that the AI is plodding ...
I'm getting everything out of Singers as there is no Mersing and the AI can't field enough Anti-Naval soon enough to stop the Sir Robin.

Now in some respects this is good. The AI will accomplish a major objective (Singers) which will allow it to keep going ... OTOH, I'm getting too much out.
Ditto china. The AI is just moving too slow. It's letting me get into positions there too easy.

Having said these two things, I also understand some of the limitations. My thoughts would be:

1. Not sure Mersing is the best solution here, but for Ironman it means advances in the DEI should be faster. Kuching/Sambas/Palembang timetables should be faster. Ironman AI should have assets to allow that to happen. This would mitigate the slow march down the Malay peninsula as there isn't much we can do about that.
2. China: ditto. I think some of the time tables should be pushed forward. Again, Ironman assets should support a faster time table...

In both cases I'm thinking like move them up a week within the first month ... something like that.




Andy Mac -> RE: I need Help - AI Testing (3/8/2016 2:59:03 PM)

Interesting

I am trying to get a base AI that works without wasting too many assets and covers itself better in the expansion phase

I will then do some variants based off of that solid base which will expand more quickly !!!



Andy Mac -> RE: I need Help - AI Testing (3/8/2016 3:39:30 PM)

Pax did you lose anything at all bugging out of Singapore ??

In AI v AI tests - the AI takes a long time over Singapore because the AI actually reinforces Singers most of the time.

This is why I need live testers AI v AI simply doesnt show everything !!

Thanks

Andy




Andy Mac -> RE: I need Help - AI Testing (3/8/2016 3:39:57 PM)

I think the new bases in Malaya help the AI advance in a more logical fashion as well



PaxMondo -> RE: I need Help - AI Testing (3/9/2016 1:54:57 AM)

quote:

ORIGINAL: Andy Mac

Pax did you lose anything at all bugging out of Singapore ??

In AI v AI tests - the AI takes a long time over Singapore because the AI actually reinforces Singers most of the time.

This is why I need live testers AI v AI simply doesnt show everything !!

Thanks

Andy



One xAK ... that's it. I got every unit out of Singers that can be bought out ... and that xAK was on the last day of the exodus. It was carrying a crappy IND BN.

Again, I chose to bug out rather than fight because I know that Singers is a major script that the AI has to complete. Also, in an Ironman game, not sure I could hold Singers anyway. I haven't looked at the Allied side in the 'base' Ironman in so long that I don't remember what they get.
I do know that in Nasty, Nasty I can't hold it. :)



PaxMondo -> RE: I need Help - AI Testing (3/9/2016 2:13:26 AM)


quote:

ORIGINAL: Andy Mac

I think the new bases in Malaya help the AI advance in a more logical fashion as well

They do, but it is a plodding advance. That's ok really, but just need to find a way to make the Sir Robin more risky. That's why I'm thinking move up the time table for Sinkawang/Sambas/Pontiak. If the NW corner of Borneo is lost, and Netties are present, the Singapore Sir Robin is toast.

Anyway , just some thoughts.


Otherwise, AI is doing well. 1/5/42 Temuloh has fallen as has KL. I figure another 2 weeks before the IJ gets to Singers unopposed.

Hasn't shown up in Burma yet. I'm planning to defend at Katha (+2 terrain) fall back at Myitkyina (+3 terrain). Even against Ironman, I should be able to hold there and in India. I'm strengthening Calcutta/Chittagong and Ceylon (Singers refugees all going to Ceylon).

I haven't peeked at the scripts, but I'm assuming I will need to cede PM. I'm also prepared to give up Darwin and NE OZ to Brisbane. I don't think I will have to give up anything else and I won't give up PErth. That also is getting defenses set.

So far, what I'm seeing tells me all this should be doable and the allies will be on counter attack by 9/42 ... maybe a bit sooner.

I'm mostly curious on how fast you accelerate aircraft models and get them into production. My 9/42 date is assuming I'm still facing Tojo/Oscar not Frank .... [;)]




BTW: In my mod of your Nasty, Nasty variant (I play IJ, AI is Allies), I have really accelerated some models and then I have setup groups to disband where I can replenish AI pools. Aircraft for allies that I increased or modded beyond your Nasty, Nasty point include:
P-39 - re-armed with 20mm Hispano (as the SOV did IRL and I think the P400 was armed as such) so it is much more effective and much better than P40, increased production severely in 1st 6 months and all US Army, Canada, OZ, and NZ fighter groups can upgrade to it.

P-51 series. All arrive 6 - 9 months earlier (P-51A in 9/42). Again as with the P-39 in terms of what units can use it. Production numbers are 100's/month. I figure this is in response to IJ players accelerated RnD ... P51a is a good match against Tojo ... [;)]

This is just a few things ... mostly just spins from what you already have in place ...



Andy Mac -> RE: I need Help - AI Testing (3/9/2016 6:36:23 PM)

Interesting plodding is ok for this one but I may forward deploy some zeroes and netties to make it tougher to evac

Main thing in this one is a solid expansion phase with lowere losses still taking the key bases and then really digging in for a tough defensive battle with resources this time.

If I can get an improved baseline Ai then I may steal a few of those ideas when I get back to tier 2 and 3 ironman scenarios !!!

At the moment concentrating on the key basic 3 scenarios



Andy Mac -> RE: I need Help - AI Testing (3/9/2016 6:37:53 PM)

ps thanks once again JeffK and Pax I know its hard to test v the Ai

I really miss Brady who was my AI tester he played out the 1st 6 months soooo many times !!!



JeffroK -> RE: I need Help - AI Testing (3/9/2016 7:42:24 PM)

Been a bit slow, a couple of incidents at work have left me a bit p.o'd when I get home.

I mentioned the 3 day PH attack, a week later KB is still in range of Midway ans attacking there, might have refuelled?


































DD696 -> RE: I need Help - AI Testing (3/9/2016 9:41:00 PM)

Hi Andy,

I am willing to help. On 23 Feb I restarted a heavily modified version of RA 7.9 and finished turn 135 today, so that is 16 days to do 135 turns. I'm turning out turns slower than normal for a good while now, but today I have completed (hopefully) my chemotherapy sessions and will begin to end this downhill slide.

I tend to play a passive game (Allied player) against the AI at first. I learned long ago if you defeat the Empire's navy around Port Morsby the game gets kind of dull after than. I withdraw as much as possible from Malaya to carry on the war in Burma. There are times when I get a frustrated at the bull headed AI and give them a good pounding and have some fun, then I go back a few turns and let it do what it wants with minimal opposition.

In order to help the AI in my mod I do replace all Japanese land units with the ones from scenario 1 as that is what the AI is expecting to find (as well as add in the additional ships), and I get to play with all the new toys. I beef up HI, LI, oil, fuel and resource production in Japan/Korea/Manchuria in order to make it as self sufficient as possible. I don't like having to play on hard or extra hard in order to boost these resources in the pools. I monitor Japan with tracker (and turn off Japanese garrison requirements) to prevent Japan taking major hits when I have to go in and tinker with something on the Japanese side.

I haven't played any stock scenarios in years and find them rather dull after trying out some of the good mods out there. If you think I would be of any use in your testing kindly let me know.

In my humble opinion, there should be scenarios made specifically for the AI player. This would make for a much better game and mod developers wouldn't be jumping thru hoops trying to have a scenario that works for both.



Page: [1] 2 3 4 5   next >   >>

Valid CSS!




Forum Software © ASPPlayground.NET Advanced Edition 2.4.5 ANSI
1.125