Posts: 300
Joined: 9/6/2002 From: Milwaukee, WI, USA Status: offline
So here is the situation, I'm playing as allies vs. the AI in the GC. It's March 1942 and I have a surface combat TF in the New Caledonia area. What happened over the next 10 turns was absolutely absurd and even with a flawed AI this should never happen. The AI attempted to land troops six different times at Luganville, Efate and Koumac. Each time my TF went into action and sank ALL the ships killing ALL the LCU's. The Japs. sent no DD, CL, CA to defend the helpless TF's and no Jap. carriers were within a thousand miles. They were all in the DEI area as far as I could tell. Well it was nice to easily destroying the Jap. TF's but it was way to easy. This should not happen. Why would the AI keep sending unprotected TF's to their deaths. This makes playing the AI totally unfulfilling. I didn't expect the second coming of Isoroku Yamamoto but this is ridicules. Well that's my rant and I now know it's time for PBEM.
< Message edited by sdhundt -- 11/8/2010 11:53:37 PM >
But I will say that when I notice that the AI gets into some wierd loop like that I just back off, let it go, and " work with it ". I have to play the AI atm because of my work schedule, but the vast majority of the time it is still a lot of fun. Just have to give it a break now and again.
_____________________________
" Gentlemen, you can't fight in here! This is the War Room. " President Muffley
Posts: 4776
Joined: 1/26/2003 From: Vienna, Austria Status: offline
Time for PBEM.
And Cap Mandrake couldn´t have said it better. The AIs beating chessmasters were able to compute close to every possible situation in advance by relying on the few different properties of the meeples and the simple layout of the board.
Now take the different properties of a knight and compare them to the computations neccesary for a TF of your choice in WitP AE. And thats only one small aspect where Chess and WitP have anything in common.
Re-reading the above I wonder how the AI is able to do anything at all.
< Message edited by LoBaron -- 11/9/2010 6:53:02 AM >
ORIGINAL: sdhundt So here is the situation, I'm playing as allies vs. the AI in the GC. It's March 1942 and I have a surface combat TF in the New Caledonia area. What happened over the next 10 turns was absolutely absurd and even with a flawed AI this should never happen. The AI attempted to land troops six different times at Luganville, Efate and Koumac. Each time my TF went into action and sank ALL the ships killing ALL the LCU's. The Japs. sent no DD, CL, CA to defend the helpless TF's and no Jap. carriers were within a thousand miles. They were all in the DEI area as far as I could tell. Well it was nice to easily destroying the Jap. TF's but it was way to easy. This should not happen. Why would the AI keep sending unprotected TF's to their deaths. This makes playing the AI totally unfulfilling. I didn't expect the second coming of Isoroku Yamamoto but this is ridicules. Well that's my rant and I now know it's time for PBEM.
Apparently AI scripts lack a function that "recons" player assets at a target area at first place, and adjusts force assignments dynamically based on the opposition (and availability of excess forces from other scripts or unused assets), or does prevent the script from starting of the enemy exceeds some threshold strength. That might need AI to be able to look through FOW to a certain degree to work well -- alternatives to that would be to cumbersome (sort of "a memory" of spotting and movements).
Respectively, a function that would reinforce a script with additional units withdrawn from other scripts that detect less force needs, or quits the script would have prevented maybe everything but the first blunder in the series.
The devs have stated that for now the AI is as good as it can be with the present engine. I still hope for an AE patch that enhances the scripting capabilities by new features, but likely WITP2 might be more realistic hope. AE obviously has reached a huge complexity and the risk of any modification to the AI screwing up how it works is probably high.
The alternative is to roll back a few turns and move your surface TF elsewhere...
< Message edited by janh -- 11/9/2010 10:10:31 AM >
So here is the situation, I'm playing as allies vs. the AI in the GC. It's March 1942 and I have a surface combat TF in the New Caledonia area. What happened over the next 10 turns was absolutely absurd and even with a flawed AI this should never happen. The AI attempted to land troops six different times at Luganville, Efate and Koumac. Each time my TF went into action and sank ALL the ships killing ALL the LCU's. The Japs. sent no DD, CL, CA to defend the helpless TF's and no Jap. carriers were within a thousand miles. They were all in the DEI area as far as I could tell. Well it was nice to easily destroying the Jap. TF's but it was way to easy. This should not happen. Why would the AI keep sending unprotected TF's to their deaths. This makes playing the AI totally unfulfilling. I didn't expect the second coming of Isoroku Yamamoto but this is ridicules. Well that's my rant and I now know it's time for PBEM.
That's your most important lesson. The AI is "script driven". It can't "think" except to look at the situation and try to engage the appropriate "script". But you've played over 100 turns, and during every turn the situation becomes more unpredictable for the script writer (Andy). Just how many "script variants" based on all the possible moves and outcomes in a game this complex could YOU write? With this programming it could take billions of "if/then scripts" just to finish the first year with the kind of competent play you are hoping for.
The AI is primarily a learning tool..., a tireless opponent for you to practice with. Some players, who have no other option, have learned how to coax more extended play out of it by carefully avoiding it's foibles and not exploiting it's weaknesses. But to really experience the potential of the system, you need another human mind on the other side. Someone just as sneaky and goal driven as yourself. A boxer can punch bags and shadows for years..., but until he steps into the ring against a live opponent he will never know if he's any good. It's time to drop the towel and step into the ring. Good luck to you.
Posts: 242
Joined: 10/29/2010 From: Massachusetts, USA Status: offline
So what difficulty level are you playing on? The manual suggests going up to HARD once you're familiar with the game. I won't vouch for the efficacy of this as a solution, but you might want to try it, if you haven't already.
Also,if you want a game that centers on the Southwest Pacific and doesn't seem to have the AI do anything too stupid as the Japanese, you might give Uncommon Valor a look.
So what difficulty level are you playing on? The manual suggests going up to HARD once you're familiar with the game. I won't vouch for the efficacy of this as a solution, but you might want to try it, if you haven't already.
Also,if you want a game that centers on the Southwest Pacific and doesn't seem to have the AI do anything too stupid as the Japanese, you might give Uncommon Valor a look.
Posts: 300
Joined: 9/6/2002 From: Milwaukee, WI, USA Status: offline
I also think the game was MUCH harder in Witp, specifically when you started the game at the Guadalcanal scenario and then went to the end of the war. The troops were already in the Historically correct positions so the AI couldn't screw up by not placing them correctly. To bad the awesome job that was done on this game didn't transfer over to a GC scenario that starts right before the Coral Sea battle.
Here I am commenting in the forum when I can't even play the game at this point. (changed computers--can't get the P-something file to be found).
My impression is that if one plays the computer, playing Japan is best. It is hard for the AI to be too incompetent as the Allies in 1942. And, later in the war, the mass of Allied units can be overwhelming, whatever strategy.
On the other hand, my impression is that the Japanese can continue to build, as the AI, enormous numbers of aircraft even late in the war. So, if you continue, perhaps you will find, though a bit unsatisfactory, a long road ahead
_____________________________
Number one principle: The inherent worth and dignity of all people.
Posts: 496
Joined: 3/11/2008 From: ummmm... i HATE that question! Status: offline
quote:
ORIGINAL: Capt Hornblower
So what difficulty level are you playing on? The manual suggests going up to HARD once you're familiar with the game. I won't vouch for the efficacy of this as a solution, but you might want to try it, if you haven't already.
Also,if you want a game that centers on the Southwest Pacific and doesn't seem to have the AI do anything too stupid as the Japanese, you might give Uncommon Valor a look.
Forget hard. Play 'Ironman' scenario. Then Japan gets all the escort ships they ever dreamed of, plus a few. Far less unescorted convoys to pummel. And of course the AI still gets its unlimited supply of aircraft as soon as a new type is available.
Actually, if you want to play PBEM, AND want to play 'ironman', give me a call. I'll be Japan.....
Why I always argued to some players, play a campaign right out of the box, as an Allied player you focus on defensive strategy for a year while building your pilots pool and learning to get a supply chain moving. As a Japanese player its a little more intense, there will be times you have to restart to learn the industrial side more effectively, (restart due to idiotcy like me who expanded industry to fast etc)
as Japanese the main goal = defend your empire, and keep your industrial going. as Allies the main goal = rebuild a supply line to Australia/New Zealand/India/Pearl Harbor - and wait until new Carriers arrive.
As a result you will many hours drawn into the game, going into months - many beers will be drank and many wives will secretly want AndyMac to disappear.
_____________________________
ASUS Maximus IV Extreme-Z Intel Core I7 2800k Corsair Hydro Heatsink Corsair Vengeance DD3 24GB EVGA GTX 580 Western Digital 1.5TB Raid 0 Windows 7