AI Disappointment (Full Version)

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



Message

Rogue187 -> AI Disappointment (2/4/2018 5:43:52 PM)

I have restarted my game after my first attempt last week and got a major victory by winning the Battle of Midway 2 weeks after the attack on Pearl Harbor. But since then, the AI keeps sending its carriers against Midway. A couple of weeks ago (in game) there was a single carrier that sortied against Midway, but my carriers were too beaten after doing battle with 3 carriers to be of any benefit. Now, it looks like the last 3 carriers of the KB are coming for Midway again. This is really frustrating. Not because its ahstorical, but because its a waste. Right now I have TF Z as well as a few CA/CL TFs running round the DEI blowing up unescorted invasion fleets. Heck, just today, I sank an invasion fleet with a report of 20K ground casualties. While this would be cause for celebration, it almost seems too easy. Where are the Japanese surface ships? Where are the carriers? The KB should be in the DEI and/or South Pacific supporting the advance of invasion fleets, not wasting time bombing Midway. I don't know if this is a result of me holding Wake, or just the fact that there are 20 random AI scrips and I got one script that has a focus on Midway. Maybe I need to be more patient, but I am thinking about starting one of AndyMac's May 42 campaigns just to give the AI a better chance by starting the game at the height of Japanese expansion.



rustysi -> RE: AI Disappointment (2/4/2018 6:09:44 PM)

Unfortunately the AI is the AI. I've yet to play a game where the AI is 'king', so to speak, but once you understand how it 'plays' you'll get a pretty good game from it. There're various things you can do to facilitate this. Many have spoken of this here before. For me, in the early game its easiest to just let it do its thing. You can destroy it later.[:D]

Remember, there're only so many resources that may be put into a game.

quote:

thinking about starting one of AndyMac's May 42 campaigns


Pretty sure Andy just rewrote many of the scripts. Keep in mind that IIRC even he said there wasn't much he could do with Japan in scenario 1. She just doesn't have that much with which to work. If that's what you're playing, may I suggest scenario 2, it gives Japan more.



Rogue187 -> RE: AI Disappointment (2/4/2018 6:23:18 PM)

I've been doing the December 8 campaign, scenario 6. I loaded up the new Dec 7 starts and found the PH attack to be way more devastating than in real life. I wanted the "Nimitz" experience by walking into PH as it was after the attack. Maybe when I get better and am willing to spend another 10 hours setting up turn 1, then I will try with a more effective December 7th surprise.



jwolf -> RE: AI Disappointment (2/4/2018 6:40:35 PM)

Note that you still have Force Z intact, which is of course way unhistorical. Against a human player that is fine, but against the AI then IMHO it's better to go with a historical Dec 7 start and that will (usually) take care of Force Z. The AI's attack against PH will always take place as ordered, but results can vary depending on the rolls of the dice.



adarbrauner -> RE: AI Disappointment (2/4/2018 7:06:30 PM)

You know the possible answers:

1st, play against a human

2nd, shift to Andy Macs Ironman hard, very hard etc.

3rd, don't punish too much the AI; what Rusty said above;

finally:ask Andy (or other suitable) to custom modify the AI in your game for added challenge; he wrote recently he's willing



cardas -> RE: AI Disappointment (2/4/2018 9:14:29 PM)

As the other posters have said the unfortunate truth is that the AI just isn't very smart. It has many bad habits that can easily be abused, even unintentionally. The options adarbrauner gave you are pretty much it, though one more option is to simply see how badly and quickly you can defeat the AI making use of all its faults. I don't think making a reasonable AI would be impossible because the game is somehow too complex however, I just think it wasn't a high priority for the development team.

I'll add one thing though if you are playing the AI: never load a savegame because you did something stupid. Live with it



Aurorus -> RE: AI Disappointment (2/4/2018 9:17:29 PM)

The AI tends to play better on defense. The way the AI works is that it is set objectives within timeframes. So long as the game-date remains in these timeframes, it will continue to send assets against that objective, even if it does not have sufficient or appropirate assets to secure the objective. This often results in the AI's forces being destroyed piecemeal. As others mentioned, it is best, when playing as the allies, to avoid being overly aggressive with the AI during the Japanese expansion phase- until June 1942 or so.

I recommened PBEM. These games are so much more enjoyable with the right opponent. You may find it frustrating to play against a very experienced player, though as the allies, there is time to learn and early mistakes will not doom you. Post a request seeking an inexperienced opponent. Oftentimes, these PBEMs with inexperienced players, do not last more than 4 or 5 months. That is OK. You will gain valuable experience for when you do find a reliable opponent to complete the whole war. I am happy, actually, that most of my PBEM opponents did not complete the game. I learned new things each time and am now able to play against very good veteran players, such as Apbarog, who is committed to completing the game.



adarbrauner -> RE: AI Disappointment (2/4/2018 9:25:13 PM)


quote:

ORIGINAL: Aurorus

The AI tends to play better on defense. The way the AI works is that it is set objectives within timeframes. So long as the game-date remains in these timeframes, it will continue to send assets against that objective, even if it does not have sufficient or appropirate assets to secure the objective. This often results in the AI's forces being destroyed piecemeal. As others mentioned, it is best, when playing as the allies, to avoid being overly aggressive with the AI during the Japanese expansion phase- until June 1942 or so.

I recommened PBEM. These games are so much more enjoyable with the right opponent. You may find it frustrating to play against a very experienced player, though as the allies, there is time to learn and early mistakes will not doom you. Post a request seeking an inexperienced opponent. Oftentimes, these PBEMs with inexperienced players, do not last more than 4 or 5 months. That is OK. You will gain valuable experience for when you do find a reliable opponent to complete the whole war. I am happy, actually, that most of my PBEM opponents did not complete the game. I learned new things each time and am now able to play against very good veteran players, such as Apbarog, who is committed to completing the game.


Hey Aurorus, Bobo is looking for to restart the contest, did you notice?



Aurorus -> RE: AI Disappointment (2/4/2018 9:36:37 PM)

I have communicated with Bob privately. Unfortunately, I have 2 games at the moment, which is the absolute most that I can do, so I cannot resume the game with Bob at the moment. Bob felt he was doing very poorly in that game. I think he can be too hard on himself and overly critical of his own play. Bob, if you read this, you were doing fine in that game... lol.



crsutton -> RE: AI Disappointment (2/5/2018 5:04:21 AM)

I practiced against the AI for a couple of weeks when the game came out ten years ago but started playing with human opponents and never looked back since. Find yourself an opponent that is around your skill level and go at it. You won't regret it.



Chris21wen -> RE: AI Disappointment (2/5/2018 5:49:28 AM)


quote:

ORIGINAL: larrybush

Did you put the Japanese to Hard or Very Hard? Did you update with the updated AI files by Andy Mac?



Putting the AI on hard or vary hard does not prevent the AI from continually attacking the same target nor does changing to Andy Mac updated files although it's much, much better.

The only way is to let the AI have it's target. If you have TFs in the area pull back once the initial combat has taken place or select a pure defensive mode. For air ops do not set any Naval maissions.



Timotheus -> RE: AI Disappointment (2/5/2018 12:36:09 PM)

Play Andy's scenarios. He is the AI guru and has improved the scenarios considerably. For both sides.

In fact, he is STILL WORKING NOW to improve them. Kudos to him, the man is underappreciated for everything he has done and is doing for the community.
Also, have you installed the patch to modernize the game?

Here is the link to Andy's scenarios (and other game assets by other users also):
http://www.matrixgames.com/forums/tm.asp?m=3088504



Rogue187 -> RE: AI Disappointment (2/5/2018 1:26:52 PM)

I do have everything patched up to the latest patches. Don't get me wrong I don't blame AndyMac and I don't mind things not being 100% historical. I kept TF Z alive. Cool! Good for me! Now it up to the Japanese to hunt it down before it disrupts their invasion plans. I think I should see BB/CA TFs, CV TF, everything, trying to hunt down that force. And if that is too much, since the Allies have little Naval power to stop such a push, having strong surface TFs escorting their invasion fleets for their protection would be something I would expect. In the early war, the Allies are at the disadvantage. Poor Naval strength, obsolete planes, poor morale/leadership, etc. When I stop an invasion force, it should feel like a huge victory. If I delay a LCU and hold on to land, it should be a victory. But its not fun to watch every invasion succeed even though 3/4 of the fleet was sunk and my ships ran out of ammo. Then to see the KB divided (which is ok) but bombing Midway? Its not important at this point! I could see parking the KB off Wake to cover a (2nd, 3rd, 4th...whatever) invasion attempt and to draw out the American carriers, but to abandon the DEI to a strong surface TF and not attempt to engage it just seems disappointing.



Lecivius -> RE: AI Disappointment (2/5/2018 2:02:35 PM)

PLay Andy's improved AI games. The first 2 years will keep you entertained [;)]



Mobeer -> RE: AI Disappointment (2/5/2018 9:43:09 PM)

At least in the original, I never found the AI scripts to be a weakness, but rather how the AI at a lower level goes about implementing those scripts. For example in 2 games:

1) Japanese (me):
- take Port Moresby, garrison
Allies (AI):
- builds up Thursday Island
- takes dominance in skies over Coral Sea
- clears out IJN ships
- invades Port Moresby successfully

2) Japanese (me):
- take Port Moresby, garrison
- take all other Papua NG bases, leave empty to act as a sort of trip wire warning of Allied intent
Allies (AI):
- builds up Australia
- tries to run transports piecemeal to Port Moresby
- takes terrible losses in surface combat
- get handful of troops ashore after 2 years of trying
- never takes the base

In each case the AI high level plan was basically the same, and also a good plan.

I have seen similar behaviour elsewhere at other bases - good plans badly executed. I don't see that improving the scripts is the real improvement required.



Yaab -> RE: AI Disappointment (2/6/2018 5:06:48 PM)


quote:

ORIGINAL: Rogue187

I have restarted my game after my first attempt last week and got a major victory by winning the Battle of Midway 2 weeks after the attack on Pearl Harbor. But since then, the AI keeps sending its carriers against Midway. A couple of weeks ago (in game) there was a single carrier that sortied against Midway, but my carriers were too beaten after doing battle with 3 carriers to be of any benefit. Now, it looks like the last 3 carriers of the KB are coming for Midway again. This is really frustrating. Not because its ahstorical, but because its a waste. Right now I have TF Z as well as a few CA/CL TFs running round the DEI blowing up unescorted invasion fleets. Heck, just today, I sank an invasion fleet with a report of 20K ground casualties. While this would be cause for celebration, it almost seems too easy. Where are the Japanese surface ships? Where are the carriers? The KB should be in the DEI and/or South Pacific supporting the advance of invasion fleets, not wasting time bombing Midway. I don't know if this is a result of me holding Wake, or just the fact that there are 20 random AI scrips and I got one script that has a focus on Midway. Maybe I need to be more patient, but I am thinking about starting one of AndyMac's May 42 campaigns just to give the AI a better chance by starting the game at the height of Japanese expansion.


Stop using game's universal supply and try to reload your USN ships with US-made supply only. For CA Houston it is either Manila or an AK to reload her main guns. See the difference? No Sorebaja, no Darwin for you. Your CA/CL TF is now a one-shot weapon. Create house-rules for yourself to limit powergaming. No fuel in xAKs, no level bombers bombing LCUs in hexes where combat takes place, no replacement aircraft for AVG as in RL etc. AI is as lost as a drunken Kempetai in LA and can't compete.



Rogue187 -> RE: AI Disappointment (2/6/2018 11:30:03 PM)

It occurred to me today that with all of these in house rules, recommendations to not be too aggressive with the AI, etc, what is the point of playing the game before May 1942? If the AI isn't good enough to react to the player and you have to make so many sacrifices in order to give it a fighting chance, why play any game against the AI from Dec 7th onward?



Aurorus -> RE: AI Disappointment (2/7/2018 12:14:25 AM)

quote:

ORIGINAL: Rogue187

It occurred to me today that with all of these in house rules, recommendations to not be too aggressive with the AI, etc, what is the point of playing the game before May 1942? If the AI isn't good enough to react to the player and you have to make so many sacrifices in order to give it a fighting chance, why play any game against the AI from Dec 7th onward?



Much of this game is logistics, and you want to gain experience preparing the supply and fuel path and depots from the U.S. to the various fronts. There are logistics bottleheads that will cause you serious problems later and slow you eventual advance if you do not address these issues in the early war period. For example, supply flows poorly to Darwin, and if you wish to use Darwin as a depot for a DEI invasion, you will need to develop the bases leading to Darwin. Much the same can be said about the route through SoPac to Australia. You also want to control your pilot training program and develop your pilots to your preferences. Much of the allied advantage derives from their enormous supply and fuel reserves. You will have assets sufficient to overwhelm Japan from 1944 onward, but if you do not have the fuel and supply built up in the correct places, you will be unable to use much of this vast armada.

Playing through a few times will also make you familiar with the allied force pool, and the problems that can ensue later if you start off poorly. While the U.S. force pool is vast, the same is not true for the Australian, British, and Indian force pools. So you want some idea of how to slow the Japanese without losing assets that cannot be replaced later.



Macclan5 -> RE: AI Disappointment (2/7/2018 12:44:19 PM)

Disclaimer: I think I may be deemed an apologist for the AI - so be it. [8D]

The AI as I recall randomly selects one of 12 different aggressiveness scripts at game start. Having played through 3.5 games I have been impressed with the variations I have experienced.

(The Midway variant was indeed one of my experiences as well; A Normantown invasion was another; a deep push into Canton Island and Fiji was a third)

But..

1) Most players (Allied or Japanese) have the benefit of almost perfect historical hindsight.
We bought the game because of our interest in the history of the subject. That translates to the tendency to rush troops, supplies, logistic support to key bases you deem critical to your long term objectives. The Allies did not have such perfect foresight in 1942.

2) What is your real objective ?
What is wrong with the pure objective of massively defeating the AI given the above ?
One experienced veteran poster responded to me once - his objective was to win a decisive victory by December 1942 (Victory Points Ratio). I inquired if this included the invasion of Tokyo - it did not; but it is his preference - great if he so enjoys.
I define my game objectives as the capture of Tokyo. I play on despite achieving the required ratio of VP points. I do find the AI very capable in defense through 1944.
I tend to disagree with "allowing the AI" early wins - but that is my preferred play style (*see 4 below).

3) House-rules.
Many good ones mentioned above.
Only ship fuel in tankers
Only rearm ships where an AKE or AK is present
Obey or adhere to stacking limits. Use DaBabes mod that imposes much harsher stacking penalties
One I personally think should be more often implemented : Off map ship movements of supply and fuel REQUIRE an escort; at least 1 for every 10 ships. Minimum would be a Armed Merchant Cruiser (AMC) or DD / DE. Rationale - this reflects the German menace of Uboats / German Raiders in the Atlantic through 1943. Your would be surprised how precious those escorts are in the early days and how long it takes to assemble critical mass to ship "off map" when you want to optimize i.e. Eastern USA to Capetown supply routes.

4) No re-load challenge.
The AI never gets hurt feelings if you reload a turn. I get it.
However depending on your personal objectives - don't reload. Make a house rule.
Or limit your reloads - 1 'mulligan' per year maximum.
The game is incredibly detail rich. They ability to pump out 'one perfect turn after another' without some 'brain dead settings'; further without the benefit of a reload adds another layer of complexity to playing against the AI. You will suffer reversals.
Nimitz never had a 'mulligan' [8D] He took calculated risks. Emulate Nimitz. Don't reload past turns.

Bottom line - a game against the AI is what you choose to make it.

A game against a human opponent may indeed be your best ultimate solution; and may be humbling as well [8D]




RichardAckermann -> RE: AI Disappointment (2/7/2018 1:29:57 PM)

I do partially agree. The AI does a fine job most of the time. But there are weak spots. The AI does not respond well to threats or kill zones. I can win the game as japan by using almost only subs alone. Another time I captured and held the solomons, which became a happy target shooting ground for my aircraft, as mainly unprotected allied AKs tried to invade there for months.
I think a good AI should challenge you without having house rules to play "dumb". If creating a good AI on a game this scale is possible is another question.



adarbrauner -> RE: AI Disappointment (2/8/2018 7:37:46 PM)

A game-winning tip: play 3 days turn;

the AI cheats; he can modify missions at the start (!!!) of his turn, maybe even at the start of each night-day impulse, everyday, while we are bound to the orders we gave;

please, if you try this remember to tell how many ships AI's Wildebeests have torpedoed, if you play Japan historical, in first two-three turns (remember please to keep a clean language in this eventuality) and how fast have lost your carriers, if you play Allied (well, the same for Japan if you are not hyper careful);



Yaab -> RE: AI Disappointment (2/9/2018 3:48:33 PM)


quote:

ORIGINAL: Rogue187

I have restarted my game after my first attempt last week and got a major victory by winning the Battle of Midway 2 weeks after the attack on Pearl Harbor. But since then, the AI keeps sending its carriers against Midway. A couple of weeks ago (in game) there was a single carrier that sortied against Midway, but my carriers were too beaten after doing battle with 3 carriers to be of any benefit. Now, it looks like the last 3 carriers of the KB are coming for Midway again. This is really frustrating. Not because its ahstorical, but because its a waste. Right now I have TF Z as well as a few CA/CL TFs running round the DEI blowing up unescorted invasion fleets. Heck, just today, I sank an invasion fleet with a report of 20K ground casualties. While this would be cause for celebration, it almost seems too easy. Where are the Japanese surface ships? Where are the carriers? The KB should be in the DEI and/or South Pacific supporting the advance of invasion fleets, not wasting time bombing Midway. I don't know if this is a result of me holding Wake, or just the fact that there are 20 random AI scrips and I got one script that has a focus on Midway. Maybe I need to be more patient, but I am thinking about starting one of AndyMac's May 42 campaigns just to give the AI a better chance by starting the game at the height of Japanese expansion.


Have you tried updating the AI and deleting the non-historical scenario variants?
http://www.matrixgames.com/forums/tm.asp?m=4257473



Rogue187 -> RE: AI Disappointment (2/9/2018 4:13:24 PM)


quote:

ORIGINAL: Yaab


quote:

ORIGINAL: Rogue187

I have restarted my game after my first attempt last week and got a major victory by winning the Battle of Midway 2 weeks after the attack on Pearl Harbor. But since then, the AI keeps sending its carriers against Midway. A couple of weeks ago (in game) there was a single carrier that sortied against Midway, but my carriers were too beaten after doing battle with 3 carriers to be of any benefit. Now, it looks like the last 3 carriers of the KB are coming for Midway again. This is really frustrating. Not because its ahstorical, but because its a waste. Right now I have TF Z as well as a few CA/CL TFs running round the DEI blowing up unescorted invasion fleets. Heck, just today, I sank an invasion fleet with a report of 20K ground casualties. While this would be cause for celebration, it almost seems too easy. Where are the Japanese surface ships? Where are the carriers? The KB should be in the DEI and/or South Pacific supporting the advance of invasion fleets, not wasting time bombing Midway. I don't know if this is a result of me holding Wake, or just the fact that there are 20 random AI scrips and I got one script that has a focus on Midway. Maybe I need to be more patient, but I am thinking about starting one of AndyMac's May 42 campaigns just to give the AI a better chance by starting the game at the height of Japanese expansion.


Have you tried updating the AI and deleting the non-historical scenario variants?
http://www.matrixgames.com/forums/tm.asp?m=4257473



Yes, I am playing the December 8th start. I know it has 20 AI script variations and the one I got seems to have a KB focus on Midway. After playing a bit, I can now see how powerful TFZ is without it being actively hunted by the Japanese. There are just so many juicy, unescorted invasion fleets running around that would be excellent target practice. My preference would be the historical Japanese moves for about the first 6 months, then open ended after that. That way my goal would be to slow the Japanese advance until better equipment comes on line. After that, it can do anything it wants.

I really need to do a May 1942 start to fight the war I want to play. I am just having trouble spending ANOTHER 10 hours doing set up again!



rustysi -> RE: AI Disappointment (2/9/2018 6:04:08 PM)


quote:

ORIGINAL: Rogue187

It occurred to me today that with all of these in house rules, recommendations to not be too aggressive with the AI, etc, what is the point of playing the game before May 1942? If the AI isn't good enough to react to the player and you have to make so many sacrifices in order to give it a fighting chance, why play any game against the AI from Dec 7th onward?


You will still learn many things just by playing. For instance, if you play as in a PBEM, you'll learn where your SLOC's should go. How many troops you need here or there, and how and where to get them from. What bases to build to what levels. How to get resources to said bases and from where. You'll see how your pilot training progresses and what adjustments to make. I can't even list all that you'll learn just by playing, and you'll be less frantic playing the AI. BTW that's how the Allies should play as they really have little to fear in the long run. My biggest goal as a Japanese player will be to get the Allies to react before they're ready, its the only way I can win. Beat the Allies piece-meal.

If nothing else, as long as you keep certain things in mind, the AI is a good learning tool. Heck I've been 'learning' for about five years now.[:D]




Page: [1]

Valid CSS!




Forum Software © ASPPlayground.NET Advanced Edition 2.4.5 ANSI
1.859375