So, the very hard AI is cheating, right?
Moderators: wdolson, MOD_War-in-the-Pacific-Admirals-Edition
So, the very hard AI is cheating, right?
Wondering if I did anything wrong here.
in a two day turn, I committed total 56 Oscar to sweep Moulmein. Half of them come at 20k ft the other half at 25k ft.
The result is very disappointing. I lost 20 fighters while AI only lost 2.
Here is an example
I have numerical advantage, height advantage. But I suffered a 0:4
in a two day turn, I committed total 56 Oscar to sweep Moulmein. Half of them come at 20k ft the other half at 25k ft.
The result is very disappointing. I lost 20 fighters while AI only lost 2.
Here is an example
I have numerical advantage, height advantage. But I suffered a 0:4
Re: So, the very hard AI is cheating, right?
Oscars are not that good and you are slow in taking Burma.
Seek peace but keep your gun handy.
I'm not a complete idiot, some parts are missing!
“Illegitemus non carborundum est (“Don’t let the bastards grind you down”).”
; Julia Child

I'm not a complete idiot, some parts are missing!
“Illegitemus non carborundum est (“Don’t let the bastards grind you down”).”
- Marauder11
- Posts: 62
- Joined: Fri Nov 15, 2019 9:25 am
Re: So, the very hard AI is cheating, right?
In a way yes. The very hard setting makes it harder to defeat the AI units and it makes it easier for the AI units to defeat the players units. Also on very hard the AI units won't run out of supplies.
For every Napoleon there is a Wellington.
-
Chris21wen
- Posts: 7719
- Joined: Thu Jan 17, 2002 10:00 am
- Location: Cottesmore, Rutland
Re: So, the very hard AI is cheating, right?
The AVG units (flying tigers) are some the best air units the Allies have during the early stages, there have been fighting since early 41 in China. easily on par with Japanese army units, plus aircraft (H81/P-40 or Hurri) are much better than the Oscars especially at the height you are using them.
-
Chris21wen
- Posts: 7719
- Joined: Thu Jan 17, 2002 10:00 am
- Location: Cottesmore, Rutland
Re: So, the very hard AI is cheating, right?
A tip on the combat report.
The in game report is a copy of a text file located in the save folder. I use it all the time. It has three advantages of the in game, you can post text and not a pic, it's text so you can edit it and you can read it and move around the game map. Of course the latter works best if you are in a window and/or have more than one screen.
The in game report is a copy of a text file located in the save folder. I use it all the time. It has three advantages of the in game, you can post text and not a pic, it's text so you can edit it and you can read it and move around the game map. Of course the latter works best if you are in a window and/or have more than one screen.
Re: So, the very hard AI is cheating, right?
Tcao, did you watch the combat animations? Who got the "dive" bonus in air combat - the Oscars or Allies?
BTW, the early Oscars have weak MGs. I bet you damaged several Allied aircraft but none were destroyed outright in the air combat. Watching combat animations will also tell you if you land damaging hits on Allied fighters.The Allied fighters you engaged are more durable than Oscars.
BTW, the early Oscars have weak MGs. I bet you damaged several Allied aircraft but none were destroyed outright in the air combat. Watching combat animations will also tell you if you land damaging hits on Allied fighters.The Allied fighters you engaged are more durable than Oscars.
- Platoonist
- Posts: 3042
- Joined: Sun May 11, 2003 4:53 am
- Location: Yoyodyne Propulsion Systems
Re: So, the very hard AI is cheating, right?
There was a Japanese pilot who complained that the early mark Oscars (Ia) with their armament of two rifle-caliber machine guns weren't too far distant from the Sopwith Camel era in some respects. This quickly gets upgraded to two heavy machine guns in the Ic but is still inferior in comparison to Allied fighters carrying at least four. P-40 pilots reported their planes could "take it" from the two machine guns of the Oscar, but if they hit an Oscar in return with all six machine guns it really blew up.Yaab wrote: Sat Apr 06, 2024 6:53 am BTW, the early Oscars have weak MGs. I bet you damaged several Allied aircraft but none were destroyed outright in the air combat. Watching combat animations will also tell you if you land damaging hits on Allied fighters.The Allied fighters you engaged are more durable than Oscars.
The plane was extremely maneuverable yes, but the AVG didn't fight that way. They focused instead on hit-and-run engagements, using their superior speed and firepower to counter the Oscar’s agility.
Re: So, the very hard AI is cheating, right?
I watched the combat animations and that's where I got a strong impression that AI is cheating.Yaab wrote: Sat Apr 06, 2024 6:53 am Tcao, did you watch the combat animations? Who got the "dive" bonus in air combat - the Oscars or Allies?
BTW, the early Oscars have weak MGs. I bet you damaged several Allied aircraft but none were destroyed outright in the air combat. Watching combat animations will also tell you if you land damaging hits on Allied fighters.The Allied fighters you engaged are more durable than Oscars.
Oscars were diving on P-40.
And there were lots of message says, Oscar fire behind H81 at 5,4,3,2,1. then nothing happens
several message shows "Critical hit on H81" but the H81 was lightly damaged in the animations and they are still in the fight.
I guess the very hard AI get the bonus on 1, dice roll deciding if it is hit or not, 2, double or even triple endurance rating so it is very hard to shoot down.
Btw. half of the Ki-43 I commit into the battle are the Ic, they have 2 x 12.7mm
overall, I sent 14 x Ia, 28 x Ib (with 1 x 12.7mm+1 x 7.7mm) and 14 x Ic in two day air battle
Re: So, the very hard AI is cheating, right?
Copy that.Chris21wen wrote: Sat Apr 06, 2024 6:46 am A tip on the combat report.
The in game report is a copy of a text file located in the save folder. I use it all the time. It has three advantages of the in game, you can post text and not a pic, it's text so you can edit it and you can read it and move around the game map. Of course the latter works best if you are in a window and/or have more than one screen.
Thanks for the tips
Re: So, the very hard AI is cheating, right?
BTW, this is a tiny scenario "Battle of Burma". So next week when I got more time I can run a test. With more data we can have a clear picture on the very hard AI bonus.
using the same save file, I am going to running the game at hard AI then normal AI. I will also switch to Allies side, facing the normal IJA AI and very hard AI.
running the battle several times and record how many A/C are lost at the end of the day.
using the same save file, I am going to running the game at hard AI then normal AI. I will also switch to Allies side, facing the normal IJA AI and very hard AI.
running the battle several times and record how many A/C are lost at the end of the day.
Re: So, the very hard AI is cheating, right?
I used that save file run a quick test.
Very Hard 3 times, Easy 3 times
AI is cheating.
Above is the kill ratio at very hard level, below is at easy level.
There is only one air to air engagement in this theater, that is the IJAAF sweeping at Moulmein.
Very Hard 3 times, Easy 3 times
AI is cheating.
Above is the kill ratio at very hard level, below is at easy level.
There is only one air to air engagement in this theater, that is the IJAAF sweeping at Moulmein.
Re: So, the very hard AI is cheating, right?
It sure looks like AI is cheating. I am always suspicious of small data sets. Any chance you could run the test 5 or 10 more times and post the results to see if these results are typical or if the difference is smaller with a larger data set.
I found with the randomness in the game it sometimes took 10 runs to confirm expected results.
I found with the randomness in the game it sometimes took 10 runs to confirm expected results.
Re: So, the very hard AI is cheating, right?
You should also show what settings (weather, fog of war, etc) are used. I would also be curious to see the full combat results file from each of your attempts.
- Platoonist
- Posts: 3042
- Joined: Sun May 11, 2003 4:53 am
- Location: Yoyodyne Propulsion Systems
Re: So, the very hard AI is cheating, right?
On Very Hard is it cheating per se or is it called a "combat advantage"?
Re: So, the very hard AI is cheating, right?
Looks like human is cheating in easy mode too. Interesting to see results at Easy, Historical, Hard, Very Hard, and maybe H2H also. I always wondered how impactful the "combat advantages" are at Very Hard level but have not done the work since I have been immersed in PBEM games where my opponents don't need any advantages to humble me.
- Platoonist
- Posts: 3042
- Joined: Sun May 11, 2003 4:53 am
- Location: Yoyodyne Propulsion Systems
Re: So, the very hard AI is cheating, right?
I don't think the developers have ever disclosed what those difficulty advantages are exactly. Loaded dice in all combat rolls perhaps? However, by selecting "Very Hard" as written in the manual it seems to me that a player is consenting to play on an uneven field. Which is not quite the same as the AI "cheating" where a player assumes the game routines are fair and impartial for both sides and then finds that they're not.pbiggar wrote: Thu Apr 11, 2024 12:15 am I always wondered how impactful the "combat advantages" are at Very Hard level but have not done the work since I have been immersed in PBEM games where my opponents don't need any advantages to humble me.
Re: So, the very hard AI is cheating, right?
Yep, AI cheating would be Hard AI getting combat advantages from Very Hard AI setting.
Still, knowing what those "combat advantages" really are wouldn't hurt. In Tcao's example it may mean higher exp/skill of enemy pilots, better CAP positioning or something else. How is a player supposed to improve his game if he doesn't know what is his deficiency vs beefed up AI?
Still, knowing what those "combat advantages" really are wouldn't hurt. In Tcao's example it may mean higher exp/skill of enemy pilots, better CAP positioning or something else. How is a player supposed to improve his game if he doesn't know what is his deficiency vs beefed up AI?
Re: So, the very hard AI is cheating, right?
"Combat advantages" what I have deducted trying VH level are apparently "loading the dice", which I don't like.
Hard is OK since it gives AI logistical advantages but no advantage in combat rolls.
Hard is OK since it gives AI logistical advantages but no advantage in combat rolls.
"To meaningless French Idealism, Liberty, Fraternity and Equality...we answer with German Realism, Infantry, Cavalry and Artillery" -Prince von Bülov, 1870-


Re: So, the very hard AI is cheating, right?
Against the AI, carriers never run out of sorties nor torpedoes.
Seek peace but keep your gun handy.
I'm not a complete idiot, some parts are missing!
“Illegitemus non carborundum est (“Don’t let the bastards grind you down”).”
; Julia Child

I'm not a complete idiot, some parts are missing!
“Illegitemus non carborundum est (“Don’t let the bastards grind you down”).”
Re: So, the very hard AI is cheating, right?
Aye, let's put it in a more diplomatic way. Very Hard AI has a huge "combat advantage"Platoonist wrote: Wed Apr 10, 2024 10:46 pm On Very Hard is it cheating per se or is it called a "combat advantage"?![]()
AI difficulty.jpg
From my observation, AI A/C got survival bonus as it is much harder to be hit , if it is hit it can take a lot of punishment before go down in flame.Yaab wrote: Thu Apr 11, 2024 3:00 pm Yep, AI cheating would be Hard AI getting combat advantages from Very Hard AI setting.
Still, knowing what those "combat advantages" really are wouldn't hurt. In Tcao's example it may mean higher exp/skill of enemy pilots, better CAP positioning or something else. How is a player supposed to improve his game if he doesn't know what is his deficiency vs beefed up AI?
BTW, what does the "Critical hit" mean? Does it mean a hit in engine part, flight control or cockpit?
anyway, like mentioned before, in VH mode the combat animations show lots of "Oscar Dive Down", "Oscar fire behind H81 at 5,4,3,2,1". then nothing happens. "Critical hit" on this and "Critical hit" on that but H81 is still in the fight.
In easy mode, the message like that will guarantee the H81 is either destroyed or withdraw battle with dive down and smoke trail it.
Another advantage VH AI has, based on my rough impression, the human player A/C will experience lots of mechanical issue forcing them withdraw from battle. A 15 vs 6 numerical advantage will quickly shrink to a 8vs6 more evenly battle.
Just want to let you know that I am still working on this. I was distracted by a different issue.pbiggar wrote: Thu Apr 11, 2024 12:15 am Looks like human is cheating in easy mode too. Interesting to see results at Easy, Historical, Hard, Very Hard, and maybe H2H also. I always wondered how impactful the "combat advantages" are at Very Hard level but have not done the work since I have been immersed in PBEM games where my opponents don't need any advantages to humble me.
When I do the testing, I was surprised to find out that it produce the same combat result again and again. it is the same number of A/C lost in air to air, same number lost in OPs. Same number of Pilots got KIA, and it is even the same guy promote into an Ace, same guy got shot down then MIA.
I realized that the game creates a seed as soon as the turn ends. so the combat result will be the same.
To create a different seed, one has to slightly change the configuration. For example, I keep all the fighters sweep mission the same, but changed the recon plane's target, it generates a different battle result.
So I am still working on this test. Hopefully I can post a report next Mon or Tue.
Last edited by Tcao on Fri Apr 19, 2024 5:13 pm, edited 1 time in total.


