carriers at war AI great?

[tomcat666]

I have heard it mentioned by some that the AI in this game is great, that it is perhaps the best in wargaming history.

yet I came across this review from the armchair general.

To quote:

"
I agree with most of what Mr Faubert says about the game. It is fun for the first go around as each side but I would submit that the AI is too timid and does not take chances.

I found that as the USN player if one pushed hard and kept the carriers near land bases (especially in the three south Pacific scenarios) one could always win the scenario with little loss. The Japanese carriers would not close with land bases and a few transfers from the US carriers to the fronline land bases allowed for very effective strikes against the Japanese.

As the Japanese the USN AI is pretty hopeless. Once you understand the search plane routine you can easily keep the range open and simply sink the US carrier and surface forces. Surface battles should always be sought by the Japanese, the carrier and land base air soften up the heavy units and the Japanese heavy units then move in and mop up.

One thing that the game does very well is remove the omnipresent position of the player. At dawn you launch the search planes and set your course for where you "think" the enemy will appear. Once contact is made you then launch the strikes and, like the real admirals on the ships, you then wait for the results. Launch too far away and the planes may have to abort, hold back and the enemy may get in close enough to launch the killing strike first. You must play the role of the admiral,...not the bomber pilot,....and much is out of your control.

Quite correctly the players cannot control the strikes and must hope the AI air bosses will send enough aircraft against the priority targets. Like the admirals, captains and CAGs you sit next to the speaker and wait for the results.

Its fun for a while,...but after a few runs you tuck the game away to wait on a boring Sunday with a couple of hours to kill."


Now I ask what is going on here?, some say the AI is ruthless and efficient, some others say it is dumb.


Or are there compatibility issues with AI, where on some systems it does not work correctly.

This is evident with a professional reviews of Gary grisby's: world at war series, one source said the AI was timid, another two sources said the AI was very aggressive and tough.

so I ask the experienced gamers here, before you put down hard money on a game, and where AI is very important to you , how do you tell the truth on this issue?

[decaro]

... so I ask the experienced gamers here, before you put down hard money on a game, and where AI is very important to you , how do you tell the truth on this issue?

Despite all the reviews I find via googling the game, I don't think there is anyway to tell for sure how good the software is until it's too late and you already bought it!

Although mainstream game sites are usually unbiased, player reviews can be more revealing. But sometimes players have no idea what they're doing w/the game and can steer you away from a good title.

[Gregor_SSG]

There are a number of interesting points raised by the question and the comments made in it, which I will try to address.

The AI in CAW can still beat me, and I've been playing the game since the first Apple II release. It has to be said that I'm not, even after all these years, a brilliant wargamer, but that said familiarity has not bred contempt.

If you read through the posts on this forum, you will find a large number of posts from gamers confessing that the AI beat them, not just once but a number of times. I submit that that is a highly unusual situation, and evidence that the AI is doing a great job.

Lastly, it sounds like the player complaining about the AI is using some completely ahistorical tactics to get some of his results. Carrier planes did end up at Henderson Field, but the USN did not use their fleet carriers as barges, deliberately landing multiple squadrons on Henderson and then cowering out of range. We don't want to stop lots of planes landing at Henderson, because they could have ended up there if their carrier was sunk.

This is a difficult area, and sometimes there are no clear cut answers. However I feel that when you start a scenario, at least against the AI, *if* you want a fair fight then you buy into the historical situation and allow this to constrain your behaviour. In other words, don't do things as a commander that would get you sacked the next day.

This is only a suggestion, people are free to play the game they way they want to and we don't want to cripple the game by trying to prevent them. However, if you want a fair comparison, then play a historical wargame historically.

Gregor

[Adam Parker]

ORIGINAL: tomcat666

yet I came across this review from the armchair general.

To quote:

"I agree with most of what Mr Faubert says about the game. It is fun for the first go around as each side but I would submit that the AI is too timid and does not take chances.

I found that as the USN player if one pushed hard and kept the carriers near land bases (especially in the three south Pacific scenarios) one could always win the scenario with little loss. The Japanese carriers would not close with land bases and a few transfers from the US carriers to the fronline land bases allowed for very effective strikes against the Japanese."

****
.....so I ask the experienced gamers here, before you put down hard money on a game, and where AI is very important to you , how do you tell the truth on this issue?

This review is wrong IMO when it comes to the Jap AI (I've yet to really play vs the US AI).

The Jap AI does indeed close with land bases - and invade them. The transfer of US planes to fortify land bases has been addressed IIRC.

The latest patch addressed replayability. The spottting information tweak really leaves FOW massively in effect now - too in effect?

As for my personal experience and enjoyment vs the AI see my quick thread here tm.asp?m=1512308

[decaro]

I have played several different scenarios as the IJN vs. the US (AI) and found it much less challenging, possibly because of the play (im)balance.

I can say that CaW has an aggressive AI, but I'm not sure if it's great or just good; sometimes I think my losses have as much to do w/luck as anything else.

I will say with some certainty that the CaW interface is unmatched.

[Warspite3]

From me playing the last few days, I have lost the coral sea scenario historical variant as the US everytime. Not even close and I have played it about 8 times so far. So I think the AI is really good to be beating me over and over at the same scenario or else I would have certainly won by now.

[Reg]

Or you are doing something VERY wrong......

[decaro]

ORIGINAL: Warspite3
... I have lost the coral sea scenario historical variant as the US everytime. Not even close and I have played it about 8 times so far ...

That's a tough scenario to play as the US; are you meeting the AI head on? Try moving your carriers west and "park" them under some clouds. Then let the AI TGs come to you -- i.e., within LBA -- and (eventually) you'll see better results.

[OldBoney]

Here’s my two cents worth on this subject. First I think the AI performs pretty well in the context of the simulation (I've mostly played as the US). I base this on two questions:

If I do the same thing does the AI respond predictably? No I’ve replayed the same scenario many times and found it played out differently most of the time.

Is it possible to "learn" a scenario? Perhaps in general terms but the crucial factors are when and what to attack. The FOW level with the patch makes for a lot of guessing.

Aggressiveness? Isn't this a matter of setting the response actions on the War Cards. Sure I've seen the IJN lose a couple of fleet carriers early in a scenario and go into preservation mode. Probably more historical than a wargamer who seeing he's losing decides to gamble everything.

"Best Ever" is over the top but I think the AI is really pretty good and a game of maneuver like CAW needs a flexible AI opponent.

[Oleg Mastruko]

ORIGINAL: Warspite3

From me playing the last few days, I have lost the coral sea scenario historical variant as the US everytime. Not even close and I have played it about 8 times so far. So I think the AI is really good to be beating me over and over at the same scenario or else I would have certainly won by now.

Coral Sea is really unbalanced scenario, try playing as IJN, then comment.

CAW AI is decent - it comes to the battlefield, it offers battle (even if the smartest thing would be to retire, but then there would be no game [;)]) and unless destroyed, it lingers around offering the player the opportunity to smack it some more in round 2 (or be smacked again).

However.....

- AI never pursues weaker enemy foces (to make the player see how REALLY brutal naval warfare should be)
- AI is bad at using land bases, especially at concetrating air assets as far forward as possible (visible with US AI more than with IJN AI because US aircraft are shorter ranged)
- AI does not use feints. Once you see AI TF going somewhere it will continue going there, unless seriously mauled (it will not change the course just to fool you)
- AI does not know to use baits

So, even though the listed shortcomings are expected from ANY AI, I honestly can't see how CAW AI would be the "greatest ever". It's decent, but nothing more than that. People who think CAW AI is gold should try playing human player(s) with medium experience to see what is some real challenge.

[decaro]

Oleg: I basically agree w/your conclusions re the CaW AI except this one ...

- AI never pursues weaker enemy foces (to make the player see how REALLY brutal naval warfare should be)

On more than one occaision, the AI (as the IJN) would target fixate on a decimated fleet and hang on to it like a pit bull until every ship in the TG sinks, leaving an icon of a sinking ship on my screen. But the AI sometimes does this to the point of ignoring more important threats to it, i.e., my CV TGs.

And yes, the CaW AI -- like most chess AIs -- doesn't know the meaning of the word "sacrifice"; although it will not bait, it will readilly take the bait, apparently w/o much consideration/processing.

[greywulfe]

I've played a lot of wargames in my day and I do find the AI challenging for the most part. In the early days I am able to pretty much trounce the USN in any scenario. The USN is more difficult, especially in the Coral Sea - which I eventually won by trial and error.

Obviously using hindsight you already have an advantage. Which is why I am able to destroy the USN during Midway, only because I know the American carriers were there!

I think if we had more scenarios, including random or 'what ifs' it would post more of a challenge to experienced players.

[Gregor_SSG]

Somewhere in this discussion the bar seems to have been boosted somewhat higher than it deserves to be. The correct comparison for CAW AI is against other wargame AIs. If we could produce an AI that could consistently beat the best human players, then we'd all be working for some secret government think tank.

In comparison with other wargame AIs, I think that CAW is the best performer, especially as it produces its challenge without cheating of any kind. It has to be said that carrier warfare is an ideal subject for producing an AI, but that said it's still a great system that can provide a continuing challenge.

Gregor

[decaro]

ORIGINAL: greywulfe
... I think if we had more scenarios, including random or 'what ifs' it would post more of a challenge to experienced players.

Greywulfe, did you patch-up? If so, there's a new random check box "hiding" in plain sight right in the middle of the main screen, but it's small and you may have missed it!

There are already several "what if" scenarios, i.e., what if a missing certain carrier was now available for an operation, etc., but Im still waiting for more new scenarios.

I can wait.

[tomcat666]

   How does the AI in HTTR compare?, i heard that it is flexible as well.

If you believe panther games it has the best AI. It is probably a greater achievement in programming since the nature of the game is more complex

[decaro]

I have HttR, and though the AI is good, it's not as aggressive as most players would like. Supposedly Panther Games corrected for that in CotA, but I'm still waiting for BftB; after I buy it, then I'll know for sure.

I'm not sure about comparing CaW w/HttR or any other real-time land-based strategy game; the only other naval game I have is the turn-based UV, but almost everyone plays it by PBEM.

What real-time naval strategy games are there to compare w/CaW/CaW's AI?

[MarkShot]

I have been working on Harpoon 3 and am still just in learning mode.

It is realtime. The level of detail is infinitely greater than CAW/CCAW, but so is the learning curve and the strategic options. So, the two games target different types of players. Of course, Harpoon, is much broader than just WWII carriers and unlike CAW, the amount of third party content is staggering.

In terms of AI challenge, I think the AI in CAW is adequate. Somewhere's I wrote a thread of tips. The SP will keep you amused. Perhaps, when the original was released it represented the very best of AI programming. However, I think these days there are probably many other comparable examples of AI programming. The same with the scenario editor. I am sure at the time when it was first released, it represented unsurpassed power and flexibility. However, these days what it offers is pretty much par for the course. If you look at Sub Command or Dangerous Waters from Sonalysts, you will find an equally powerful and flexible scenario design capability for naval combat with these games.

[vinboni]

When I play as Japan AI isn't adequate I think. It's very easy to win in many different scenarios, but when I play as Allied is much more difficult.

[MarkShot]

Well, it is a very rare game indeed that can give a full challenge on two sided scenarios when played from either side.

If a game beats you playing either side than you have either found:

(1) A very well coded AI.
(2) A game that uses computer cheats to improve the player challenge level.
(3) Or sadly, you the player are truely clueless.

[decaro]

Re 3# -- sounds like you read Reg's earlier post.