AI Development and Our Hobby

[pzgndr]

Someone find me a game where a strategic level AI is genuinely capable of:
1) Launching a coordinated large-scale amphibious invasion.
2) Effectively responding to a large-scale amphibious invasion.
3) Changing a nation's military unit production to counter what is actually occurring on the battlefield.
4) Changing a nation's technological research path(s) to counter what is actually occurring on the battlefield.

Well, SC2 has these features right now. The combination of a good generic AI plus scripted events (including customized events for Allied or Axis AI) plus scripted AI behavior based on game conditions allows a modder with sufficient patience to develop a challenging computer opponent. For better or worse, SC2 with its comprehensive editor does in fact do all these things. Not perfectly of course, but one needs realistic expectations for playing against a dumb machine.

SC2-WaW is even better. Several improvements to the generic AI make it noticeably better. A few changes to some script structures provide some more flexibility. And the editor is more comprehensive than it was before, allowing modders to adjust many game parameters that were previously hardwired. Keep an eye out for this to be released soon.

In retrospect, it's taken several years since SC was first released for Hubert Cater to grow his game and get the game and AI to this point. CEAW will likely experience similar growing pains on its own path to success. While many basic features of both games are similar, there are enough differences to make each game unique. And we should all encourage this! There may be a natural tendency for SC2 to become more CEAWish and for CEAW to become more SC2ish over time, but we do not want or need two exact same games where one has hexes and the other has tiles. (As if. [8|]) Allow the game developers to be creative and impress us with new wonders. [8D]

Right now the main problem with computer game AIs is that they hardly "learn from mistakes"

Another problem is programming/scripting the AI to focus on a particular grand strategy that integrates planning with research and production priorities. That would be nice. And yet another "problem" is getting the AI to perform feints and other deceptive maneuvers as deliberate actions. (AI bugs don't count!) Maybe someday we'll see a cunning computer opponent worthy of the title "grand strategist."

[Warfare1]

ORIGINAL: pzgndr

Someone find me a game where a strategic level AI is genuinely capable of:
1) Launching a coordinated large-scale amphibious invasion.
2) Effectively responding to a large-scale amphibious invasion.
3) Changing a nation's military unit production to counter what is actually occurring on the battlefield.
4) Changing a nation's technological research path(s) to counter what is actually occurring on the battlefield.

Well, SC2 has these features right now. The combination of a good generic AI plus scripted events (including customized events for Allied or Axis AI) plus scripted AI behavior based on game conditions allows a modder with sufficient patience to develop a challenging computer opponent. For better or worse, SC2 with its comprehensive editor does in fact do all these things. Not perfectly of course, but one needs realistic expectations for playing against a dumb machine.

SC2-WaW is even better. Several improvements to the generic AI make it noticeably better. A few changes to some script structures provide some more flexibility. And the editor is more comprehensive than it was before, allowing modders to adjust many game parameters that were previously hardwired. Keep an eye out for this to be released soon.

In retrospect, it's taken several years since SC was first released for Hubert Cater to grow his game and get the game and AI to this point. CEAW will likely experience similar growing pains on its own path to success. While many basic features of both games are similar, there are enough differences to make each game unique. And we should all encourage this! There may be a natural tendency for SC2 to become more CEAWish and for CEAW to become more SC2ish over time, but we do not want or need two exact same games where one has hexes and the other has tiles. (As if. [8|]) Allow the game developers to be creative and impress us with new wonders. [8D]

Right now the main problem with computer game AIs is that they hardly "learn from mistakes"

Another problem is programming/scripting the AI to focus on a particular grand strategy that integrates planning with research and production priorities. That would be nice. And yet another "problem" is getting the AI to perform feints and other deceptive maneuvers as deliberate actions. (AI bugs don't count!) Maybe someday we'll see a cunning computer opponent worthy of the title "grand strategist."

Very well said!

I have noticed that in modding scenarios in Panzer General 1 (yup - from 1993), that if I give the AI more recon units (I also increase their movement, spotting and fuel levels), the AI tends to make smarter decisions.

In large strategic games such as these devise ways to give the AI "eyes".

[Warfare1]

Slitherine just released info related to what will be included in the first patch:


We wanted to keep everyone up to date with our plans for Commander. Here is a list of issues we are planning to investigate in the short term. We can't promise these will all make it in as you can't tell how difficult some things will be until you start trying to do them, but we'll do our best!

* PBEM saved game names
* PBEM password protection
* PBEM feedback on opponents turn
* Chat in TCP/IP games
* AI for DDay landings
* AI on normal difficulty setting
* Manpower levels for the UK
* Scroll speed
* Map centering before combat
* Nationality & leader icon clarity to show who has moved
* Zoom level during the AI turn
* Turn summary message issues
* Destroyer & battleship graphical mix up
* Editor
* Screen mode issues, failing to set up 1024x768 on some machines
* Oil abundance in Africa

There are lots of other things we'd like to look at, but these are the ones we think we can get done quickest and that will have the most benefit for the first patch. We're estimating this will be ready for mid July.

http://www.slitherine.com/forum/viewtopic.php?t=3624

[Warfare1]

ORIGINAL: Marc Schwanebeck

I think the problem is also that some wargamers expect they get an AI that is as capable as their best wargaming buddy (that does not cheat). This is nonsense and way to high expectations. Scientists burn billions each year to let a robot do things a 2 year old could do better, and no the robot does not do it on his own too [;)]

If a game AI "cheats" that is no bad thing at all. You can not copy human cheating over to computer cheating. Human cheating is negative, for sure. Computer "cheating" is necessary. Computers only do what you tell them, they do not do the thinking.
Right now the main problem with computer game AIs is that they hardly "learn from mistakes", unlike humans. Your MP buddy will do this. Once we are at a stage where computers game AIs can learn certain things we will see better artificial opponents.

There is certainly room, sometimes a lot of room for improvement with some games AIs, but the expectations should not be set too high too. We have had also several cases where gamers simply did not bother to raise the difficulty level and then complained about a bad AI when they play on "easy level". Some people are extremely good at finding patterns of behavior within games and thus have no problems to defeat the AI even when it cheats extremely. It will be extremely hard to please these people with a good AI until scientists and coders have developed a real AI that has human patterns of thinking and learning

Absolutely agree.

The AI simply cannot think like a human. Doubt it ever will.

The AI will always need help. Gamers expecting games where the AI does not "cheat" will never have a very playable game.

Thus, the game creator/designer can then use his knowledge of devious tactics and tricks and program them into the AI side. This, added to subtle forms of AI advantage (such as free invasions, more troop levels, spotting advantages, etc), along with good editors will give any wargamer a challenging game.

[dinsdale]

ORIGINAL: Marc Schwanebeck

I think the problem is also that some wargamers expect they get an AI that is as capable as their best wargaming buddy (that does not cheat). This is nonsense and way to high expectations. Scientists burn billions each year to let a robot do things a 2 year old could do better, and no the robot does not do it on his own too [;)]

I think you're overestimating the demands Marc. I don't think anyone will expect a Big Blue equivalent for wargaming anytime soon, but simply a competent opponent. There's a huge problem in claiming "AIs can't be as good as humans" when the AI in question (and I am not commenting on CEAW) is unable to execute the most basic calculation-driven "intelligence."

If a game AI "cheats" that is no bad thing at all.

It is if it destroys suspension of disbelief. It can in the extreme cases, render rules irrelevant and strategy worthless. For example, if an operational Napoleonic AI suffers no fog of war, then the potential subgame of cavalry screens skirmishing would be pointless. AI cheats are of course necessary, but they need to be well thought out and part of an existing competent set of rules, not a x10 resource bonus because the AI is unable to cope with attrition or inflation.

Cheats should not be a crutch, they should work with the AI as aid for it's weakpoints. That can't be achieved very well if the AI in question is rubbish to begin with. (Again, not talking about CEAW in particular.)

Right now the main problem with computer game AIs is that they hardly "learn from mistakes", unlike humans. Your MP buddy will do this. Once we are at a stage where computers game AIs can learn certain things we will see better artificial opponents.

We're not really going to get there though are we. Not unless wargaming is able to reuse breakthroughs researched elsewhere. FPS and RTS are less dependent on AI smarts as the PC is able to out-compete the human in reaction time and calculation, two areas which can dramatically improve the appearance of an AI in a real time game or sim. So with the major market not needing the depth of decision making and learning, where is wargaming going to get the money together to get to the stage you say?

--------

While there are some interesting points in this thread, it contains the usual strawmen thrown out whenever Ai is mentioned

1) "There isn't even a chess AI which can beat humans and that's a simple game." Sorry, but what utter rubbish. 90% of chess players can be consistently beaten by chess game AIs. They do it without giving the PC 4 Queens and respawning bishops. Further, the skill level of a wargamer, playing a game for perhaps a few months, cannot be close to even home chess players who have been playing the game since childhood.

2) AI's can't be reused. Not wholesale they can't, but let's not pretend that code/design/algorythym reuse between games is impossible because frankly it's insulting. There may be market issues which prevent leveraging technology or building upon reusable components, but there's no technical barrier which would have prevented code and techniques in Chariots of War to have been built upon for subsequent Slitherine releases and reused in this game.

3) Tactical and Strategic are so dramatically different that one can never have a good AI. Again, untrue. There's no difference between strategic games and tactical when it comes to movement and combat. While forward planning and managing strategic options is a different kettel of fish, strategic is the same as tactical on the board.

The greatest impediment to AI development though is cost. Wargaming in particular is victim of one-man-bands: games where one or two programmers do everything. Would anyone expect their mechanic to come plumb-in a toilet? Just because two people use a wrench, it doesn't make their jobs equal, and it is the same in programming. AI (and UI for that matter) are highly specialized disciplines which are unlikely to be mastered by programmers who have to do everything.

Thus it should be no surprise that AI and UI are typically the very worst facets of most wargames. It's less to do with impenetrable research boundaries and more to do with expertise.

Finally, IMHO the second greatest problem with AI development is us. There's simply not enough of a selling point when a game has a better than average AI. If it were, Panther games and SSG would outsell everything else by Matrix 10:1. We do not buy games because the AI is good, we buy games and hope. So in all fairness, why should Slitherine cut back on features or spend money on AI which is unnecessary to sell their games?

[IainMcNeil]

Just to clear up one fact - we do not use any of the AI from Chariots of War in our games. That AI is dead and buried and will never be returned to. AI is so game specific anything other than generic route finding algorithms are useless to other games and route finding is not even really what you woudl call AI.

[Warfare1]

Finally, IMHO the second greatest problem with AI development is us. There's simply not enough of a selling point when a game has a better than average AI. If it were, Panther games and SSG would outsell everything else by Matrix 10:1. We do not buy games because the AI is good, we buy games and hope. So in all fairness, why should Slitherine cut back on features or spend money on AI which is unnecessary to sell their games?

Hi

Overall in your post you made some valid points.

I wonder if game sales in our hobby isn't also related to the type of game being offered?

SSG games to date may have excellent AI, but I am less interested in para operations or single scenarios, then I am in large strategic level games, such as the one by Matrix (UV), Slitherine (CEaW) and Battlefront (SC).

It seems that a majority of wargamers prefer the larger strategic level games. This has less to do with AI, and more to do with game preference.

If SSG decides one day to apply their AI expertise to a large strategic game of WWII, then they will no doubt garner even more attention.

For some reason I like the simplicity of the CEaW and SC game designs....

[decaro]

Overall, some very thoughtful -- and very long -- posts.

BoA has a decent AI, but it abstracts many features and gives the user fewer tactical options. However "Athena" -- its AI -- works well; there's even an extended AI time feature that's very effective, not unlike the "think on opponent's time" option on my Kasparov Chess Computer, which I set to beat me most every time (think Mr. Spock on that old Star Trek episode).

Also, I think it's easier to mod ball and musket and other primitive warfare -- i.e, no air, subs, etc., -- than most modern conflicts.

Some time ago I bought the original Superpower , a modern sim that boasted a neural net AI that was not pre-scripted and supposedly learned; the result was that the oddest pairs of nations went to war w/each other, always ending in WWIII.

Sid Meir's Civ AI usually got my goat, whether I won or not.

[jcorbin]

ORIGINAL: Marc Schwanebeck

Finally, IMHO the second greatest problem with AI development is us. There's simply not enough of a selling point when a game has a better than average AI. If it were, Panther games and SSG would outsell everything else by Matrix 10:1. We do not buy games because the AI is good, we buy games and hope. So in all fairness, why should Slitherine cut back on features or spend money on AI which is unnecessary to sell their games?

We wargamers obviously represent a very small portion of the overall strategy gaming market, especially when you factor in so many RTS titles. But when you look at the sales for the turn-based segment of this same market--in particular the "Civilization" series and the "Rome: Total War" series--these titles clearly indicate that we have a segment that might be willing to crossover into wargaming if the following things occurred:

1) Wargame developers received enough funding to improve their presentation. For example, I am currently playing Strategic Command 2 for the first time and the very first thing I noticed is the poor UI design. The game is definitely fun, despite it's many abstractions, but the UI is not fun at all--in particular the non-intuitive toolbar running along the right side of the screen. Of course, you seem to have one guy (Hubert Cater) doing all of the programming, undoubtedly on a shoestring budget, so this is probably to be expected. One drools at the possibility of an investor throwing money at a small company like Fury Software in order to improve the game's presentation. Meanwhile CEAW has the classic hexagon look. Many young non-wargamers will take one look at this presentation and say "Now that looks like a wargame my father used to play back in the seventies. I'm not really interested." SC2, as I understand, changed its look from the classic hexagon to isometric tiles. Is it flashy? Obviously not. Does it differ from the traditional look? Yes. Would the non-hexagonal look appeal more to non-wargamers? My guess would be yes.

2) Wargame developers created products that mainstream gamers would find more appealing. An obvious example would be more streamlined play. This appears to be what CEAW has attempted. What worked so well for the Panzer General series years ago might, in a 21st century incarnation, work even better now. Time will tell. Certainly the "Hearts of Iron" series suffers from its initial complexity (until you figure out all of the game's nuances) and, in so doing, only caters to the niche market.

3) Wargame developers created better AI. And now we're back to the reason I started this thread. Marc's point above is well taken, but I strongly believe that a really good AI will always remain a clear selling point, especially when coupled with points 1 and 2. Slitherine certainly knows this, judging from the references on the following patch candidate list, which has a few proposed AI fixes:

http://www.slitherine.com/forum/viewtopic.php?t=3624

[dtx175]

To get some insight in the difficulty of programming AI to "think" and "see" a game board like a human, I recommend "How Computers Play Chess."  It's ideas are applicable to computer wargames as well. 
 
AI can't "see" the map - it only "knows" what the programmer has assigned as numerical representations on a map (everything a computer "knows" is numeric -ultimately just 0s & 1s).  Similarly, each unit on each hex of the map represent a different number. 
 
To give some sense of this, one of the programmers for Big Blue (the IBM supercomputer that played world-class chess) did his entire PhD thesis on the value of a knight in the center 4 squares of the board.  Wargame maps and their units are far more complex than a chess board and we unfortunately don't have people doing PhD theses calculating the value of a invading division in the hexsides of the coast of France.
 
Also, the programming for computer chess began in the 1960s and chess programs of today benefit from these decades of development.  Chess pieces and chess boards remain the same and the computer algorithms can be steadily refined.  In contrast, wargame maps and their pieces constantly change. 
 
As I play entirely against the computer, I always want better AI - but I accept that better AI is a huge challenge.   

[dinsdale]

ORIGINAL: Warfare1

Hi

Overall in your post you made some valid points.

I wonder if game sales in our hobby isn't also related to the type of game being offered?

...If SSG decides one day to apply their AI expertise to a large strategic game of WWII, then they will no doubt garner even more attention.

Absolutely, I think you're expanding on the point I was trying to make: AI is hardly a consideration.

For reasons you point out about SSG (Korsun Pocket, BiN BiI) could have the greatest AI on earth and it won't necessarily make a difference to someone buying it if they're not interested in the scale or the subject. So it shouldn't be a surprise that games which seem to have better than average AI's aren't necessarily the best sellers.

------------

ORIGINAL: Iain McNeil
Just to clear up one fact - we do not use any of the AI from Chariots of War in our games. That AI is dead and buried and will never be returned to. AI is so game specific anything other than generic route finding algorithms are useless to other games and route finding is not even really what you woudl call AI.

Iain, are you really stating that game AI is unique among almost all code in that none of it is reuseable? I offered CoW as an example, that perhaps if process had been put in place then, that a significant amount of work could be reused and improved rather than thinking that each game must reinvent the wheel.

---------------

ORIGINAL: JoeDBoA has a decent AI, but it abstracts many features and gives the user fewer tactical options.

I'd be interested in why you think BoA has a simpler ruleset than CEAW.

---------------

ORIGINAL: dtx175
AI can't "see" the map - it only "knows" what the programmer has assigned as numerical representations on a map (everything a computer "knows" is numeric -ultimately just 0s & 1s). Similarly, each unit on each hex of the map represent a different number.

One might as well also explain that surround sound and 3d effects are also handled by what is ultimately just 0s and 1s. I'm not sure what the point is.

To give some sense of this, one of the programmers for Big Blue (the IBM supercomputer that played world-class chess) did his entire PhD thesis on the value of a knight in the center 4 squares of the board. Wargame maps and their units are far more complex than a chess board and we unfortunately don't have people doing PhD theses calculating the value of a invading division in the hexsides of the coast of France.

Why is BigBlue relevant? It was an exercise by a large team in trying to beat the greatest player on earth. Until everyone playing wargames is the equivalent of a world chess champion, then I hardly think the detail required for that task are relevant.

Also, the programming for computer chess began in the 1960s and chess programs of today benefit from these decades of development. Chess pieces and chess boards remain the same and the computer algorithms can be steadily refined. In contrast, wargame maps and their pieces constantly change.

Do you think that two wargames are more or less related to one another than the relationship between your cell phone and car? How about the difference between Amazon sales and Dreyfus Insurance risk assessments?

There are no technical hurdles to exponentially better AI today, only economic ones. If a wargame with great AI sold 5 million copies, the industry would be tripping over itself to emulate the success and technology would be utilized to make that happen.

The barriers to better AI are economic, nothing else.

[YohanTM2]

Give me a break pzngdr, your are a fanboy for SC2 (and tester I believe) and it is a very smelly pile of droppings. SC was a great game, for beer and pretzels, and consumed many an hour of PBEM effort for myslef and the lads.
 
SC2 was a joke, it was unplayable and I wish I had followed Irish Dragoons advice and not bought.
 
But i must admit, it wins hands down for PBEM at this stage as CEaW blows chunks for PBEM. I was on the phone explaining to an opponent last night that he had to check the value of his convoys each turn to see if they had been attacked, let alone where an enemy sub may even have been seen.
 
Add to this no replay for air attacks so you don't know wher his planes are.please patch soon and you can re-open your turn in PBEM 800,000,000 times with no cheat code pickinh it up

[geozero]

Okay... so since we are dealing with a STRATEGIC level game, perhaps one way to look at the AI issue would be sort of similar to real life military chain of command...

Have several "layers" of AI decision making. Upper most level would make decisions or "orders" to t he lower layer, then that layer would take info and make decisions, and so on... let me explain in layman's terms.

Suppose you play AXIS and the AI opponent is the Allies.

Typically, Germany invades Poland first. The Allied AI upper layer decides to a) declare war, or b) not declare war. IF decision A is made then it orders it's military and production to a war footing (If "B", then Allies remain Pacificst for now). This changes IF Germany declares war on Allies of course... assuming that decision A is taken, the Allied AI now has to look at the Situation... there's no combat yet in Africa, so nothing is done there. But a lower AI layer could decide to A) use bombers to attack Germany, B) have France attack in the West, C) send out the fleet in Search & Destroy missions against Axis subs and surface shipts, etc, etc.

Let's say decision B above is activated, Now a lower level AI decides which French units will attack and which will remain in defensive positions. Which targets present a better attack outcome? An undefended German city? A weaker Axis unit? etc.

That's how AI should work. Yes it takes programming. If, Then, Else commands as noted elsewhere. When enough of these scripts are programmed the opponent will make very good decisions, whether it is what type of production to build, where to attack or defend, when and where to invade, etc. We were programming this kind of stuff way back in the 80's using Basic.

Puts on fire retardant suit...[:D]

[pzgndr]

Give me a break pzngdr, your are a fanboy for SC2 (and tester I believe) ...
SC2 was a joke, it was unplayable and I wish I had followed Irish Dragoons advice and not bought.

That's a cheap shot Yohan. I made some specific comments about AI, since that's the point of discussion here. You didn't mention if you actually agreed or disagreed with any of the points I made?

SC2 "was" unplayable, or "is" unplayable with the latest patch? Is far too easy to play an initial release of a game, criticize the AI, and then proceed to ridicule the game. Many are doing just that with CEAW. It's a bit unfair, based on unrealistically high expectations for what a computer opponent is capable of. How nice it would be if all games were released bug-free and with perfect AI. (That's another issue.)

Truth is AI development is hard. Especially for a grand strategy game like this with its many combinations and permutations of possibilities which need to be scripted for. SC2 still has some weaknesses, even after years of development since SC was first released. You expect perfection? CEAW is brand new, with many of the same AI complaints players had about SC when it was released. If you expect CEAW AI to become perfect in a few patches then I've got a bridge in Brooklyn to sell you. I expect Slitherine will eventually get there but it's going to take time. Give them time.

It's true I have been a playtester for SC/SC2/SC2-WaW, full disclosure there. Do you think maybe I have some insight after all these years? I'll admit to being a fanboy of the WWII ETO grand-strategy genre, which is why I locked onto SC when it appeared years ago. If Slitherine had released CEAW then, I'd probably be hip deep in CEAW, but SC was all there was back then. After being disappointed with Computer Third Reich back in the 90's, my quixotic quest has been to find something like Advanced Third Reich for the PC with a challenging computer opponent, or to use a game editor and make it myself. And I am doing exactly that, with the new SC2-WaW editor.

Back on subject. I believe the combination of generic AI plus event scripting plus AI scripting can and will result in some challenging computer opponent games for this genre. SC2 and GGWAW are doing OK. CEAW will get there. I'm optimistic that WiF will get there, but that's going to be a beast of an AI. And by "there" I mean challenging, not perfect. You still need a human opponent for the ultimate challenge.

[Erik Rutins]

ORIGINAL: jcorbin
1) Launching a coordinated large-scale amphibious invasion.
2) Effectively responding to a large-scale amphibious invasion.
3) Changing a nation's military unit production to counter what is actually occurring on the battlefield.
4) Changing a nation's technological research path(s) to counter what is actually occurring on the battlefield.

Out of curiosity, have you played Gary Grigsby's World at War: A World Divided? Significantly more complex and takes more time to play than CEAW, but some folks swear by it and the AI is definitely above average. You can never really have too many WWII grand strategy games. [;)]

[decaro]

I'd be interested in why you think BoA has a simpler ruleset than CEAW.

I didn't say it had a simpler ruleset/engine, but BoA doesn't encompass the entire world or take into account a myriad of unit types -- only infantry, lt. infantry, militia, rangers, Indians -- no air combat whatsoever, and only frigates and transports, not including its worthless bateaux.

In fact, BoA has so few "pieces," I'm tempted to compare it w/my chess computer; the longer the AI "thinks," the more effective it becomes.

As for difficulty settings, BoA can compromise on FOW, fighting ability, etc., but supply rules and wx are set in stone, at least until the next patch!

[targul]

"Out of curiosity, have you played Gary Grigsby's World at War: A World Divided? Significantly more complex and takes more time to play than CEAW, but some folks swear by it and the AI is definitely above average. You can never really have too many WWII grand strategy games. [;)] "
 
Own the game and have played it.  AI is absmal as is game.  I regrettfully bought that pig while waiting for HOI2 since they released it a couple days prior.  All of Grigsby stuff is ameturish but this was one of his worst.  I swore not to buy his stuff before WAW's release and again I bought the oh this one works.  Hopefully I will never again make such a horrible and disappointing mistake.

[IrishGuards]

This is just my opinion .... I hardly ever play against the AI ... other than to see what happens .. damage .. tech .. or develop a strategy ....
As far as other games go .. I am not here to say what a Game Developer wants when they design and put a game out ..!!
I am a Wargamer ... This means I play against people .. Yep .. Might be a strange concept .. Dunno .. [X(]
They might even be throwing dice around and smack talkin .. OK ...
All I know is that there is no way a computer has the "Experience" of playing some of the best games to ever see the light of day .. [:)]
Has the Comp played .. WiE .. TSS .. WiF .. EiA .. WV .. The scope and scale of these games are massive ..

Give me an AI that that can even remotely begin to understand .. what the hell is going on .. I will play against it anytime ..
Most of the best games .. and by that I mean ... games that can develop .. have a good enough baseline that with adaptability and flexability by both the game and the game player .. will continue to have flavor ...
If you have a game like World in Flames ... [&o] ... and you have played this game .. I know guys with 35 years Exp in wargaming and because each Start is a variable presents new and different aspects of what the pointy units in your OOB can do ..
If you think that a computer will even begin to understand ..
Economic .. Political or Military .. great .. but once they are all heaped together .. what a riot .. [8|]

A strategic level game is .. shall we say strategic ... [:D]
Until the first turn happens and the results of your " Strategy " .. are in effect .. [:'(]
You can have the best AI or strategist in the world .. This does not mean they will be able to keep up with a seasoned Wargamer ..
The Tactical situation almost always takes precedence compared to the Overall " strategy" ...
I know by Experience .. how you gonna give that to a Comp ..[8|]
I play games for lots of reasons as I am sure others do ..
But .. Once the first turn hits the ground .. all those plans and such you have made ..
I'm gonna build this .. do that ... go right out the window .. why because you adapt to what is happening on " Board ".
IDG

[Warfare1]

ORIGINAL: geozero

Okay... so since we are dealing with a STRATEGIC level game, perhaps one way to look at the AI issue would be sort of similar to real life military chain of command...

Have several "layers" of AI decision making. Upper most level would make decisions or "orders" to t he lower layer, then that layer would take info and make decisions, and so on... let me explain in layman's terms.

Suppose you play AXIS and the AI opponent is the Allies.

Typically, Germany invades Poland first. The Allied AI upper layer decides to a) declare war, or b) not declare war. IF decision A is made then it orders it's military and production to a war footing (If "B", then Allies remain Pacificst for now). This changes IF Germany declares war on Allies of course... assuming that decision A is taken, the Allied AI now has to look at the Situation... there's no combat yet in Africa, so nothing is done there. But a lower AI layer could decide to A) use bombers to attack Germany, B) have France attack in the West, C) send out the fleet in Search & Destroy missions against Axis subs and surface shipts, etc, etc.

Let's say decision B above is activated, Now a lower level AI decides which French units will attack and which will remain in defensive positions. Which targets present a better attack outcome? An undefended German city? A weaker Axis unit? etc.

That's how AI should work. Yes it takes programming. If, Then, Else commands as noted elsewhere. When enough of these scripts are programmed the opponent will make very good decisions, whether it is what type of production to build, where to attack or defend, when and where to invade, etc. We were programming this kind of stuff way back in the 80's using Basic.

Puts on fire retardant suit...[:D]

Good post.

CEaW is programmed in JAVA. Does that make a different to say, using C++? Forgive me, since I don't know anything about programming.

Does using JAVA mean that things such as "IF, THEN, ELSE" can't be used?

[Warfare1]

ORIGINAL: targul

"Out of curiosity, have you played Gary Grigsby's World at War: A World Divided? Significantly more complex and takes more time to play than CEAW, but some folks swear by it and the AI is definitely above average. You can never really have too many WWII grand strategy games. [;)] "

Own the game and have played it. AI is absmal as is game. I regrettfully bought that pig while waiting for HOI2 since they released it a couple days prior. All of Grigsby stuff is ameturish but this was one of his worst. I swore not to buy his stuff before WAW's release and again I bought the oh this one works. Hopefully I will never again make such a horrible and disappointing mistake.

targul:

Did you buy the World Divided version of the GG game?

I had it earmarked for purchase, since I thought the game had matured to the point where the AI would be a good opponent.

Where, in your opinion, does the AI fail? Is it poor even on a higher difficulty level?

Thanks.