CRAP AI, Other BUGS ,Fix Or implement IPX/TCP!!

[PeterNZer]

Well, I'll try and help some more

Cpt. S

(a point by point breakdown follows)

- Size of map and pts: well BTS increased the pts more for multiplayer than any other reason, I guess there was some demand there. Think of the ability to use extra points as an optional mod. Just like in Quake or many other games, there's all sorts of stuff you can use and tack on, doesn't mean the game was designed to have those things there. It does seem you have encountered a sitation which the AI doesn't handle well, that is of huge forces on a huge map. I guess you could conclude the AI isn't to well suited to this scale of operatoins, and I have absolutely no problem with that whatsoever. The game is supposed to simulate at most.. what.. a Batalion? So generally that's, waht 2 or 3 companies of infantry, some heavy weapons support and maybe a platoon of Armor. With up to 5000 points the AI can have the option on a lot more stuff and much bigger forces.

Now remember, The tac-AI works fine, that's the AI for handling in-turn stuff if i recall correctly. That's the shooting, firing etc etc. What you seem to be saying is that the strategic/operational AI bit, the 'big plan' AI isn't working so hot with larger troops.

Really what your post boils down to is "It seems the operational AI can't handle the larger maps and forces? eg..<insert examples> Any comments BTS?" And i'm positive if you posted like that you'd have got some worthwhile reply. However your post was pretty jumbled, so the question got lost amongst the vitriol.

Anyway, where was I. Well you have two choices really. Play with huge points and maps vs' the AI, or not. I'd suggest you try 'not' and at the same time, ask politely what's going on or if you are missing something. If you're playing with the default points in a scenario, well you have to expect some odd things to happen. If you are playing the scenario exactly as it is listed with no tweaking, and STILL facing trouble, well you have a legitemate concern.

Meanwhile, I suggest you try a couple of smaller maps and point values and see how the AI does. I've always found the AI to be pretty sharp and much much better than other games I've played. What I like best is I will see other players in multi-play make mistakes the AI would -never- make. That's pretty cool.

- ICQ: Well I've had icq since it was released and not had a single problem. If you want it, just get the latest version, turn of 'show ip' turn off 'publish to white pages' or whatever it is, then the only people who will know you are online will be those who are already on your list.

If PC security is a general concern then you need to pick up.. black-ice or whatever that program is.. it monitors all ingoing-outgoing traffic as well as pings, fingers, malformed packets etc etc.. and is very effective at stopping crap happening. Even without ICQ you are VERY likely to be probed. I could sit my PC for an hour and find up to 20 probes from all over where the script-kiddy has run probes over thousands of random IP address. So, moral is, got **** you need to save and keep secure, either get a firewall or BlackIce. If you haven't got those already then installing ICQ would be like adding a pinhole to a piece of swiss cheese. (cool metaphor huh?)

(o, i'm in london by the way, was in Auckland).

- PBEM: I dunno why you find it so annoying. I like to surf, get a turn arrive by ICQ autodownload, go play, pop back, surf some more and send it. It's pretty relaxed

I think one thing I'm not clear on is whether you are using the default points or not? If you're not, well, you have to expect odd things to happen.

Perhaps if you could clearly outline the scenario, pts per side, and AI's actions, we can then all compare our own results.

eg

1) Whooping o' the Asses

pts. 4000 each

result: ai creamed my pasty butt on turn 3

At the moment it's .. well it's bassically really hard to decipher what you're getting at from your posts. So a few clear examples and we'll see. If it turns out others are seeing much the same thing we have a pattern and Steve/Charles can pop in and say what they think. If it's a case of you being the only person who is seeing that problem. Well who knows! Or if it's an intermitent problem, well you can't have everything. I don't think a better game AI has been written. But feel free to try yourself.

PeterNZ

[Henri]

<BLOCKQUOTE>quote:</font><HR>Originally posted by CPT STRANSKY:

SS vs North novas battle. the computer doesnt even attack me with infantry till about the 35-36 turn. 40 turn game.

This has happened 4 times now.

<HR></BLOCKQUOTE>

Well, it happened more than 4 time is WW2.

Once in a major offensive, half of the Canadian forces made a wrong turn and ended up miles away from their objective. It took them much longer than the duration of a scenario before they figured out why there were no Germans opposing them.

Similar things happened to others including the Germans, when orders got mixed up or didn't arrive, or when they received contradictory orders, or when Murphy's law just took over.

I haven't played this particular scenario SS vs Novas, but All or Nothing should be played as the British; even a human has trouble managing all the British convoys in this scenario, so it is not a wonder that the PO would have trouble too.

I don't know why you keep playing this scenario when you find it so rotten: are you playing for fun or trying to make a case against the game? If the former, try a scenario such as Fear in the Fog, Villers-Bocage (as the Germans) or Lelorey (available at various sites), if the latter, you are wasting your time

Excuse me if this sounds patronizing, but as others have pointed out, posting a message with a confrontational tone is not conducive to a calm discussion (when one spits in the air...). Bang on your computer screen, throw your CD against the wall, yell at your wife or any passer-by, then take a deep breath and come on this forum to discuss calmly. Try it, you'll like it

Henri

[DmanTblad]

I've forced myself to read thru Cpt Stransky's posts a couple of times. I think he is complaining about what PeterNZer hit on in his second post here. That is the function of tac AI compared to strategic/operational AI. I've seen BTS post somewhere that there are 3 different levels of AI in the game. And, that the Tac AI is the only one that is completely finished. Could explain why the computer fairs better on small maps and not quite so well on larger ones.

Like everyone always says, do a search and most questions/problems have already been answered in great detail.

I for one, have no fears that BTS won't stop working on improving the game. Just be patient, and your large battles will surely get more interesting.

[Cueball]

I struggled through the post as well. There is another valid point that was mentioned.

I had the computer assign me my forces several times, then went through each unit and added up the point totals.

Often times it came up short. Not 20 or 30 points, but sometimes 200 in a 2000 point game. I couldn't even attribute that to air support, as Axis air support is 300+

I don't really mind it so much, as the variance makes it interesting, but there could be a bug with the automatic force selection process. For 200 points, I could have had another Panther

[This message has been edited by Cueball (edited 08-30-2000).]

[Sparks]

Um, so wheres the amiga version??? <cackle>

yes, i was a Amiga guy till the bitter end, and now im paying for it, i should have wasted my time with DOS and win so that i would be making more money today.

An amiga 1000 should be included so we can shoot at it or use it to chock jeep wheels heck the cpu wont even put out enuf heat to keep our UK friends tea warm.

Again i have to side with the team here, the site you purchased the product said in plain english NO REFUNDS NO RETURNS AS IS, your lucky they are not money grubbing half baked coders raping our money belts like so many others that are ruining the software gaming market.

Sparks

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

Its not the end of the earth, but i can see it from here!

[RMC]

"I want to win the Iron Cross."

"here, you can have one of mine..."

Why on earth anyone would pick the Prussian ass from "Cross of Iron" as a online monicker is beyond me. Do we have a SGT Steiner to be the good guy?

Can't be bothered to read before asking? Well BTS shouldn't be bothered trying to figure out what you are saying in those unorganized "stream of consciousness" posts....

[JoePrivate]

I think the intent of the original poster has some merit. The AI while attacking has some shortcomings, it seems unable to plan, coordinate and execute even a reasonably coherent attack.

While attacking I have observed the AI bunch all his infantry, FOs, support weapons etc in narrow avenues of approach and mill around, going forward and backward, for several turns, an obvious and easy target for artillery. Instances also when I have looked at the AI's forces and it has not kept the platoon's command structure intact at all. It doesn't seem to adhere to the principle of laying down suppressing fire before maneuvering. Personally I have yet to see the AI use HE in it's arty missions, I won't debate on whether that's a good thing or bad thing but at some point the suppressive effect of artillery HE should be used.

Sometimes it will use it's armor to good effect. In the first battle of the Villers Bocage op I saw it try and flank the German right with two tanks and a HT. Many other times however I have seen it commit it's tanks piecemeal through the same ambush zone, again and again, or the same as the infantry mill around like it is unsure of exactly what to do.

On defense or even meeting engagements, the AI is enjoyable and will penalize sloppy play, not a walkover by any means.

I'm sure most of us have seen these things. I'm not a progammer so I can't say how much effort is required to code a better attacking AI for how much gain. I'm not even sure I expect it in CM but there is room for improvement and a discussion is a good way to start.

[Capt. Toleran]

The long, stream of consciousness posts the original poster has been writing kinda remind me of the Unabomber... Does Ted Kaczinski have Net access?

[Guest Big Time Software]

<BLOCKQUOTE>quote:</font><HR>I'm sure most of us have seen these things. I'm not a progammer so I can't say how much effort is required to code a better attacking AI for how much gain. I'm not even sure I expect it in CM but there is room for improvement and a discussion is a good way to start.<HR></BLOCKQUOTE>

This is a good place to give a hint at what lies in the dozen or so AI threads that the good Captain doesn't see any reason to look into...

AI programming is the most difficult thing to do PERIOD. Not just games, but ANY application you care to name. The problem is compounded in games because they require HUGE volumes of AI programming.

The other thing to note is that nobody, no team of PHDs or anybody else, has made an AI capable of human standards for a game of this depth and variability. Deep Blue, although impressive, is not even in the same ballpark and look how much effort went into that just to play one game with set rules, conditions, and no inherent variables.

We could do nothing by code AI for 12 months and it would still not be up to the level of human with a basic understanding of tactics in EVERY GAME IN ALL CIRCUMSTANCES. To suggest that this is somehow possible means that whoever is doing the suggesting has NO idea what they are talking about.

Yes, CM's attack AI has its problems. As we have stated over and over again... it is going to be a fact of life for decades most likely. We are proud of CM's AI because we are more objective. We know what a computer is capable of, we see how bad the AI is in other games, and also do not judge the AI using human standards.

This is not to say that we can't make improvements to the existing AI (we can to some extent), but one has to balance out the time to code AI with all the other things. If we gave you the following choice, what do you think the majority of games would choose:

1. A marginally better attack AI with CM2 in 2 years.

2. A slightly improved attack AI with CM2 in 1 year.

I'd rather go with #2, and I think if the choice were real so would 99.9% of you gamers.

Reality is a bitch to deal with when you don't understand how reality works. Fortunately, we *do* understand reality and that is why CM kicks ass. We didn't waste 6 months of development time only to find out that the AI was not much better than it was after 2 months of work. This is not a problem that can be overcome with suggestions and good intentions. Reality has to be satisfied in the end.

Steve

[Chupacabra]

My stock response to people who complain about the AI -

You think you can do better?

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

Soy super bien, soy super super bien, soy bien bien super bien bien bien super super.

[jpinard]

Stransky is a loser. Get lost you troll.

[Teamski]

That's just it, what would you expect? I've seen some pretty inept AI's out there worse than CM, that's for sure. How 'bout Close Combat. Yeah, now that's a tight AI! Nothing like running a new scenario each time just to find those same ol'Panthers holing themselves up in the middle of town, AGAIN, moving back and forth as if they were rocking themselves asleep, AGAIN. That's a pretty damn poor AI!

I don't know how anybody can create an AI to be human. The AI in CM is about as good as you can reasonably get. I'm sure a few tweaks can be squeazed to get the agression out of it, but with situations being so complex within a battle, it would be tough for an AI to react to them all. It will be quite a while until that "human" AI appears.......

Just enjoy the experience and a few PBEM games..........

-Ski

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

"The Lieutenant brought his map out and the old woman pointed to the coastal town of Ravenoville........"

[eba]

I played a QB the other day, American Airborne glider company defending against SS Armor. Small map, 2000 points, attack, AI as Germans.

The AI attacked immediately with a coherent plan. Tanks on both flanks and infantry up the middle, which made perfect sense given the map -- wheat fields on either flank and heavy, heavy woods in the middle.

Since the routes of attack were pretty clear, I pulled off a great ambush on one flank using one AT gun and a bazooka team. On the other flank, though, the AI kept its tanks about 200 meters back, out of my best AT range and proceeded to blast my position for several turns. I kept wondering if the tanks would ever advance and thought they behaved a little timidly, but on reflection concluded that they probably did the right thing. They finally advanced once their infantry moved up through the woods in my center. They had a great time smashing my infantry at close range until I was able to take them out with bazookas and a Gammon bomb.

In retrospect, I think the AI did a pretty decent job. The tanks didn't come roaring in like in the movies, but doing so wouldn't have made sense under the circumstances. In fact, the only reason I was ever able to get at them was because they finally had to close with me to capture the objectives. They waited until they had some infantry support and then took their best shot. The AI also used a lot of HE artillery fire on my infantry - just about every turn after their infantry made contact.

In the end, the QB just demonstrated, accurately, the weakness of armor without infantry support. The tanks couldn't close in until their infantry showed up and they ended up not having enough infantry to do the job. All in all, though, I thought the AI did a pretty decent job.

[This message has been edited by eba (edited 08-30-2000).]

[JoePrivate]

I hear you Steve, I have read some of the previous discussions. It isn't a big issue for me as CM does so many other things superbly, namely the TacAI which is arguably more important for the WEGO system. I think also most players will eventually gravitate towards play against human opponents. Cheers.

[Joe Shaw]

Steve I think the AIs in CM are very good and better than anything else out there. They are not perfect and likely never will be but for anyone to expect a home computer to compete with a human (with the possible exception of Lower GI Tom, of course) is ludicrous. I think the answer is found in the words of the Capt. himself: <BLOCKQUOTE>quote:</font><HR>ive been wargamming for 20 years and i have yet to find a good War game.<HR></BLOCKQUOTE> And he likely never will in that case. Too bad, he's not seeing the forest for the Tall Pines tiles.

Joe

[CavScout]

<BLOCKQUOTE>quote:</font><HR>Originally posted by jpinard:

Stransky is a loser. Get lost you troll.<HR></BLOCKQUOTE>

Come on now. Who's trolling now?

Cav

[Zamo]

Yeah really. Despite the fact I found his post rambling and difficult to follow (I really hate it when people compromise English due to laziness), he has valid concerns and is due the same respect we would give anyone.

I personally don't share his oppinion, but he certainly has the right to express them in an open forum.

Zamo

[tailz]

<BLOCKQUOTE>quote:</font><HR>Originally posted by JoePrivate:

I think the intent of the original poster has some merit. The AI while attacking has some shortcomings, it seems unable to plan, coordinate and execute even a reasonably coherent attack.

While attacking I have observed the AI bunch all his infantry, FOs, support weapons etc in narrow avenues of approach and mill around, going forward and backward, for several turns, an obvious and easy target for artillery. Instances also when I have looked at the AI's forces and it has not kept the platoon's command structure intact at all. It doesn't seem to adhere to the principle of laying down suppressing fire before maneuvering. Personally I have yet to see the AI use HE in it's arty

B]<HR></BLOCKQUOTE>

Actually, the AI continues to suprise me with its capabilities. QB winter action, probe IIRC... pushed the center and my right... plastered me with HE in the center, THEN pushed a single platoon along the left map edge. (If I didn't know better I'd swear the TacAI SOB was trying to 'edge creep' me, but I don't think it'd try a 'gamey tactic') I'd be curious to know how often the AI is programmed to purchase FO's. (might explain the lack of AI use of arty) Has anyone noticed a difference in how much arty the AI will purchase depending on if it's playing axis or allied? (historically, by '44 German artillery was waning in effectiveness due to lack of logistics more than anything) Attacking or defending? I think this would be as interesting a case study as the guys carrying out their firing range tests. (don't look at me to do it though, I'd rather be playing than experimenting) The CM AI may not be the best... but it is better than anything else to come down the pike in a LONG time. (probably ever!)

Keep your powder dry.

[Wayne]

One solution to the basic problem of AI timidity in attacking is to give it an advantage in force size. I remember playing the CE scenario in the Gold Demo as Allied vs. AI and giving the AI a 100% advantage. I got one good pasting. Clearly the AI shines on defence and that's how I usually play against it. This may get boring after a while but the truth is, that's how most combat actions were, one defender against an attacker. Cm is still the best combat simulator/wargame on the market and I play it religiously, meaning I pray alot when playing. I love it even though I've got a slow cpu and the large to huge games are almost impossible to play. Nothing's perfect but CM comes pretty close.

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

Blessed be the Lord my strength who teaches my hands to war and my fingers to fight.

[gaffertape]

Ooh, I liked that one, Joe:

"Can't see the forest for the Tall Pines tiles".

Very nice.

[aka_tom_w]

This is a great Game

it has the Best AI I have ever played against

and given these two chioces:

'1. A marginally better attack AI with CM2 in 2 years.

2. A slightly improved attack AI with CM2 in 1 year.

I'd rather go with #2, and I think if the choice were real so would 99.9% of you gamers.'

I will happily choose # 2 thanks for the great game Steve and Charles. (not withstanding the non-modeling of Long range German High velocity flatter trajectories and German Zeiss gunnery optics for improved long range accuracy and LOS and LOF traveling right through live and dead non-smoking vehicles)

Ok I 've already been labled a "whinner" on the later issue.

I can live with it.

(posted in good humour)

-tom w

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

<BLOCKQUOTE>quote:</font><HR> "Remember that no dumb bastard ever won a war by dying for his country. He won it by making the other poor dumb bastard die for his country."

G. S. Patton

<HR></BLOCKQUOTE>

[This message has been edited by aka_tom_w (edited 08-30-2000).]

[CavScout]

<BLOCKQUOTE>quote:</font><HR>Originally posted by Zamo:

Yeah really. Despite the fact I found his post rambling and difficult to follow (I really hate it when people compromise English due to laziness), he has valid concerns and is due the same respect we would give anyone.

<HR></BLOCKQUOTE>

Without knowing where he is from, his poor English could be because it is not his native tounge. Who knows..

Cav

[Guest Big Time Software]

Joe,

I meant to make it more clear when I used your quote that I thought you were making excellent points. I reread what I wrote and I guess it could look like I was taking issue with your thoughts. Not so

Cpt. Stransky is from NZ. I have noticed a bit of a difference on how NZers post, but the two posts from the Cpt. were exceptionally difficult to read and follow. The points he raised, thanks to PeterNZ's follow up, are nothing new. That is why I suggested that he read the stuff that has come before. But apparently our words are valuless unless they are freshly posted?

Steve

[Major]

Nail all trolls

Greetings to Lindan! Come on, send me your next turn!

[eba]

tailz:

In the QB I described above, the AI purchased four Arty Spotters: three 105mm and one 81mm. My guys were under arty fire for the last 10-12 turns of the battle.