Lanzfeld Posted February 1, 2000 Share Posted February 1, 2000 Just a question about the AI in the final product. When we build our own maps, how will the AI know what to defend and what is important? Will the placement of the objective flags clue the AI into what it is to go for? Lets say I make a map with a bridge or road through the woods checkpoint, will the AI recognize this as a nice place for an ambush? Thanks, James Link to comment Share on other sites More sharing options...
von shrad Posted February 1, 2000 Share Posted February 1, 2000 Good Question [This message has been edited by von shrad (edited 01-31-2000).] Link to comment Share on other sites More sharing options...
Guest Madmatt Posted February 1, 2000 Share Posted February 1, 2000 The AI is pretty smart but it does key on the the objective flags and a good scernario designer will learn where to put them to make the AI do what he wants. There are also tricks a designer can do with terrain to make the AI do something specific (i.e. line an axis of attack will impassible vehicvle terrain to force the AI to attack up a lane for instance) Once you guys get your teeth into the editor you will learn how to get the most out of it but like anything you will need to test your creation until you get the requireed results. Madmatt out... ------------------ If it's in Combat Mission, it's on Combat Mission HQ! combathq.thegamers.net Link to comment Share on other sites More sharing options...
Guest Thomas William Davie Posted February 1, 2000 Share Posted February 1, 2000 MadMatt; I thought one of the explicit design decisions was that the AI would NOT key in on objective flags???? How's this for a question; what happens in a game if you do NOT assign any objectives? Man, I'm bummed about what you've just said. It sucks. Tom Link to comment Share on other sites More sharing options...
Guest Big Time Software Posted February 1, 2000 Share Posted February 1, 2000 Tom, Don't take Matt's words litterally. Yes, the StratAI pays attention to the flags. If it didn't, how the heck could it ever hope to win? Think about it. Humans have to pay attention to flag possession as well, or they too will lose the game (most likely). What I am sure you are thinking about is suicide charges to capture flags. No real tactics, nothing remotely "strategic", just blind dashes to take/hold flags. This is how most of the AIs out there behave, but it is NOT something you will see in CM. Hardly. The AI does not pay attention to flags any more than you or I would. If it can take one it will, if it can't it will shift its forces towards something different. It will NOT just make a beeline for the nearest flag and call it a day. If a map has no flags at all the AI will not have a clear sense of purpose, and neither will the human player. Both will be equally confused as to what it should be doing, either on the offensive or defensive. Just like a human, the AI will try to make the best of it. If defensive it will set up facing the expected enemy advance in the best positions possible. If on the offensive it will assault more or less in a broad offensive until it hits resistance and then adjust accordingly. Again, JUST like the human in the same situation would. Er... hope that makes sense. If not, please figure out where the problem is because we see none Steve PS There are no flags in Operations, so the above can not be BS or Operations wouldn't work. And they work just fine, thanks very much [This message has been edited by Big Time Software (edited 02-01-2000).] Link to comment Share on other sites More sharing options...
Doug Beman Posted February 1, 2000 Share Posted February 1, 2000 That makes perfect sense, Steve. Battles with no clear, concrete objectives are nothing more than "kill the other guy" and I, personally, have had enough of that playing RTS games. The Brits learned in N Africa the hard way that concentrating solely on eliminating the enemy, while neglecting possession of vital ground, is a very bad thing. This tactic repeatedly allowed the Germans to lure the Brits into killzones, and then kill them with firepower emplaced on the same vital ground the Brits were ignoring. DjB Link to comment Share on other sites More sharing options...
CoolColJ Posted February 1, 2000 Share Posted February 1, 2000 Steve But don't operations have objectives like make an offensive push and capture a town? If so, wouldn't it make sense to have a flag in the town? ------------------ CCJ aka BLITZ_Force My Hompage ----> http://www.geocities.com/coolcolj Double your immersion with my Tweaked Textures and Saving Private Ryan sound set mods for CM!! Check out my music while your there! Link to comment Share on other sites More sharing options...
Guest Thomas William Davie Posted February 1, 2000 Share Posted February 1, 2000 Steve; it makes perfect sense, and no there is no problem, but my question should perhaps be rephrased 'How much emphasis will the Strategic AI place on the acheivement of objectives .vs. dealing with tactical threats or avoiding going into what *MAY* be a tactical threat'? I thought that this was one of the purposes of fuzzy logic, no? What would I do with no objectives? Recon, recon, recon and lots of arty that's what. Air power too, if I had it. suppose there's a nice fat juicy objective at the end of a nice roading winding through some dense woods. I know what you or I would probably do, but is the Strategic AI going to deal with this in an intelligent manner? I *do* understand that you're not utilizing waypoints, but I fail to see how randomly generated games in Combat Mission will be any different than in any other tac WW2 game other than an avoidance of the victory hex rush (or victory objective rush). Again, I do understand that a randomly generated game is different than one that is carefully constructed by a human, but so what? is there a priority on dealing with a tactical threat or a priority on acheiving objectives? [On taking a city; I guess there's a lot of failed objectives in Grozny] Tom ------------------ email address: thodavie@videon.wave.ca Link to comment Share on other sites More sharing options...
Guest Big Time Software Posted February 1, 2000 Share Posted February 1, 2000 CCJ, there are no objective flags in operations because the end purpose is either to hold the whole map and/or eliminate the enemy. Possessing x hill or y block is irrelevant. Tom, the StratAI behaves with a great deal of flexibility. While it does indeed try to go for flags, it will not do so blindly. It might take a round about way because it offers a better/safer approach, it might do a double envelopment, or it might just line 'em up and go for it. Totally depends on the map's terrain, the forces at its disposal, and what it "feels" like doing at the time. As far as the "hex rush" problem, if CM doesn't have it in hand crafted scenarios then why would it have it in randomly generated ones? To the StratAI there is ABSOLUTELY no difference between the two. Even though the StratAI and OperationalAI have undergone many improvements since the demo, the behavior overall in the demo is pretty dynamic and yet focused. If it wasn't people wouldn't be able to replay the thing a dozen times against the AI, yet that is exactly what people are doing. So in fact you have some "proof" right in front of you. The Strat, Operational, and Tac AIs all work together to kick your butt. The StratAI pays attention to what the others want to do and also what they are capable of doing as a whole. If a unit is in a great killing spot and is laying waste to the enemy it will probably not be moved no matter what the StratAI wants. Conversely, if the StratAI wants a whole Platoon to attack on the left flank the Operational AI will decide how to do that as best as possible. It can even recind StratAI orders if they are "out to lunch". This then might cause the StratAI to change its plans by bombarding something, moving reserves, calling off the attack, pushing somewhere else, etc. In short, there is no magic formula. The Fuzzy Logic base simply does what it feels it should in the given situation. In one case that might be to do an all out last push for a flag or it might be to encircle it and not move an inch further. All depends, just like it would for a human. It has human like reactions to the game as it develops. No scripting either in the general sense or in the map specific sense. And that is why CM's AI kicks ass. I've played plenty of games and I never know what the bastard CM AI is going to do, and when it does it I am often caught off guard. Not because it is swamping me in some sort of lemmings like attack, but because it is flanking, supressing, laying down covering fire, establishing killing grounds, etc. Moon just had a battle where one AI tank was constantly driving back and forth behind some houses sniping at his vehicles, at the same time scoring kills and not allowing Moon the chance to get effective return fire on the bugger. No AI I have ever seen in my entire life can do something as devious and spontaneous as that. Steve [This message has been edited by Big Time Software (edited 02-01-2000).] Link to comment Share on other sites More sharing options...
Gromit Posted February 1, 2000 Share Posted February 1, 2000 Steve: Thanks for the AI detail! Now you have got me slathering again... Patiently waiting, Gromit Link to comment Share on other sites More sharing options...
hnh3_cm Posted February 1, 2000 Share Posted February 1, 2000 Why all this talk about Artificial Insemination? Link to comment Share on other sites More sharing options...
Guest Madmatt Posted February 2, 2000 Share Posted February 2, 2000 Umm, yeah...What Steve said... Thats what I meant all along... Madmatt out... p.s. Who would EVER take my words literally?!? ------------------ If it's in Combat Mission, it's on Combat Mission HQ! combathq.thegamers.net Link to comment Share on other sites More sharing options...
Guest Thomas William Davie Posted February 2, 2000 Share Posted February 2, 2000 ----quote---- It might take a round about way because it offers a better/safer approach ----end quote---- Can I take this to indicate that the Strat AI has a specific ability to interpret terrain features in a heuristic sense? I mean the ability to look for certain patterns and tendencies; if not I don't see how it could do what you say. But then, heuristic patterns are used in advanced virus scanners, so I could believe this to be the case. ----quote---- As far as the "hex rush" problem, if CM doesn't have it in hand crafted scenarios then why would it have it in randomly generated ones? ----end quote---- Because Steve, there *IS* a difference between randomly generated scenarios and ones that are handcrafted or tweaked. There is no play balancing in a randomly generated scenario by definition. In a hand crafted scenario, you could eliminate it entirely by superimposing terrain between starting locations and objectives *if* the Start AI is able to take terrain into account when formulating a movement path to take. Hex rush is a bad term; call it objective fixation or what have you; it will be more important in randomly generated games than in hand crafted ones by virtue of the fact that the hand crafted ones have been tweaked for the purposes of play balance. But, I'm willing to accept that the Strat AI doesn't go after victory hexes like a dog in heat. And unfortunate reading on my part, if not perhaps an unfortunate comment on Matt's. ----quote---- Even though the StratAI and OperationalAI have undergone many improvements since the demo, the behavior overall in the demo is pretty dynamic and yet focused. If it wasn't people wouldn't be able to replay the thing a dozen times against the AI, yet that is exactly what people are doing. So in fact you have some "proof" right in front of you. ----end quote---- I've found watching what I perceive to be the Tac AI an endless source of fascination. I'll literally play a game step by step, doing nothing on my part, and watch how the program responds to changing conditions. My hat's off to you on this part. But the Strat AI? Well, I'm aware of the fact that I've only had access to the beta demo, and yes I'm aware of the fact that it has changed orders of magnitude since then, but to be honest I found it to do no more than dither about. Again, I'm aware that this was a beta demo, so this is not an intended insult. I found it utterly incapable of acting with any sort of concerted action or plan. Not impressive (as opposed to the Tac AI, which as I mentioned a book could probably be written about). I sure as hell hope you can pull it off, and I'm not in any hurry to rush you. Tom ------------------ email address: thodavie@videon.wave.ca Link to comment Share on other sites More sharing options...
Guest Big Time Software Posted February 2, 2000 Share Posted February 2, 2000 Hi Tom, Yes, the StratAI does look at terrain using heuristic assessments. Not as detailed as we would like, but these functions are massive CPU killers. So the answer is that it does what you think it does to the best of today's CPU power. To phrase it in a way that everybody can understand, the StratAI does not just move guys dumbly from starting location to victory location. Instead the paths it chooses to get to the victory location are chosen using logic based on its setup, task, terrain, etc. Not perfect, but it is totally dynamic. This is why nothing has to be imbedded into the maps to make the StratAI get from A to B in a sensible (though predictable) way. The StratAI is no more or less likely to get objective fixation than a human player. I still disagree about the difference between human and randomly generated maps in terms of the StratAI's performance. I for one have never felt the need to tweak a map to make the StratAI's life easier. If I make a tweak it is for general game balancing, which does not have anything to do with the StratAI directly. In other words, if I think the defender needs a better view of the attacker's main march route, and so I remove some trees, this has nothing to do with the StratAI as the change is there just as much for a human player as for the computer one. Also, there are likely to be the same number of objectives in a hand made scenario as a random one, so the Hanzel and Grettel concept of paths for the Strat AI to follow isn't an issue. Generally there are between 2 and 4 Victory Locations for a battle, which is hardly enough to make the StratAI take a specific course of action. The problems you see in the Beta Demo boil down to a lack of operational coordination. The StratAI's logic used for march/attack routes and counter attacks is being compromised by the OperationalAI. In the Beta Demo once units hit significant resistance they fall in love with the concept of firing from a safe distance This is the OperationalAI at work here, not the StratAI. Like I said, the Tac and Op AIs can override the StratAI. In the case of the Beta Demo, too much so This is the area we fixed the most. To illustrate what I mean is this (think Last Defense, German AI)... The StratAI plots advance routes and assigns missions to the various units at its disposal, as well as their starting locations. Then they go along their merry way using the Operational AI. When a task force achieves its objective or needs the current one to be changed, it asks the StratAI for something new. The StratAI can also change the tasks and routes based on changing conditions as the game progresses. The OpAI is then obligated to follow these changes within certain parameters. All along the TacAI is there to help individual units behave when meeting enemy units. Again, the TacAI is the only part that is available to both human and computer players alike. OK, say the right flank bumps into a bunch of units behind the wall. The OpAI puts the StratAI's orders on hold and sets up positions to engage the enemy. The TacAI may or may not do this before the OpAI has a chance to. Now... here is the problem with the Beta Demo. The StratAI then realizes that it needs to do something so it "asks" for volunteers based on their abilities to perform the desired function (assault usually). Unfortunately, the OpAI keeps units from raising their hands and instead instructs them to keep on firing. Units not already engaged, lightly engaged, or of nominal combat value at long ranges, are tossed into a hasty attack formation. So HQ units and spotters are OFTEN selected because they are genearlly the least likely to be effectively engaged. That is why you often see these units leading attacks. Now, with each "wave" of uncoordinated attacks, the StratAI gets a bit less likely to listen to the OpAI and starts ORDERING units to disengage. However, in the Beta Demo the StratAI is a "wimp" and too often takes SCREW YOU for an answer So maybe it manages to get only one in nine squads to disengage, then sends it forward. You know what happens next! Compounding this problem in the Beta Demo is that the StratAI is not as skilled at looking at its watch to see what units can do what and when, so it far too often will get two volunteers and send them forward "ASAP" instead of holding one back to wait for the other. OK... So where does the AI as a whole stand right now? See all the problems noted above? Wonder why we are aware of 'em in such detail? 'cause we fixed the root problems to our satisfaction (we would of course love 6 months to do nothing but AI ). The AI doesn't beahve perfectly in all situations, but it certainly behaves MUCH better than it does in the Beta Demo as a whole. Let me put it another way... I just played my first human vs. human battle against Fionn, as the Germans, in a city, where he had Panthers and I had crappy Sherman 75s. I won, although it was a very tough fight. Fionn is as good as he appears to be. A little more bad luck for me and I would have lost. Yet the AI has beat, or nearly beat, me on several occasions. At the very least it has challenged me greatly. Again, it ain't perfect, but it will clean your clock if you don't use your head as if you are playing a human Steve [This message has been edited by Big Time Software (edited 02-02-2000).] Link to comment Share on other sites More sharing options...
Zamo Posted February 2, 2000 Share Posted February 2, 2000 Gentlemen, esp. Steve, This thread has filled me with a born-again sense of excitement regarding CM. The last week or so I have been more or less "coasting" in my anticipation. To hear that the AI is going to make it's decisions like this has got me all a-giggle again... Will this blasted game NEVER reach my sweaty little paws??? Zamo Link to comment Share on other sites More sharing options...
Fionn Posted February 2, 2000 Share Posted February 2, 2000 Ahem, I might point you to look at the score Steve . Your "celebrations" are a little premature methinks . The game was a draw from what I can see .. See, (and I knew this better than you since I've PBEMed a lot ) CM gives a LOT of points to the side with the greater kill total. Since I had a 2:1 kill ratio but lost the Victory Locations I still would have gotten a draw ... CM REALLY does lead to phyrric victories. If I'd evacuated all my men off the field at that stage I'd still have gotten a draw due to casualty levels alone . Uurah Now forgive me as I mutter about Sherman 75s and the damned "Front Lower Hull" penetrations they keep getting on my lovely Panthers LOL . Seriously though guys, as a gamer I've got to say that the AI in CM thinks like you and I do. It gives weight to objectives but doesn't follow shortest path routes etc.. If you look at the course of a game from an overhead view you'll actually see that the AI actively searches for your flanks and tries to bring forces to bear on your flanks and pass other forces into your rear. I've lost count of the number of times that I've found enemy tanks making their way through scattered trees and forest paths 600 or 700 metres away from the nearest fighting as the AI tries to get around my flanks. Hell, in one game where I lost 9 tanks and 3 HTs ( 3 Panthers, 6 Pz IVs and 3 HTs) in the first 70some seconds I lost virtually all to 2 57mm AT guns.. The AI had put 2 M10s and a 57mm AT gun directly in front of my start positions and then put one 57mm AT out to each side ( about 45 degrees left and right of my position). When I started to move out the ones on my flanks picked off the Panthers while the ones in front went for the Pz IVs. I'd have been surprised if any human player had picked such good ambush shots which maximised the potential for getting side shots. It won't beat you all that much since it sees every scenario with NO knowledge of how the scenario plays OR how you play.. As you play the AI you'll get to know how it thinks etc so you'll have a major advantage over it BUT I must admit I've seen a lot of good to average players doing more silly things than the AI and I think the AI really does rate as an average to good human player who will push you close in most battles and even win a few if you make mistakes. It's the first AI which I've ever been beaten by in a "fair" game (IOW being beaten when you're outnumbered 20 to 1 isn't a surprise but being beaten when both sides are equal is fair). It doesn't beat me often but every so often it gets a lucky break somewhere and then it capitalises . It's hungry to win. I think Steve and Charles don't want people expecting "Deep Blue II" BUT the AI is a damn sight better than I've seen in any commercial wargame and I wouldn't worry about it being too easy if I were you . [This message has been edited by Fionn (edited 02-02-2000).] Link to comment Share on other sites More sharing options...
CoolColJ Posted February 2, 2000 Share Posted February 2, 2000 Yeah I think the AI will be nasty in randomly generated games, operations and any scenrio you have not played yet! It will surely kick my booty Especially since I have not played a single PBeM yet! Steve your a genius! All your AI posts should be compiled togther to forma TGN article/FAQ I love it - everytime you write one of these AI posts, I get all warm and fuzzy! ------------------ CCJ aka BLITZ_Force My Hompage ----> http://www.geocities.com/coolcolj Double your immersion with my Tweaked Textures and Saving Private Ryan sound set mods for CM!! Check out my music while your there! [This message has been edited by CoolColJ (edited 02-02-2000).] Link to comment Share on other sites More sharing options...
Guest Thomas William Davie Posted February 2, 2000 Share Posted February 2, 2000 Fionn; I've played the scenario 'The Last Defense' 3 times. Once .vs. you, and you handed me my ass on a platter (congrats). The other time .vs. a human was against Lee Contini, and although I got a draw, felt that he handily won the game. The only time I've played solo in a complete game against the AI was the very first time I played, and I got an overwhelming victory by doing nothing (as the Americans). But, since then, I've stepped through the scenario about 35-40 times completely. Sometimes as the US, other times as the Germans and some times dual pbem just to see how the computer will play itself. Again, I think the Tac AI is a study in extremely interesting behaviour. It is really niec to see how the game reacts to changing micro conditions. However, having watched it's overall 'plans', I'm not impressed at the Strat AI. Me thinks that it dithers to a large degree. I see no sense of coordination in terms of attacks. I do note though, that this is the beta demo, and it is only one scenario. At the very least, it is promising, and I make the assumption that for a demo that is essentially 3 months old at this time, things have been vastly improved upon by now. I'm not knocking the game; I'm making comments based upon what I can see. Tom Link to comment Share on other sites More sharing options...
Darwin Posted February 2, 2000 Share Posted February 2, 2000 Thomas, you are aware that there are 3 scenarios avaliable to you. Right? Link to comment Share on other sites More sharing options...
Guest Madmatt Posted February 2, 2000 Share Posted February 2, 2000 Ahh, It seems perhaps what I said in the earlier post was misunderstood and looking back over my words I can see how easily that could occur. I did not mean to say that the AI blindly goes after objectives, what I meant was that the AI takes them into consideration just like a human would. But as Steve said the AI is not driven to obey the larger scale aspects of what a Objective represents and will sometimes heed to the more direct and local (tactical) situations. Allow me drive this point home with a little example from a scenario I was testing. This scenario had a column of German troops marching up a railline with trees on both sides. The track opened into a wide and relatively open clearing with a few scattered buildings. The whole area (other than the open field) was dense woods and forest with a few scattered dirt paths going this way and that. BTW this was the shown in a recent POTD on CMHQ The traintrack continued past a small group of buildings and then exited at the top of the map. In the field was large force of American infantry (having breakfast it seems) and a few support vehicles. When I played as Germans I quickly realized that I had a much superior close in firepower advantage and made a bold assualt across the field with motars and MG's covering my advance. Casulties ran high and I was involved in some perilous close combat but I eventualy took the main objective (some buildings) with my main attack while a flanking attack through the edge of the tress ran into some tough resistance near a secondary objective. Eventualy morale on the Americans started to waver and after an aborted counterattack they were bloodied enough to surrender. I on the other hand was just as beaten and nearly out of ammo, it could have easily gone either way but luck shed on me. This was one of the funnest games I had yet played of CM. The most fun when I then played as Americans as the AI totally wowed me. I played again and took command of the Americans and now I knew EXACTLY the forces up against me. I deployed into a good defense in depth and even was able to establish a reserve counterattack force of a platoon to exploit any weaknesses I saw in the German assault. I knew from experience (having just played it) that the Germans had awesome close range firepower and my best chance to win would be to keep them at a distance and fight my battle. The AI was not in the mood to play that way. After the AI ran into my advance scouts they quickly deployed into the woods along the train tracks. MG42 fire flashed out from the trees and my own 30's answered in turn. I kept a good portion of my forces hiding in cover and buildings waiting to see where the AI was going to press. Well the Germans started dropping mortars on my main position and used this suppresive fire to manouver. After a few turns with no NOTICABLE attack movement on the AI part I began to get cocky. I started thinking that perhaps my own fire was enough to keep the poor Germans hiding in the woods unable to move out..BOY WAS I WRONG...A few more turns went by and there was a small advance made made by the Germans of a platoon or two but no concerted push...By turn 11 or so I began to redeploy my reserves toward my center as I fully expected some sort of mad rush across open ground (the same attack that I had made to such success in the previous game) or at the least a advance through the woods closest to the objective. On the next turn a lone team way off near the edge of the map detected some 'INFANTRY SOUNDS' on my extreme flank...Ahh, that must be a routed MG team or something I thought... There was no way the AI would have followed the map edge when the objective was clearly straight ahead from the start positions. The next turn those infantry sounds morphed into an entire COMPANY of charging Germans and rolled up my flank with ease! The AI had snuck 10 turns through thick trees and undergrowth to come upon a small (and undefended) access road that led into the heart of my defense. I had not bothered with flank security as I never expected the AI to perform such a move. The point is even though I knew EXACTLY what forces the Germans had and had already defended for what I thought was the best approach for an assault the AI came upon a completely different and in the end more tactically sound decission. With the limited range of combat in woods the AI tore me apart with its Submachine gun armed squads and eventualy routed my main defense line. It was only with a counterattack with my reserves that I was able to blunt any further advance to the secondary positions and casualties were extremely high. In the end I think managed to pull out a draw, not because of any tactical finesse on my part but for the fact that the AI was occupying itself with killing all my scattered remains! It was a very very very bloody battle and everyone was out of ammo and close combat was a common occurance. I had several units with only 1 or 2 guys holding out under fire while whole squads would break anmd reform in the woods..VERY NASTY...A truely wonderful scenario and you can thank Moon for creating it, but it goes to show you that the AI that is in the current build will smash your freaking face into the pavement if you dont act smart. Because of my familiarity of the sceanrio I thought play as the Americans would be a cakewalk and I got cocky and the AI for all practical purposes kicked my ass...I left the door open on my flank and was blind to it and the AI exploited it, albeit in a very time consuming and roundabout way but the effects were devastating. Funny thing is, when I played as Germans I totaly discounted that approach as untenable due to the distance, terrain and the fact that it was nowhere near the main objective... The AI saw that too but also saw that it led right into my exposed flank... I hope that sheds some light on the AI from a more 'grunts eye' point of view... Madmatt p.s. And YES you may refer to me as 'FLANK BOY' for 24 hours, starting....NOW! ------------------ If it's in Combat Mission, it's on Combat Mission HQ! combathq.thegamers.net [This message has been edited by Madmatt (edited 02-02-2000).] Link to comment Share on other sites More sharing options...
Justin Hoerter Posted February 2, 2000 Share Posted February 2, 2000 This is the best thread I've seen in a while. Very nice to hear what has changed since the beta demo and the reasons behind it Steve. Also nice to hear the brief AAR from Madmatt. Just can't wait to give it a whirl myself... J Link to comment Share on other sites More sharing options...
Sarge Posted February 2, 2000 Share Posted February 2, 2000 dear flank Boy thank you for an in depth, and fascinating account. question: if you discounted the approach used by the germans, does that mean there is no built-in "amount of ground to cover" calculator? i'm not being a samart-ass, really want to know. too many games allow you to 'calculate' the distance from x to y, and how many turns it will take to arrive. if not built-in in, there is enough information that you can figure it out. so was this a 'gut' feeling (too far) can't do it...or miscalculated? or though you can calculate (in full version YOU are testing) you choose not to? hope this isn't too complex, but i think you see where i'm coming from... thx Link to comment Share on other sites More sharing options...
Guest Thomas William Davie Posted February 2, 2000 Share Posted February 2, 2000 Darwin; yes I'm aware that there are three scenarios, why do you ask? Tom Link to comment Share on other sites More sharing options...
Guest Big Time Software Posted February 2, 2000 Share Posted February 2, 2000 Tom, the problems you keep pointing out that exist in the Beta Demo are in fact OperationalAI problems, not StratAI ones. Although it is hard for you to tell in the Beta Demo, the StratAI is actually making good decisions, but the OpAI is allowed too much freedom. Read my post again to see what was wrong with the Beta Demo and what we did to fix it. It really wasn't all that hard to fix. Couple of days actually Oh, and there were a couple of bugs that caused the StratAI to sit on its can sometimes. Those have been fixed as well. Sarge, Matt discounted that attack avenue because he had a gut feeling it wouldn't be worth trying. Kinda like the French thinking that nobody could come through the Ardennes Steve Link to comment Share on other sites More sharing options...
Fionn Posted February 2, 2000 Share Posted February 2, 2000 Thomas, StratAI was only in for about a month when the demo came out. Now it has been in for 4.. It has undergone at least TWO complete revisions since then. I think it is very safe to say that it acts much more competently now.. I suggest you play against the AI once or twice... For your info playing a PBEM vs PBEM game will NOT activate the Strategic AI. Hence, if you are saying it just mills around you'd be right since there was NO Strategic AI operating in those PBEM vs PBEM games... All that would happen in those games is that the tactical AI would take over and move units if they began being fired at by mortars. So, to recap there is NO AI vs AI functionality in the beta demo... If you want to evaluate the Strat AI (which is only a primitive version of what is now in) you NEED to play a 1 player game vs the computer. I'm right in thinking you thought that setting up a PBEM game but issuing no orders to either side would cause the strat AI to kick in right? Link to comment Share on other sites More sharing options...
Recommended Posts