Jagdcarcajou Posted February 11, 2001 Share Posted February 11, 2001 Originally posted by Tris: I've read what I could re commands and firing and seeking cover and all that. It's all been perfectly circular and we've gotten pretty much nowhere in the process. Big surprise. To sum it up at the moment: these most recent observations are just that, ingame notes from a gamer as he runs across this stuff during play. Assuming there's a solution I hope it's found and that the company moves on it. [This message has been edited by Tris (edited 02-11-2001).] Hey, Hmmm, I see no circle. Its simple: Run and Move orders allow a unit to move and shoot. Period. Sneak will not allow a unit to move and shoot, UNLESS that unit is engaged first. A truck cannot engage you, ever, so a unit ordered to sneak will do so, right up to the thing, and never fire a shot. Always been like this so far as I can tell. If your troops are affected by some sort of morale condition (Panic, Pinned, etc..) their behaviour is not as cut and dried. I threw together a little show-and-tell scenario with the situation you described. Download it and check it out if you like. It will do the same thing over and over. "http://members.home.net/jagdcarcajou/MoveTest.zip" Hope this helps out. EDIT: Apparently browsers don't like that file type, so I zipped it. Chris ------------------ What the hell is a Jagdcarcajou? CM Recon [This message has been edited by Jagdcarcajou (edited 02-11-2001).] [This message has been edited by Jagdcarcajou (edited 02-11-2001).] Link to comment Share on other sites More sharing options...
Vanir Posted February 11, 2001 Share Posted February 11, 2001 Give it up Jadg. He's not listening. ------------------ You've never heard music until you've heard the bleating of a gut-shot cesspooler. -Mark IV Link to comment Share on other sites More sharing options...
Tris Posted February 11, 2001 Author Share Posted February 11, 2001 You are mistaken, I do listen, but I tend not to accept everything I happen to hear. For instance . . . I just had a similar instance of infantry not firing when given the move order. It was up and over a hill and as they travelled along they duly spotted an enemy unit (artillery spotter) and there was even a nice little red line to show that this squad was targeting . . . but it never fired, just kept moving toward its assigned objective. Only after it reached its objective did it bother to fire. So, as I said before there's a problem with the various movement commands related to targeting while on the move. There ought to be a way to give tareting orders and have those obeyed while units are, at the same time, on the move. At best it's a hit-and-miss process at present. Link to comment Share on other sites More sharing options...
Jagdcarcajou Posted February 11, 2001 Share Posted February 11, 2001 Originally posted by Tris: So, as I said before there's a problem with the various movement commands related to targeting while on the move. There ought to be a way to give tareting orders and have those obeyed while units are, at the same time, on the move. At best it's a hit-and-miss process at present. Hey again, This is another misunderstanding. If you set up a test scenario with various types of move situations you can clarify this stuff. An artillery spotter presents no immediate threat to the infantry (direct fire) so they move into position then begin firing (more accurately since they aren't moving). If they had spotted an enemy squad they would have fired upon them, assuming they weren't already heading for cover. If you had designated the target, as opposed to the AI finding it during the action phase, they would have fired the whole time they advanced. While these may not be the results you are looking for, they are at least predictable. Chris ------------------ What the hell is a Jagdcarcajou? CM Recon Link to comment Share on other sites More sharing options...
Guest Big Time Software Posted February 12, 2001 Share Posted February 12, 2001 Tris, As others have stated, SNEAK is designed to NOT engage ANY enemy unit UNLESS fired upon. Otherwise your units would hardly ever do any actual sneaking because more often than not there is something available to shoot at. Simple as that. You wrote: I just had a similar instance of infantry not firing when given the move order. It was up and over a hill and as they travelled along they duly spotted an enemy unit (artillery spotter) and there was even a nice little red line to show that this squad was targeting . . . but it never fired, just kept moving toward its assigned objective. Only after it reached its objective did it bother to fire. Units try to move instead of fire. That is a standard behavior. If the unit isn't able to effectively engage a target while on the move it will wait until it has paused for some other reason. I suspect that the infantry unit you used was not very capable of firing on the move (say... German Pattern 44 Rifle). Or there could have been any number of situations present that caused the behavior you described, but short of seeing the move itself it is hard to guess what it could be. Bottom Line... A battle is a chaotic environment. If you expect precision and absolute predictibility, and totally consistant results, you are playing the wrong game. A ridged game system, as opposed to the dynamic CM environment, would not only be less realistic but far more frustrating to play. Again, there might not be a ready solution given AI state of the art. However, I did offer, through an email conversation with Matt some weeks ago, to connect BTS with a friend of mine who develops recreational software and has studied fuzzy-logic theory for some 25 years. This person may have an idea or two which could benefit this project. So far I've received no word back on this overture. I suppose my offer fell on deaf ears. So be it. It did not fall on deaf ears. We have enough experience with programming, game development, and AI to know that what you offer is not practical. We have tried to work with professional AI guys before and it goes nowhere. Unless your friend is programming AI for games, similar to Combat Mission, and has full access to the entire game code base, there is nothing to discuss. Even if he has such experience, the effort necessary to even make a tiny improvement to the AI's behavior would be totally out of proportion to any possible benefit. I'm not saying that your friend couldn't offer any sort of help that might improve CM, but it is not likely without a full blown partnership. And even then it is more than likely to not produce noticably better AI. Simply put, we thank people for wanting to help, but we have been doing this long enough to know what can work and what can't. Steve Link to comment Share on other sites More sharing options...
Tris Posted February 12, 2001 Author Share Posted February 12, 2001 Fuzzy logic theory's not my suit so I'll let that pass in general. I doubt, however, that anyone 30 years old or so can have the same appreciation for this complicated field than a professional programmer who's dealt with it specifically for the past quarter of a century or more--and with pretty much the same application, at that. This isn't to offer up an argument to your position, just to point out the incongruity of the logcal direction your statement wished to go. Neither is it to suggest that the world's foremost authority in the field could come up with something better than what Charles has already, though the latter person might have beneficial insight to impart, at the least ought to serve as a useful foil for discussion. The infantry in my example was a standard GI squad. Its morale was okay, no action in its immediate vicinity--the nearest shots being fired were a mile south of it. As I said, when it reached its assigned objective it stopped, all the while maintaining its LOS/targeting red line to the German FO, then let loose with a volley that ended it all. Another funny occurance in v1.12 which I just got through witnessing a couple of minutes ago: I set my Greyhound to HIDE in scattered trees with orders to AMBUSH a marker 122m distant in clear terrain. This was to catch unawares what I supposed to be a "convoy" of German HT's headed laterally left to right across the M8's front. Sure enough, three more HT's soon appeared and the one the Greyhound had originally sighted (with LOS to it the whole while) rolled right over the ambush marker . . . only the Greyhound never fired (verteran crew, by the way, still no firing in this neighborhood, only a mile to the south). I am not trying to pick on you, Steve, or your game system. It's a splendid thing to play around with. Lord knows I invest appreciable spare time doing just that. Maybe it's coincidance I'm running across so many seeming anomolies with this latest version. Link to comment Share on other sites More sharing options...
Guest Big Time Software Posted February 12, 2001 Share Posted February 12, 2001 Tris: I doubt, however, that anyone 30 years old or so can have the same appreciation for this complicated field than a professional programmer who's dealt with it specifically for the past quarter of a century or more--and with pretty much the same application, at that. In theory, you are of course correct. But theory means nothing in this case. Practical issues, which have nothing to do with the skill of your friend vs. Charles, means that there is little that your friend can do to improve Combat Mission. Unless you are a programmer, or lead development teams and tried to do cross discipline work, this is not an easy concept to explain. I'll try Your friend could be the most brilliant AI programmer in the entire world, but unless we did nothing but have him work with Charles, day in and day out, on Combat Mission for at least 6 months, there is little chance that he could do anything to help improve its overall performance. That is the practical reality of the situation. Charles could EASILY make the AI much better than it is right now. But since about 40% of the nearly 3 year development cycle was devoted to AI, we would have to hold off releasing CM2 for at least a year to see significant improvement overall. And maybe it would even run on the current highend hardware. AI programming takes a ton of time to do. The more complex the environment, the more AI programming needs to be done. At some point you hit the wall of diminishing returns. Where every day spent doesn't significantly increase the quality of the AI unless large numbers of days are invested. We are at that point. Doing one day more of AI programming to improve a single behavior will likely produce no consistantly visible results. Unfortunately, AI does NOT sell more games. Even if CM's AI was at the level of the average multi-unit strategy game (which we feel is poor to terrible) I doubt our sales would be any lower. And if the AI was significantly better, I doubt they would be any higher. Customer happiness AFTER the purchase would be worse/better depending on the skill of the AI, but it would not affect sales. This has been true since games (not just wargames) have first graced computers. So we can only do so much and remain in business. I think if you took a pole of CM gamers they would agree that CM's AI is top dog by far. Therefore, we have already "over invested" our precious development time into CM's AI. Obviously, we made it what it is because WE would not be happy with a crappy AI. And since we have the freedom to release only what we are happy with, CM's AI kicks butt. It isn't as good as we could, in theory, make it... but there is something called "reality" and we have pushed the envelope about as much as we can for the moment. I am not trying to pick on you, Steve, or your game system. It's a splendid thing to play around with. Lord knows I invest appreciable spare time doing just that. Thanks Just keep in mind that while Charles might not know as much as your friend in terms of Fuzzy AI programming, Charles knows more about wargame programming than he does. And he certainly knows more about CM and the Big Picture of our development schedule, market, and business model. And it is on that basis that I say, quite confidently, that there is nothing that your friend can PRACTICALLY offer Combat Mission. Maybe it's coincidance I'm running across so many seeming anomolies with this latest version. It must be. Except for some tweaking of the smoke usage behavior, Charles made NO changes to the types of behavior you have mentioned thus far. So if you are seeing it now with 1.12 it would be just as likely that you would see it in earlier versions. Steve Link to comment Share on other sites More sharing options...
Tris Posted February 12, 2001 Author Share Posted February 12, 2001 Re game play I assume that's the case (vis-a-vis changes made since v1.05). So be it. Re fuzzy logic: I understand your thrust, really I do. There simply are no easy answers as it's all pure theory, per se, until it's applied in game and "hit and miss" comprises the marching orders for the day. The programmer I reference doesn't write wargame software but sports. Still, its the work of a behavioral science of sorts and the principles do not vary much from that base. Anyway, as always I look forward to whatever improvements to the system that can be delivered to us out here. We do progress. Link to comment Share on other sites More sharing options...
Tris Posted February 12, 2001 Author Share Posted February 12, 2001 Jagdcarcajou, thank you for your effort. I DL'ed your test scenario and as you say the outcome (based on three tries here) is predictable. I then set up something similar but with an Sd Kfz 7, SPW 250/1 HT and truck as the transports. I further loaded these vehicles with two squads and a platoon HQ. I find it interesting that the advancing (ordered to sneak) squads will fire on both the truck and Sd Kfz 7. What happens is that they identify the riding squads and apparently count these as threats even though they are still on their transports and thus cannot yet fire. Again, I appreciate you taking the time. Link to comment Share on other sites More sharing options...
Guest Big Time Software Posted February 12, 2001 Share Posted February 12, 2001 Tris: The programmer I reference doesn't write wargame software but sports. Still, its the work of a behavioral science of sorts and the principles do not vary much from that base. Once again, in theory. In reality, they are as different as Warcraft II is to Combat Mission. Totally different, inside and out. And it is what is "inside" (i.e. the design and coding) that is the most important thing to keep in mind. Programming trashes theories and good intentions faster than anything else in the known universe For example, Close Combat is about the closest game to Combat Mission out there. In theory. In reality they are so far apart from each other that there can be no meaningful direct comparisions. Assuming that CC uses sound Fuzzy Logic principles (I don't know if it does or not), if Charles talked with the CC AI programmer I bet he could offer him very little in terms of improving CC. And vice versa. And these two programmers share FAR more in common, both in terms of experience and game type, than anything else out there. Again, unless your friend worked with Charles FULL TIME for many, many months there would be no noticible improvements worth such an investment of time. Anything short of this is a waste of both your friend's and Charles' time. And it has nothing to do with the intelligence, skill sets, or creativity of either person. Steve [This message has been edited by Big Time Software (edited 02-11-2001).] Link to comment Share on other sites More sharing options...
Guest Big Time Software Posted February 12, 2001 Share Posted February 12, 2001 Tris: I find it interesting that the advancing (ordered to sneak) squads will fire on both the truck and Sd Kfz 7. What happens is that they identify the riding squads and apparently count these as threats even though they are still on their transports and thus cannot yet fire. Correct, although I would be surprised if this happened at anything but close distances. The logic though is quite simple. The infantry may not be a threat at this particular second, but the next it could mean the death of the squad. If the probability of the enemy spotting the Sneaking unit is high, then so is the chance that the unit will start opening up. Get some range between the two, or significant cover, and the likelihood that the Sneaking infantry would continue to hold fire should be much higher. Steve Link to comment Share on other sites More sharing options...
Jagdcarcajou Posted February 12, 2001 Share Posted February 12, 2001 Originally posted by Tris: Again, I appreciate you taking the time. Hey, No problem Tris. Glad I could help. Chris ------------------ What the hell is a Jagdcarcajou? CM Recon Link to comment Share on other sites More sharing options...
Tris Posted February 12, 2001 Author Share Posted February 12, 2001 Not sure what might constitute "close" but I've just gone back in there and increased the distance from 50m to 150m. First, the US platoon HQ opened up on the truck from 176m (HQ was positioned somewhat to the rear of its command elements). First US squad then fired on the Sd Kfz 7 from 146m, third US squad followdd suit on the truck over on the right flank from about 155m as this German transport backed up. Again of interest, the SPW 250/1 meanwhile targeted both first and second US squads in succession. While second US squad returned this fire first US squad ignored this threat and instead kept up its fire on the thus far non-threatening Sd Kfz 7, per its original targeting orders. Finally, third US squad let loose on the truck with a couple of bursts (ammo 38). At this juncture the German infantry squad aboard the truck dismounted and ran off to the left, meanwhile third US squad continued to fire on the truck, not this dismounted German squad. I know more about squad behavior with movement orders and targeting in mind now, but I'm not sure I see a whole lot of consistency to this behavior--neither is it particularly intuitive on the surface. As you say, Steve, and rightly so, one wouldn't be satisfied in the long run if this behavior was scripted instead of left to fuzzy logic (though I suspect there is a script of sorts in there, for priorities if nothing else, with fuzzy logic then taking over at some level), and to a point unpredictable behavior does lend an air of realism to the proceedings. It also lends it share of frustration. Link to comment Share on other sites More sharing options...
M Hofbauer Posted February 12, 2001 Share Posted February 12, 2001 Michael Dorosh wrote: Once one of your units spots something, it is considered spotted by all units on the board. Your men could "see" it. That is incorrect. Spotting is calculated for each unit on it's own, there is no CC-like beehive telepathic mentality in CM. Link to comment Share on other sites More sharing options...
Recommended Posts