Fionn Posted February 23, 1999 Share Posted February 23, 1999 Rock, I think you're talking about Panzer Elite there by Wingssimulations. BTS seem to know Teut . A little friendly rivalry perhaps . I won't abuse the board by going into much detail about PE (and since it is a tank sim it won't really be competing with CM anyways) but as a tester of PE yep it can surprise you and a lot of work is going into the AI thank god . I'm glad to see some companies still care about AI. I've seen FAR too many games fail on AI yet still garner plaudits. AND all too few magazines have a dedicated AI evaluation in their reviews which doesn't help matters. Securing from AI rant mode now Link to comment Share on other sites More sharing options...
HAL Posted February 24, 1999 Share Posted February 24, 1999 Hey Fionn do happen to know the URL for Wings Simulation? Psygnosis doesn't have any references to Panzer Elite on the Web Site anymore. Until your post I was afraid that they had canceled the project. Sorry to go off topic. HAL [This message has been edited by HAL (edited 02-23-99).] Link to comment Share on other sites More sharing options...
Marko Posted February 24, 1999 Share Posted February 24, 1999 Open the pod bay doors, HAL: http://www.wingssimulations.com Link to comment Share on other sites More sharing options...
Guest Bobb Posted February 24, 1999 Share Posted February 24, 1999 I am with Marko. In my opinion ANY ATEMPT TO FORCE A SCENARIO DESIGNER INTO "GOOD" DESIGN PRACTICES, is ill concieved. A scenario designer should be able to author his creations with as many tools as can be practicably layed at his feet. It the scenairo designer's responsibility to produce a good design. If it be bad, so be it. The more we work to limit the potiential to do bad designs, so we limit the potiential to do good ones. Assume a fantastic AI. A designer might want to provide some lameness. Let one of the players tasks be to discover a weakness, or have such a weakness or stupidity show up somewhere, perhaps at random, so that if stumbled upon or otherwise discovered, there lies opportunity. Every opportunity to spice up scenario design amplifies the richness of play. Random elements expotentiate that into replay. The final line on AI design is that it is a contrivance, so that even the best concievable example is going to have "personality" characteristics that become apparant after sufficient exposure. If the "language" of scenario design is rich enough, then the designers thereof will have at their command elements, which can be used to counter or otherwise compensate and thereby vastly enrich the possibilites of design and therefore play. Hidden Victory Locations or Waypoints or Holding Zones are ways to make AI units converge, move or stay in place. If these locations be assigned a range of values with which the AI interacts; if they be avalilable as random elements as well as with time limits, activation and expiration triggers, then a scenario author can design in such rich variety as to even make playing his own work a challange and filled with surprises. Can one not evision a hierarchy of AI decisions about place values with fuzzy acouterments included, where the Visible VLs basically rule according to their value, but in concert with the hidden ones with random elements and time limits all entertwined? Must be a programming nightmare as well as a tough set of sticks for juggling as a scenario designer. But I have faith! Waypoints persay make more sense if one can assign them to a specific unit. Such a unit could be sent on a scouting mission or be assigned to be the unit selected to an occupation mission. Otherwise VLs can act as waypoint system for general use by the AI. A designer should be able to assign a unit to a holding location with leaving "instructions". Hold at all costs, Hold for X turns, Spot and Withdraw. Teams should be available for outposting or listening post locations. I recall a mention of a player being able to hear the enemy tanks etc. Does the AI have the ability to "get nervous" in hearing range of the player's tanks? Do the Germans hear the talkitive, noisy greenhorns fresh from the reppeldeppel better than veterans? I'll repeat for emphasis, If Outpost Teams are made available, defence is a much more realistic exercise. Where whole squads are the only option, then a lot more of the player's strength is exposed to immediate loss than with outposts. Units can be positioned for potiential flank attacks or for safe withdrawal if a frontal attack is discovered by outposts. This same type of team would also be available for point in movements. I know that this game is quite apparantly in far more capable hands than mine so I speak these lines with a temorous heart; what I say is more in hope of offering encouragement to these intrepid three than in expectation that I can contribute - the nature of that encouragement being that, My God! what a game we have growing here! It makes me believe it can accomplish all things for all wargamers! To what other wars shall we next see it applied? It makes me so excited that I can't contain what I otherwise would be smart enough to keep to myself. I am exceedingly impressed with depth of background that this design crew possesses and its appreciation of detail and nuance; and, perhaps as importantly, a sense of the limitations that individual perspective and available information impose. Opposed to my opening line I finish by saying PROGRAMMING AND BASIC SYSTEM DESIGN CONSIDERATIONS OUTWEIGH ELEMENTS, WHICH THOUGH OTHERWISE DESIRABLE MAY BE EXCESSIVE PONDEROUS. (probably like this posting) Therefore, if game design is a coocoo clock, I shall not look for eggs therein. However, this coocoo looks so real that I might just sneak a peak. [This message has been edited by Bobb (edited 02-23-99).] Link to comment Share on other sites More sharing options...
Guest Big Time Software Posted February 24, 1999 Share Posted February 24, 1999 I'll try to answer Bob's post without going point for point... : A scenario designer should be able to author his creations with as many tools as can be practicably layed at his feet. We agree to a point. Where we draw the line is where these tools start to interfere with the game itself. CM's AI will be fluid and dynamic, even if it isn't as smart as a human. It will take advantage of situations that NO scenario designer could ever dream of, no matter how good, even if its decisions are the best. So the more "rules" the designer has at his disposal, the more it is going to bind the AI into static thinking. This is bad. Normally, AIs aren't dynamic so these "rules" are the ONLY thing that drive the AI's decisions. As stated in my message above, this is why AIs suck. We strive to have an AI that is much better than "sucky", so AI crutches are not on our agenda. : If the "language" of scenario design is rich enough, then the designers thereof will have at their command elements, which can be used to counter or otherwise compensate and thereby vastly enrich the possibilites of design and therefore play. : A designer should be able to assign a unit to a holding location with leaving "instructions". Hold at all costs, Hold for X turns, Spot and Withdraw. We disagree with this. Basically you are trying to preprogram the AI player, instead of it doing this for itself on the fly, based on events as they unfold. What you are talking about is scripted AI and that is something that will never work with wargames. The AI has to be able to think for itself or it will suck. Plain and simple. Scripting is just a fragile, easily broken crutch for poor AI. In the end, it doesn't really make it any better, and can even make it worse. So far as we are concerned, the MORE AI tools you give the scenario designer, the weaker it will make the AI. Well, at least the kind of AI we are making. And since CM will have randomly generated maps, we HAVE to make an AI that works without crutches, because there will be no designer available to hand tweak maps. If we fail to do this, all the invisible markers in the world won't make the AI better. : It makes me believe it can accomplish all things for all wargamers! Well, now I have to figure out some UBB code... where is that... ah! Here it is We don't think CM will accomplish EVERYTHING that wargamers want. After all, we have to leave SOME stuff out so we can actually finish the darned thing However, Combat Mission will, for SURE, do more than any game that has preceded it. In fact, it already largely does. Steve [This message has been edited by Webmaster (edited 02-24-99).] Link to comment Share on other sites More sharing options...
Guest Bobb Posted February 24, 1999 Share Posted February 24, 1999 Blazes Steve, your're fast. You can make a cogent answer faster than I can make a muddled statement. I suppose we can agree to disagree a little. I may be missing a little in the computer generating capability so that I don't quite understand what that does to scenario design. My viewpoint is rather confined to scenario design for one game only. So that experence probably colors it with a limited palate. I just think it would be nice to be able imprint some features into the behaviour of the AI that would cause some disjunctions that would otherwise not occurr. If we leave so much up to the AI, it may take some of the fun out of authoring a scenario. After all why should you three hog all of the creativity? I have complained bitterly of some scenarios that I have played wherein it seemed that I, the "commander" on the battle field, was being micromanaged by some jerk in higher command who had no notion of what it was like to be there working under his stupid orders. This has been the case in historical battle often enough. As a player of CM I may impose such "orders" on my own units through command incompetance. I think it would be nice to be able to arrange for such things to occurr on the part of the AI. There, does that not express the ultimate faith in your design? To ask for the ability to dumb down, at least here or there in a scenario with a measure of considered moderation, the games AI must be some exercise in highly placed confidance. Link to comment Share on other sites More sharing options...
Guest Lokesa Posted February 24, 1999 Share Posted February 24, 1999 with deployment zones you should be able to guide the AI strategy to a certain extant. For example you could make a very small zone requiring a squad be placed there where you would want you outpost to be. Whether the AI thinks it's a good idea to remain there is another thing Link to comment Share on other sites More sharing options...
Moon Posted February 25, 1999 Share Posted February 25, 1999 With reference to an earlier post by Fionn about the AI in Panzer Elite - reading through their FAQs it becomes clear that the AI in PE is linked to the pre-generated maps. This, obviously, is a little trick that makes it much easier to "simulate" a strong AI. I'm not saying it's bad, but it wouldn't work with random maps... BTW: I know also another game with an awesome AI (101st Airborne in Normandy) - which Fionn knows pretty good - that is using exactly the same "trick". One of the reasons there is no scenario editor for 101... It seems to me that linking the AI to a pre-defined map is the easiest way out of the AI problem. From the point of view of a player I must say that I am very satisfied with 101's AI (obviously I don't know PE yet) - it feels very much like playing against a human opponent. However, since I prefer playing against human opponents anyway, I'd trade in the strategic AI against a scenario and map editor anytime... Link to comment Share on other sites More sharing options...
Fionn Posted February 25, 1999 Share Posted February 25, 1999 With reference to an earlier post by Fionn about the AI in Panzer Elite - reading through their FAQs it becomes clear that the AI in PE is linked to the pre-generated maps. This, obviously, is a little trick that makes it much easier to "simulate" a strong AI. I'm not saying it's bad, but it wouldn't work with random maps... Agreed,another reason why a scenario editor will be delayed IMO BUT this does give the scenario designer GREAT ability to tailor the actions of supporting and enemy units in any scenario. e.g. In the first scenario we tested we felt that the Sherman platoons facing the Germans would move up to a hill on the German flank and HOLD POSITION and fire from hull-down while their infantry moved up on the German's other flank and several units made a demonstration to the German front. The Scenario designer worked on it and now the AI exploits flanks, hull-down, ultra-concerted attacks etc. The nice thing is that all of the AI actions in scenarios are commented on and argued about by sever history buffs and quite a few current and ex-tankers. The only thing that annoys me about PE is that I still get my clock cleaned in anything under a Tiger I (or possibly Panther). It's a good complaint to have . BTW: I know also another game with an awesome AI (101st Airborne in Normandy) - which Fionn knows pretty good - that is using exactly the same "trick". One of the reasons there is no scenario editor for 101... Agreed . Hardcoding the AI to a map makes it much more "intelligent" for that map BUT the ULTIMATE AI IMO is one which will utilise tactical concepts ad precepts for making its decisions. This will allow the player to use the AIs own doctrine against it realistically and overall will make a much better game. I know this is probably a holy grail BUT if a company decides to programme an AI like BTS has done and makes this public commitment and has the track record BTS has with CM(as far as openness and honesty go) and other games they have made I'm willing to bet we'll be pretty happy with the results. I don't think an amazing AI is all that difficult to create, I just think most company's realise that for all that extra cost and delay they won't gain a marked number of increased sales. Remember most publishers do NOT aim to bring you the best game. They aim to bring you the most profitable game. Lose any other illusions. It seems to me that linking the AI to a pre-defined map is the easiest way out of the AI problem. From the point of view of a player I must say that I am very satisfied with 101's AI (obviously I don't know PE yet) - it feels very much like playing against a human opponent. You'll like PE. It is tough, yet realistic and a fair few of the testers are tankers or military. One of the BEST things is the openess to tampering with files, images etc. One crazy tester (name witheld for psychiatric reasons ) has made a house with "Kilroy was here" scrawled all over it. It's good to see he's testing the most important aspects of the game isn't it ? NOT ! I just think we're all lucky to have a few good games coming onto the market soon. And in three separate spheres so we can justify buying all three of them (important consideration to those of you "attached" , you know what I mean ). Combat Mission Panzer Elite Road to Moscow Feel the drool .. [This message has been edited by Fionn (edited 02-25-99).] Link to comment Share on other sites More sharing options...
Recommended Posts