Battlefront.com Posted February 19, 2014 Share Posted February 19, 2014 Steve, if you were to stop producing new theatres after RT (and maybe Black Sea; 'pends how far along... ) and get the engine coders working only on AI improvements for "one product cycle", then release the three upgrades at $30 apiece (equivalent to one new theatre at $55 and three $10 upgrades), I think you'd make a lot of people happy. And the "skin" guys can make packs of vehicles that don't need any/much engine input to keep them busy and make more people happy and get your "pack" income stream started (and get a head start on the next theatre's skins maybe)... Or wouldn't "one product cycle" make enough difference in the AI for you to feel comfortable charging "that much" for it? Two problems with that: 1. Making people happier doesn't necessarily translate proportionally to making more sales. Conversely, people who complain endlessly about various issue in a game doesn't necessarily translate into proportionally lower sales. On the whole people don't put their money where their mouth is either way. 2. Opportunity cost compounds the expense side of things. Which is to say if we stopped making new games for a while we would be sacrificing income for that period AND carrying the same expenses AND likely not getting anything extra for the effort. I learned a long time ago that there are extremely sound reasons why the average game usually has terrible AI and the best have only decent AI (or not!). Oh, and couple that with the fact that CM's level of play is the worst of all environments for doing AI. There is no harder level of combat to write AI for than this level. Lower is focused on the individual and not coordination, higher is focused on coordination and not the individual. CM's level has to be equally concerned about both. Sucks to be us! Steve 0 Quote Link to comment Share on other sites More sharing options...
BLSTK Posted February 19, 2014 Share Posted February 19, 2014 I'm like the Mini-Me to Doctor Evil. Well then, the good Doctor needs a cat. Molester? 0 Quote Link to comment Share on other sites More sharing options...
c3k Posted February 19, 2014 Share Posted February 19, 2014 Simple solution is for BFC to create a server farm which comstantly crunches every tactical permutation. Then players would have to link to it online for single player. The TacAI would no longer be limited to consumer grade processors. We could train it by playing tic tac toe. 0 Quote Link to comment Share on other sites More sharing options...
BLSTK Posted February 19, 2014 Share Posted February 19, 2014 [quote=Whiterider;1503289It´s impossible to get tits and ass (in spanish: Imposible tener teta y culo ). 0 Quote Link to comment Share on other sites More sharing options...
Whiterider Posted February 19, 2014 Share Posted February 19, 2014 Rider, you are my hero. Now I can tell my wife I'm "brushing up" on my Spanish. For this I thank you...errr...muchos gracias! You´re welcome! (¡De nada!). Another classic proverb in spanish is "two tits are more powerful than two chariots" (pueden más dos tetas que dos carretas") Sorry for the offtopic. 0 Quote Link to comment Share on other sites More sharing options...
BLSTK Posted February 20, 2014 Share Posted February 20, 2014 I just won't tell my wife what I'm learning in Spanish. 0 Quote Link to comment Share on other sites More sharing options...
Whiterider Posted February 20, 2014 Share Posted February 20, 2014 I just won't tell my wife what I'm learning in Spanish. Wise decision because "Battles against women are the only ones you can win runing away". (Napoleon Bonaparte). returning to the topic: AI improvements are a good business in a loooong-term (everybody loves good graphics/AI) but a disaster in a short-term. Obviously Battlefront prefer to have a bird on hand than one hundred in the sky (another one: "más vale pájaro en mano que ciento volando"). 0 Quote Link to comment Share on other sites More sharing options...
BLSTK Posted February 20, 2014 Share Posted February 20, 2014 "Not tonight, Waterloo." (Josephine Bonaparte) 0 Quote Link to comment Share on other sites More sharing options...
c2yeung Posted February 20, 2014 Share Posted February 20, 2014 I have always played against AI and found that AI defense is pretty decent, especially the placement of its forces. They are well positioned and inter-connected to provide mutual fire support that costs me lot of time and manpower/equipment to chip away bit by bit. AI offense, on the other hand, leaves much room for improvement. It is either very recklessly aggressive or very docile (at end of scenarios, I sometimes saw a sizable force sitting idle way behind the line). What I would like to see is a more reactive and dynamic AI that will think on the fly depending on battle situation. I am not asking for AI level equivalent to the IBM "Blue" that plays chess with Kasparov as an equal. A most robust AI will definitely make the game more enjoyable. 0 Quote Link to comment Share on other sites More sharing options...
Battlefront.com Posted February 20, 2014 Share Posted February 20, 2014 No doubt that a better AI makes for a more challenging game. The problem is the disproportional amount of work to get that part of the game improved vs. what can be improved in other parts of the game using the same amount of effort. Steve 0 Quote Link to comment Share on other sites More sharing options...
Childress Posted February 20, 2014 Share Posted February 20, 2014 How many computer war games provide a nimble and convincing opponent on offense anyway? They're rarer than hens' teeth. 0 Quote Link to comment Share on other sites More sharing options...
Whiterider Posted February 20, 2014 Share Posted February 20, 2014 How many computer war games provide a nimble and convincing opponent on offense anyway? They're rarer than hens' teeth. Yep, really really rare. The following wargames have a challenging AI, however their weak point are graphics. Nothing is perfect. :cool: http://matrixgames.com/products/471/details/Flashpoint.Campaigns:.Red.Storm http://matrixgames.com/products/377/details/Command.Ops:.Battles.from.the.Bulge.. 0 Quote Link to comment Share on other sites More sharing options...
c3k Posted February 21, 2014 Share Posted February 21, 2014 Humans learn; computers don't. Every AI will be defeated by a human. Maybe not the first time. Or even the second or third. But eventually the human gains supremacy. Triggers are a huge step forward for the CM AI. A turing level ai is impossible for consumer computers in the foreseeable future. Skilled designers and triggers will be key. 0 Quote Link to comment Share on other sites More sharing options...
Collingwood Posted February 21, 2014 Share Posted February 21, 2014 I haven't seen a wargame AI that learns, but there are other AIs that learn. Perhaps not in the way we think of learning, but they can adjust the choices they make based on previous results. I know we're discussing wargames AI, but for Chess, "eventually the human gains supremacy" is now a long dead hope. An example of learning in computing is back propagation neural networks. Effectively you show the network a training set of data, and for each example you provide the expected answer/result. The network trains itself to a level of confidence you specify, and then you can give it further problems and it will answer based on its knowledge. Completely agree about skilled designers and triggers. I can't wait! 0 Quote Link to comment Share on other sites More sharing options...
Battlefront.com Posted February 21, 2014 Share Posted February 21, 2014 In 1992 I designed my first game AI using the then novel "Fuzzy Logic" and "Fuzzy Neural Logic" theories. On paper it's kick arse However, trying to make that work in a game is a massive undertaking. That's what sucks the most about AI. Smart game designers and programmers know exactly how they would go about things if there was a financial incentive to do so, but there is actually a financial DISincentive. We don't like it any more than you do. Yup, the first defense a game developer has when it is accused of not having a good AI is "point me to a game that does". Very, very, very few examples to point to and often they have nothing to do with the game being criticized. In our case, pointing to an operational AI is not applicable. Might as well ask for a 5 star meal fit for kings and then point to the best damned high end pizza restaurant in town and say "just do what they do". Operational games deal with vastly fewer and more simplistic "ingredients" and far lower expectations of brilliance than tactical games do. Steve 0 Quote Link to comment Share on other sites More sharing options...
Collingwood Posted February 21, 2014 Share Posted February 21, 2014 In 1992 I designed my first game AI using the then novel "Fuzzy Logic" and "Fuzzy Neural Logic" theories. On paper it's kick arse However, trying to make that work in a game is a massive undertaking. Wow, I've only applied neural nets to classification problems (e.g. handwriting recognition). I can't think how you could even begin to use them for game AI. That would be really interesting. Yeah neural nets got me very excited, then reality hit home. Same with fuzzy logic and genetic algorithms. 0 Quote Link to comment Share on other sites More sharing options...
Michael Emrys Posted February 21, 2014 Share Posted February 21, 2014 Smart game designers and programmers know exactly how they would go about things if there was a financial incentive to do so, but there is actually a financial DISincentive. We don't like it any more than you do. So let me see if I've got this right. What you are saying is that if someone were to endow BFC with, say, $10,000,000 you would hire a team of talented programmers and have them work on nothing but developing a super game AI? Michael 0 Quote Link to comment Share on other sites More sharing options...
Battlefront.com Posted February 21, 2014 Share Posted February 21, 2014 Yes. I'll even tell you where you can send the check. But we'll make sure it clears first before getting started Steve 0 Quote Link to comment Share on other sites More sharing options...
Whiterider Posted February 21, 2014 Share Posted February 21, 2014 Games with Dynamic AI (non-scripted) tend to be the most challenging because the AI has an umpredictable behaivour that changes depending on the circumstances. Still nowdays it´s something extremely rare but I guess in a few years will become fairly common. I´m very sure computer programmers would love to include it in their games, but it´s a matter of money and time. Patience is the key. An interesting interview about AI in games: https://www.idlethumbs.net/3ma/episodes/thinking-machines 0 Quote Link to comment Share on other sites More sharing options...
Whiterider Posted February 21, 2014 Share Posted February 21, 2014 One of the latest improvements in AI is the use of drivatars in "Forza Motorsport 5"; all computer-controlled opponents are modelled on other human beings using the cloud. More info 0 Quote Link to comment Share on other sites More sharing options...
Sailor Malan2 Posted February 21, 2014 Share Posted February 21, 2014 I am not an expert (by any measure), but I guess the last, best hope for mank... AI coding for this type of game will be when other demands in computing cause someone to develop a generic AI engine that can be configured to produce a game AI (as a byproduct). It seems far fetched maybe but for controller modelling for engineering applications such things exist (MatrixX or Simulink for example). A sophisticated GUI lets you string together standard functions, embed your own custom routines, and simulate the result against a model of the plant you are controlling. We need the consumer use of robots, or 'intelligent houses' to take off seriously for this to happe I think, but one day. Just compare how games have changed in depth and graphics in the last 15 years... 0 Quote Link to comment Share on other sites More sharing options...
Fizou Posted February 21, 2014 Share Posted February 21, 2014 I think the AI is good in defence already (atleast OK and as we all know, AI is hard to get right). The new triggers and the continual work to improve these will surely be a very nice addition and make the game even better. Some of our talented scenario and campaign designers are already the target of a lot of complaints that their work is to hard. This wont make the AI any easier. I like it!! 0 Quote Link to comment Share on other sites More sharing options...
Michael Emrys Posted February 21, 2014 Share Posted February 21, 2014 ... AI coding for this type of game will be when other demands in computing cause someone to develop a generic AI engine that can be configured to produce a game AI (as a byproduct). Interestingly (to me) this is the same idea I've had bouncing around in my head for several years now. I don't think AI programming has found its Einstein yet, someone who has a brilliant insight that cuts through the complications that make it difficult now. I haven't a clue what that might be or I'd patent it and move to a tropical paradise. Michael 0 Quote Link to comment Share on other sites More sharing options...
Cymru Posted February 21, 2014 Share Posted February 21, 2014 Yes. I'll even tell you where you can send the check. But we'll make sure it clears first before getting started Steve Damn, wish i had known. I just gave my last $10,000,000 to Farmville. 0 Quote Link to comment Share on other sites More sharing options...
Skwabie Posted February 21, 2014 Share Posted February 21, 2014 I entirely understand the reasons that AI development is a little low on the list, BFC is a business and we've seen over-ambitious business models die out over time during the past 2 decades or so. However, I somehow think the AI faces an exceptionally tough challenge in CM, for there're rarely other games where an AI controlled entity faces so many environmental factors like terrain types, paths, LOS/LOF considerations etc etc. IMO it wouldn't be a bad idea to pump it up a notch to give the game a bit more re-playability... 0 Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.