The Chick Parabola is the curve on which women look when you get drunk. See they are ugly.... then you drink.. they get hotter... then you drink more.. and things go down hill... you start thinking they are a goddess and.. boom. The curve falls downward again.
Soren Johnson: The Chick Parabola
- Thread starter alexman
- Start date
troytheface
Deity
- Joined
- Mar 26, 2002
- Messages
- 3,262
lol
it takes the guy like a page to write something as simple as "after you play a game for awhile you learn it and move on"
Tabula Rasa is more significant, relevant, and accurate
the evidence is clear- They like lots of line
it takes the guy like a page to write something as simple as "after you play a game for awhile you learn it and move on"
Tabula Rasa is more significant, relevant, and accurate
the evidence is clear- They like lots of line
Dale
Mohawk Games Developer
- Joined
- Mar 14, 2002
- Messages
- 7,601
Most popular != good. A game that simply relies on negative phsycology to force your players to return is not a good game.
And yet Soren is making a FaceBook game (announced by EA two days ago). Nice condemnation eh?
As far as I know he is responsible for Frontierville and not Farmville.
He is also responsible for CivII and SMAC, so he already reached the pinnacle of glory in gamedesign.
After so much good (with Civ and SMAC) he had balance with some evil (Frontierville) to keep yin and yang into balance
Brian took over Farmville early in production. He didn't have the original idea, but Farmville today is because of him.
Your point is interesting. Does that mean after so much good Sid has balance with Civ3 and Civ5?
TheMeInTeam
If A implies B...
- Joined
- Jan 26, 2008
- Messages
- 27,989
Civ has guys like jdog5000 outside looking in...he was the key component of the BetterAI mod for BTS and is a beta tester for V. I wonder how the AI would be performing if instead of BBAI, he were working on the V AI too for a few years
.BBAI isn't perfect, but it's pretty nasty compared to BTS 3.19.
AIs are not going to be competitive with humans in games like this, generally. One workaround for that is to make them less predictable. Take for example an AI modeled after human behaviors in civ V. Say it is very simple; it simple selects between canned strategies that copy human patterns and executes them. Predictable and easy, right? Maybe not, if the AI picks these strategies on a weighted basis. You would have maybe a 33% chance of getting horse-rushed, 33% chance of the AI spamming cities and playing defense, 33% chance of it playing to set up a dogpile on someone or the player...all without readily selling open borders or tipping its hand. Ultimately, if you set the game up correctly the player would have to guess a little, especially with multiple AIs.
Such an AI isn't very dynamic when you look at it, but it would heavily outperform most civ IV and V AIs. Maybe you could weight them for "flavor" without having them be 100% predictable. Beyond that most of the AI work would then be on empire optimization.
Anyway I agree that this should be a career unto itself and emphasized more heavily - the gaming industry needs it more and more.
One true statement is: Contemporary strategic single players games should be design around AI.
In civ 5 that would mean that Map and unit informations should be core design, made as easy and efficient to access them as possible.
1UPT is ok, it would be better (easier for AI a bit of it could stack units in non combat situation, thought leaving them a bit exposed this way.)
Otherwise, if access to unit and terrain information is fast, then it is possible to write Brute force tactical combat AI. Add to it some strategic AI, like using expendable scouts to scout ahead, how to make attack force group, how to move it and we can have good AI. But again, map and units information need to be design for AI from ground up.
we still do not have code, and based form what I read, Civ V design did not follow this model.
In civ 5 that would mean that Map and unit informations should be core design, made as easy and efficient to access them as possible.
1UPT is ok, it would be better (easier for AI a bit of it could stack units in non combat situation, thought leaving them a bit exposed this way.)
Otherwise, if access to unit and terrain information is fast, then it is possible to write Brute force tactical combat AI. Add to it some strategic AI, like using expendable scouts to scout ahead, how to make attack force group, how to move it and we can have good AI. But again, map and units information need to be design for AI from ground up.
we still do not have code, and based form what I read, Civ V design did not follow this model.
That's what they tried to do in Civ5. Lot of people consider it a failure.
MadRat
Cheese Raider
In CiV the problem was they programmed it so the AI doesn't play so much as to win as to play to ensure the human player doesnt. It often does things that, while they while they appear to be moves at winning are in fact just suicidal.
I agree with SJ - the AI is like a novice chess player; looking only at the board as it is now (what some call "1 move ahead" strategy). Experienced players look at every position not just for immediate gains but principally they have an end game outcome in mind. Thus, for example, and AI is offered a good trade deal, but refuses to take it because it is beneficial to a "rival", sadly that deal was for coal and now the AI cant build factories.
On the topic of navies, and having done some piece work game programming, it is indeed something to give the devs a stroke. In RL naval invasions are logistical nightmares and the asking the AI to deal with them is hard. The conecpt of Naval strategy is very different from land strategy. It isn't about beelining to towards enemy cities or units it is about notions like coastal defence, bombardment, escort, projection of power. The sea also creates massive pathing issues - it is such a massive "space" - it also acts like an extra dimension if you will complicating avenues of approach, shortest path. In fact it is this very "spatial" geometry of sea power that makes it so easy for humans and so hard for AI.
That isnt to say that land AI development is any easier but it at least it has the continuity of geography which reduces the spatial logic a tad.
Going back to the chess analogy - experienced players dont always thing of the single best move but rather in term of abstract strategy like zones of control, freedom of movement and positional strength - these are very abstract concepts. Most novice players like an AI will think in terms "what is the best move NOW" out of all the possible combinations (ie brute force). It isnt just about long term strategy but the very spacial dynamics. How do you code this kin dof abstract geometry and have it applied rationally. It isn't easy.
That being said. I think the reason a lot of folks (including myself) are so disappointed, is that CiV4 seemed to do everything a lot better with a lot more variables in play. Now this could in part be due to the fact that the game was getting so complex that even human players were unable to completely understand the system and didnt play as efficiently; thus making the AI look better. In civ5 the mechanics are easier to grasp thus the AIs short comings are more apparent. Secondly in Civ4 the alternate paths to victory other than domination are viable. Consequently the AI could win without being a military ogre.
So knowing all this - why am I disappointed. Simply because simplifying the game exposed the AI for what it was (a dumb bot) and destroyed the kind of immersion and fun you got from the earlier Civs with their "anything can happen" outlook from al the complexities in play. Civ 5s AI shortcomings are so more visibly apparent that it feels more like playing chess with an idiot than entertainment.
I agree 100% with the earlier poster about MoM - the AI was stinky (MOO series/HOMM series/EU series) but the entertaining value and the huge variability in what can happen made it still fun. These weren't smart AIs but they were smart enough to approximate an enetertaining opponent. The biggest complaint about CiV 5 by many is that it feels to "bland".. this is a case in point - there isnt this kind of uniqueness to the games as in earlier versions. The oversimplification exposed the stupid AI for what it was and it had no enjoybale trappinsg (random events, alternate wins, build your own units (SMAC/MOO), deeper diplomacy. character, game changing wonders, AI personalities). Each game of CiV 5 is the same boring linear gring that we cant help but focus on the AIs blunders.
/wall of text crits you for 10482091 damage. Loading Please wait....
Rat
I agree with you on this. If your talking Panzer General or (especially) Operational Art of War with hundreds of units, each with dozens of attributes on maps of thousands of hexes, then yes, I can see the complexity. But on CiV we have a dramatically reduced number of units - in the games I've played (before shelving it) a big nations entire army would be maxed at 10 - 15 units. And as each unit has so few attributes (strength/move/range) surely you could calculate EVERY possible outcome for a couple of turns ahead and still have processing time left over?
And as you say, at the moment the AI is making the most ridiculous mistakes that could be easily fixed. eg. Leaving workers within range of ranged units, continually moving units from strong positions to positions where they can be attacked. "Attack ranged units with mobile units that move in from outside of range". etc etc We're not talking "Deep Blue" here, thinking dozens of moves ahead, just a simple "Don't move the queen into a position where it takes taken for nothing"
I suspect the weak AI is currently down to business decision of a rushed release, not due to the technical challenge.
He's actually saying that's only the case for some games. That there are others that once you've learnt them, they become more interesting and that's when you get really hooked.
I guess the latter are few and far between, but I know exactly what he means. Civ 4 is an example of a game that takes a while to learn, but becomes more interesting the further you learn the rules. At the end of a typical shooter, the last thing you want to do is start it again. At the end of most Civ4 games I've played, I really have wanted to go back and start again and try something else.
On the other hand, Paradox games have always been like the first type to me. A couple of weeks of being baffled by them, but enjoying them, thinking this is going to be "the one". Then once I start to figure out the mechanisms, their games become quite boring and I realise my only role is as a "slider adjusting agent". X happens, move Y slider down 2 notches.... ad nauseum. And they make them so user-unfriendly and hide key information from you while bombarding you with irrelevant info just to conceal how basic they really are. "The learning curve is the game". I still keep buying the next Paradox game just in case this is "the one", though.
Well said. The AI sometimes feels more like a speed bump than an opponent.
I have never played panzer general but from what I heard doesn't the AI just defend positions on pre-set maps?
Also did the AI have full visibility of where there opponents units were?
That wouldn't be near as hard to program as a randomly generated map where the AI is expected to go on the offence.
Also did the AI have full visibility of where there opponents units were?
That wouldn't be near as hard to program as a randomly generated map where the AI is expected to go on the offence.
I doubt it. If you have 10 units in a battlefield of 20 hexes, there are more than six hundred billion (10^11) ways to arrange those units (if I calculated correctly). Some of those permutations are outside unit movement ranges, but the order of magnitude is like that. There are however heuristic technics to solve this problem: http://sander.landofsand.com/publications/AIIDE08_Chaslot.pdf
katipunero
Prince
+ 1 vote for more gambits, daggers and rogue tactics for the AI. I remember saying something before in favor of a larger ai database of tactics each of which triggers at random or at a certain percentage.
Louis XXIV
Le Roi Soleil
OK, 2 things. One, he's also responsible for Civ2. Second, the solution mentioned in the argument is one that was borrowed for Civ5.
I'm hopeful Civ5's AI will be improved massively. But there's no point in criticizing previous Civ producers ad hominem.
If you guys think that Civ V AI and how it works with 1UPT frustrates you.... you should ask those who play Civ almost exclusively Multiplayer and highly competitive.
P.S. Kinda ironic when i think that 1UPT was the feature that had me really going for Civ V before it's release...
P.S. Kinda ironic when i think that 1UPT was the feature that had me really going for Civ V before it's release...
Yes, it is 20! / 10!, which gives 6.7e11
Jerrymander
Epistemologist
But this topic pertains only to singleplayer.
jdog5000
Revolutionary
That was a very interesting read.
Zhahz
PC Gamer
I'm also not buying that 1UPT is so much more drastically hard to code for. The system involves clear rules. Programming around clear rules is easier than coding for nebulae.
And yes, at the very least, the AI shouldn't be making utterly moronic mistakes like moving ranged units into melee or out of positions of superiority. Even if the overall AI movement and coordination is lacking for whatever reason, it cannot be too hard to create a rule in code that says "don't freaking kill yourself" or "don't move out of a good position where you can inflict dmg and into a position where you're gonna get 1-shot"
Players are even less forgiving of AI that's utterly stupid. Weak AI is more tolerable if it appears to at least by trying.
And yes, at the very least, the AI shouldn't be making utterly moronic mistakes like moving ranged units into melee or out of positions of superiority. Even if the overall AI movement and coordination is lacking for whatever reason, it cannot be too hard to create a rule in code that says "don't freaking kill yourself" or "don't move out of a good position where you can inflict dmg and into a position where you're gonna get 1-shot"
Players are even less forgiving of AI that's utterly stupid. Weak AI is more tolerable if it appears to at least by trying.
Similar threads
