I was just thinking about various difficulty settings in games like GalCiv2, and how they are usually scaled against some nebulous "average" gamer out there.
Wouldn't it be cool to tailor the difficulty settings to each player? Keep a profile of each human playing the game (on the local computer) so when you ask for an "normal" difficulty game the AI would adjust the AI to give YOU fairly challenging but beatable opponents? Maybe even advance YOU in rank over multiple games as your skill improved. So when you ask for a "difficult" game after playing and winning the last 20 games in a row the AI would adjust to give you a true difficult challenge.
Of course, even without profiles there is always the option to just choose a higher difficulty setting. I just think the ability to increase in rank over multiple games would be a nice plus. It would give more of a feel of customization to the game.
I think difficulty should work as follows.
Beginner - you win. Barring that you surrender you are going to beat these pathetic opponenets silly. This setting is for learning the game, and for people with very fragile egos. Winning at this setting would give NO rank advancement to your profile.
Easy - still an almost certain win. The Ai should put up enough of a fight to give new players a sense of accomplishment by winning. Winning gives minimal advancement.
Normal - A game that requires some knowledge and skill of the game to win. Even a well played game should have some possibility of being lost, although the player should be favored as long as they play well. Winning gives "100%" advancement. (i.e. 5 wins at this level increases a rank)
Difficult - You are going to have to know what yoyu are doing and make few mistakes to win. Even with a well played game, the computer players have a good chance of winning. Winning gives 150% rank advancement
Painfully difficult - Advanced players are going to have a rough time. Most players are going to get beaten at this level, sometimes humiliatingly. Winning gives 200% rank advancement
Impossible - Even advanced players wet themselves against these opponents. Victory even for the elite player is going to be the exception rather than the rule. Winning gives 500% rank advancement (i.e. 1 win = 1 rank advance)
As far as ranks you could start out as a "sorceror's apprentice" and move up thru various ranks (top rank being Master of magic, of COURSE!)
I'd want personalities to be in different catigories rather than just "beginner, easy, normal, hard, etc." Like I'd want "rushing AI easy, hard, impossible" and "culture AI easy, hard, impossible" maybe "turtle AI simple, advanced" (easy and normal wouldn't really apply, since you could probebly still be able to rush them). You know, something other than just 'easy, normal, hard' or 'noble, warlord, king' or whatever set you can imagine. Something that better describes how the AI acts. Naturally a turtling AI would be easier than a balanced AI. Some players might be able to deal with both of those, but not the rushing AI who always just presses the offensive and never lets up.
Oh I agree completely. The AI need to come in different flavors, and maybe even have a somewhat schizo personality that might behave one way and then change personality types throughout the game. There should be honorable AI that once you have made an agreement with them, their word is their bond - but also if you break your word to them this personality type would be very vengeful. Or a doublecrosser that always did what was best for itself and would break a promise in a heartbeat. The AI's should have different motivations. There should be expansionists, turtles, rushers, those who focus on infrastructure, those who relentlessly pursue spells, and dozens of others. The difficulty and rankings (of the profile) were in addition to this.
Excatly right denry
There is a big difference between a "hard" personality that follows a certain lines of code or honor but is very smart, and a guy who randomly says "Man, my nose itches, I think I should take it out on you" and declares war. The latter is how I felt the difficulty in most TBS games is created.
(I want an AI personality that says the above line)
Hard AIs should not pounce on every mistake at lightspeed. For one thing its not entirely realistic, and secondly if they wait you may compound that mistake with another.
You know what I want more than anything when it comes to AI? A truly adaptable one complete with a certain amount of unpredictability. It always irks me that after just a few interactions or a little espionage I can tell exactly what type of AI I'm up against, and thus gain an advantage over any and every AI player because I know how they will react to things and their overall strategies.
I want an AI that constantly makes contextual decisions about its grand strategy. For example, let's pretend AI Player A starts out surrounded by five other players. It decides that being too aggressive would simultaneously annoy many of its neighbors and weaken its defenses, so it chooses to dig in like a good little turtle. However, after some time two of its neighbors (B and C) go to war, and player B gains the upper hand - but not by much. Player A then strikes a deal with Player C: A will join the fight against B in exchange for an alliance and a constant tribute of iron ore (A has very little iron in its territory, and B has a small surplus). A increases its military production and draws from its defenses to make good on the deal. This turns the tides on player B - the combined strengthens of A and C overwhelm B, who is defeated in short order.
The results of this for Player A? It gains about half of B's territory, an alliance, a steady supply of a resource it was short on, and one less neighbor (at least for now). Now, with B defeated and C an ally, A has only 3 potentially hostile neighbors. It now decides to use its increased territory and resources to build up a powerful military, stops turtling and goes on the offensive with its new ally. Once A and C defeat or ally with their remaining neighbors, A explores its surroundings and discovers that there are other very powerful kingdoms/alliances not too far away. A decides to revert to being more of a turtle again, building a strong economic foundation, and focuses on both magic and technological research; it also does a lot of quests for the shiny rewards. It maintains a strong defensive position so that would-be invaders would think twice before attacking.
From here, A could decide to go on the offensive again, or it could go down the diplomatic route, or it could continue to turtle, aiming for a magic victory. Or it could try to complete its Master Quest. At this point in the game, all of these strategies would be more or less equally valid.
Moral of the story: I want an AI that can play like that. I have never played a game with an AI with that level of versatility, and in my opinion such a believable AI alone could make Elemental memorable for a long time to come. I would even be happy if they do away with the standard types of AI (turtle, aggressive, tech, etc) and focus solely on making one truly versatile AI, maybe with some tweaks here and there for different factions and races.
The only thing I want from varying difficulties is that the AI should never, ever, ever, ever cheat. Make it do stupid mistakes one lower settings, make it ungodly strategic on hard, I don't care. Just never, never, never, ever, ever, ever, ever make it cheat. When I get added disconent in my cities on higher settings and the AI doesn't, that's cheating. When the AI produces units at twice my speed on higher difficulty or gets bonuses I don't, that's cheating. When they get free money, that's cheating. When they can see into fog of war, that's cheating.
How the difficulties are handled from there on in, I really don't care too much about.
I disagree, Luck, If you only want to play at the highest degree of difficulty WITHOUT cheats for the AI, I agree that should be an option. SO they should make it clear at what levels of difficulty the AI starts getting cheats and what those cheats are.
However, there are some players (my son for instance) that there are some games that he can beat the game even with the game getting double production bonuses and other penalties for himself. In a perfect world, a GOD AI could be created that could at optimum setting outplay almost anyone without cheats. We don't live in that world, and if they have to let the AI cheat in order to be "nearly impossible" to beat, then I think it should be included. Again, they should make it very clear at what difficulty level cheats are introduced.
Yeah, but don't you think that having cheating as an option to increase AI difficulty stymies the motivation for developers to bother creating genuinely decent and hard to beat AIs? I mean if you're gonna give in to using cheats anyway why bother spending time on improving your current AI? Why wouldn't you just stick with your crappy AI at all levels and just add more and more cheats? It sure as hell would save time and development resources.
Well I guess it depends on the discipline of the developers. I think everyone, and particularly Stardock, is VERY committed to making the best AI possible that can win "fair and square". And with RTS that is even fairly doable just due to the AI to be doing a thousand things at once but still play within the rules.
In a turn based strategy game, it is incredibly tough to make a non-cheating AI that can beat top human players. It's ability to do a thousand things at once is to some degree mitigated if not eliminated. Think of how difficult it is for a chess program to be able to beat top human players. Now take a game like Elemental where the "ruleset" is much more complex than a chess game and with multiple ways to win. Coding an AI to even give a solid player a good game "within the rules" is a tough prospect. One that can beat a strong player is doable, but HUGE amounts of resources would be required to do so <IMHO>. I would love to be proven wrong
GalCiv 2's hardest AIs cheated. I think tough (7/10?) was the highest difficulty level in which the AI didn't cheat. And GC2 has damn good AI compared to almost any other game. And considering Stardock made GC2, I don't think that's a good argument. If Elemental were being developed by a company like EA I would agree with you (but our opinions wouldn't matter anyway); but I think it's a different story with Stardock.
So I agree with Denryu. It'd really easy to take the highest difficulty level and make a few even harder ones by allowing the AI to cheat. There's no reason not to add them in for masochistic players.
And making the entire game optional is a can of worms I really should get around to create a thread about. Both things - "optionals" and the AI cheating are unnecessary crutches and excuses to avoid taking responsability. The "I SAID OPTIONAL!" claim is not a valid defense.
You know all the pieces at once, you know the positions, what's in play is finite. In a TBS, you can't think 8 turns ahead of time in the same manner (nor do you always have to).
One of the 'cheating' difficulty levels in GC2 is *called* Masochistic (I think the top is Suicidal?). I've always taken that as a sign that the devs understand they have an audience to serve there, but it's a slightly weird one.
I've played GC2 since pretty much launch time and mostly on equal terms with the AIs (Tough), so between just learning the ropes and keeping my game updated with the steady stream of improvements, I've never felt the need to let the AIs cheat. For the expansion packs, I actually found the best difficulty changes for me were the number and/or choice of AI opponents on the huge and immense maps I prefer. But then I'm more or less a turtler and more than half think of GC2 as a toy, not a game.
I would say this argument proves why a good AI (for a TBS) is MORE difficult to program. In chess, as you say, there are a limited number of moves, it is not that hard for the AI to look ahead at all possible moves for the next x number of turns and determine what is the optimal course. For the very reasons you give, it makes it difficult to make a good AI for TBS games strictly because of the nearly infinite possibilities that the program would need to examine and evaluate.
Now I completely agree with you as far as the AI being able to cheat and see into the fog of war. But giving the AI productivity/research bonuses I think is fair play as long as it is explicit for the difficulty level what is happening.
You call it a copout, I call it a valid option. SD has already bragged about the AI and how it is going to feel like playing another human. So I don't see them taking the easy path and jsut putting out a crap AI with cheats to make it competent. I expect and hope that they put out a competent AI that can be made ruthless with (cheats, advantages, whatever suits you)
Cheating AI will result in me -having- to play against it at one point or another, to get a comparable relative difficulty level to my peers. It becomes a pissing match who's playing against the most unfair odds, rather than who's actually the better player. It's the same reason I oppose this obsession with optionals, or the attitude of some people in other games when it comes to overpowered features; "Don't use it, then" - One should -never- have to neuter oneself to get a joyable playing experience.
Edit: And I never called it a copout. Don't put words in my mouth.
>.< I don't want to neuter myself for any reason. .... well, actually, I wonder if that would stop me from being distracted by every nice pair of boobies that walk by. I'd probebly be a lot more productive now that I think about it. I really don't need the aggression and other testosterone related emotions...
But stardock HAS boosted that the AI of this game is going to be very good, and they seem to be aiming to make it seem very human. So I expect it to be as such, with hopefully minimal cheating. *cheers* here is to fun AI
You don't have to do anything you don't want to do. Stardock including a few difficulty levels with cheating AIs does not in any way force you to play games at that level. If you feel obligated to do so, that's your own problem And quite frankly, plenty of your peers will be avoiding cheating AIs anyway - maybe even most. I, for one, don't enjoy playing against cheating AIs. GC2 is proof of that - years after it came out, many, probably even most, don't play on the cheating levels. The people who do that are, I think, mostly the people who are competitive with scores - or just crazy insanely good at the game.
This actually kind of reminds me of that discussion where some people wanted to prevent everyone from loading games from earlier saves, because they didn't want save/loading to tempt them or whatnot. It was a horrible argument in my opinion, and your position here is more or less the same argument in different clothing.
That argument isn't even relevant. The existence of cheating AIs has no bearing on the quality of the non-cheating AIs. In GC2, Stardock doesn't really spend dev time changing or even making the cheating difficulty levels except to keep them up to date with overall AI updates. They will develop the AI to make it as good as they can without cheating. Then once the game is out, they will continue to update the AI to be even better, without cheating. Those updates will be automatically reflected in the cheating AIs - because it's the same exact AI just with lots of advantages.
If you play at the highest non-cheating difficulty level and continuously steamroll the enemy over and over again until you fall asleep, then there is a problem with the AI, and it is totally unrelated to whether or not there are difficulties above that where the AI cheats. You may as well be saying, "If Stardock includes includes a difficulty level where all the AI units are replaced with bunny graphics, then I will continuously steam-roll the enemy over and over until I fall asleep." It's a nonsensical argument.
Adding difficulty levels is hardly adding "optionals." Based on your above argument, there shouldn't be difficulty levels. Everyone should have to play against the same level, whether it's their first time playing a strategy game in their lives, or it's the 100th time they've played Elemental.
For the sake of argument, it's important that you realize the internal conflict of interest. One one hand, you want to play against "the hardest possible", but on the other hand you don't want to play against a cheating AI - you want a level playing field against the AI, just like you want a level playing field when comparing your relative "worth" to that of your peers, opponents, rivals, enemies or whatever.
The resulting annoyance of being "forced" is detrimental to the enjoyment of the situation - i.e. the game.
"I think we should have a civilization of dragons, optional!"; "But.. Dragons are supposed to be incredibly powerful, capable of steamrolling nearly entire nations. That wouldn't make sense, contextually or lore-wise!"; "I SAID OPTIONAL!".
I'd argue that difficulty levels are options, but not optional.
Oh well with that I agree. Like in the GC2 metaverse (the ladder system) there are mechanisms in place to prevent (or make it a pain) to load previous save points of your game. I'm all in favor of that, because otherwise the ranking system would be totally broken and worthless. So long as there's a way for people like me who couldn't care less about scores and such to play and save and load however and whenever we want, I am all for there being a competitive mode that prevents it. Unfortunately the same won't really work in this situation, because many of the people who want to compare their scores will want to play on cheating difficulty levels...
I think the conclusion to draw from your post is that you need to see a psychiatrist, Luckmann ... Quite frankly I have no desire to play against "the hardest possible." I only want to play using the settings that make the game the most fun for me. In my case, that means the highest difficulty level at which the AI doesn't cheat (unless the AI is god awful and needs to cheat to even make you aware that it's there at all - but in that case I'd lose interest fast and it wouldn't really matter). It's a game, not a competition (unless you make it one). I don't find it fun to play against cheating AIs, but some people do. Please don't ask SD to leave out cheating AI difficulty levels simply because you have an inexplicable urge to turn games into a competition with.. yourself
Yeah now I see the argument. I also see the solution: therapy
I did indeed misunderstand you. I went back and read your post and it is actually very clear what you meant...
they need to do a few things to make the game difficult.
1) they need to automatically take cities when they are in position to do that. No matter how good the relationship with the player is.(humans do not care, so the AI should not care.)
2) If they get all the tributes they need from a weaker opponent, then they must immidiately kill that player off.
3) they need to make the AI focus on military, and overall make them more agressive.
4) You need to make the AI sometimes stop makeing new cities, research, building infrastructure during war. only units and some military buildings during the war phase. so that the war is over quickly and a large army can overwhealm the other player.
5) the AI needs to take into account where you have your army, and cities and how well the predicted reinforements are.
6) The AI needs to learn how to backstab an ally, when it increases there chance of winning. That also means declearing war when it is atvantageous even if there tributes are met!
7) the AI needs to prepare for a massive quick war just before they decleare war. Massing units in the prewar phase.
8) The AI needs to focus on infrastructure when they are far from an opponents cities and armies.
9) the AI needs to learn how to raid with fast units, with no other goals then to prepare for an invasion. To destroy roads, mines and lightly defended cities.
I have to say, at first I thought you were being sarcastic, but as I kept reading it became clear that you're serious...
That is a HORRIBLE idea. It would make me so angry. And you're damn right I care - why would I want to go to war with my buddy just because he left a city near our border fairly undefended (probably because he likes me) when there are plenty of hostile cities around for me to take, even if at a slightly higher cost? Doing this in many situations would actually be a dismal idea with severe repercussions. It would make the AI intolerable and bad.
Only if it's a convenient time to do so and if the opponent is significantly weaker. Oh, and if that opponent isn't acting as a nice buffer zone between a powerful warmonger and you. There are so many reasons why this should not be a universal rule, but a situational one (just like #1) that listing them all would make my hands hurt.
I also don't like this one. It seems to be the standard crutch - let's just make all our AIs angry, spearman-churning machines that do nothing but attack. For one, you can't focus on military without having a strong economy to back it up. While a belligerent military powerhouse is a valid strategy in most games, it is definitely not the only one - especially in games with various victory conditions (hey, Elemental is going to have lots of victory conditions! what a coincidence...). And if it's all the AI ever does, the game would not have nearly as much lasting appeal as it could have.
The rest of your points are much better - they are very valid points about where AIs tend to fail, much to their detriment. But the first three, wow. Please, no!
No, no, you quite plainly missunderstood me.I also find it interesting that I can't actually argue with you anymore, since you're not raising any points except "therapy". I -could- play at a lower difficulty setting, of course I can. I just don't feel that I should have to somehow neuter myself. Like the argument I raised before, it's the same basic argument that other retards raise in other mods or games;"The flying sasquatch unit is way overpowered to the point where I'm not having much fun playing the Sargasso Pirates civilization. We really need something done about this"; "So don't play the Sargasso Pirates, or don't use the flying sasquatch unit. Duh!"And yes, singleplayer games are competitions of the self. You get better and better as you play, for any variety of reasons. The only one you have to beat is either yourself, or scoreboards compared against other people. Wheter that's your piece of cake or not is another thing entirely, but not getting a comparable difficulty level to that of others is a problem, yes.If for nothing else than for reference points between games and players.
(This was originally supposed to be posted right after Pigeon's last response, before KnutAreMykland posted, hence making a lot more sense without direct quotings)
I want to live under the constant threat of being steamrolled because I don't know how strong my enemy is or where he is hiding his armies. That's not to say that I want to become a nervous paranoid wreck because the enemy is a complete sociopath.
Sorry if I equate "excuses to avoid taking responsibility" to "copout"
You seem to be arguing that being a superior player (since you seem concerned about ladders and such) means that someone who can beat the AI with no cheats is equal to or superior to someone who can beat the AI with cheats. Now I expect I am going to be accused of putting words in your moouth, but seriously is this not the concern - that in order to compete in some online pissing contest (your words) you are going to be FORCED to play against a cheating AI to compete? Well, I don't know I guess if they can soundly thump an AI that uses cheats and you can soundly thump an AI that does not use cheats - I'd tend to think they are the superior player unless you play against the AI using cheats and beat it.
But maybe the best peaceful solution so that you aren't forced to play in a way that you dont want but can still compete in the pissing contest that you obviously care about... maybe the ladders should be broken down by difficulty level. So keep a ranking ladder for the easy difficulty, the average difficulty, the hard difficulty and the impossible difficulty, etc that way no one will be forced to play in a way they don't want and they can still compete.
I'm an ambivalent latecomer to the GC2 Metaverse, and I don't think I'll be disappointed if there simply are no scores for Elemental. But if there are, regardless of whether Stardock maintains ranking ladders via something like the MV or fans do it independently somewhat like the AltMeta, I believe that a diverse set of ladders is the single best response to the otherwise-endless arguments that arise about just what sort of playstyle deserves top points.
Games as complex as even the original Civ, MoM, GalCiv, and MoO can be played and enjoyed very differently. Folks who want to compete are a subset of the total game audience, and folks who want to have their scores compared to *all* other players are a subset of the competitive folks. Sorting scores by difficulty level is just a good start, IMO. A good ladder set for Elemental would also sort for major playstyle traits such as map size and vicory type. I'd even enjoy seeing a dimension in a laddering schema that tracked games by your choice of faction and possibly other game setup choices.
Yes GW in fact as we were discussing this I could not help but think of the player database that Everquest 2 maintains.
I mean you can see who has had the top 100 magical hits, top 100 melee hits, who has the most in game wealth, who has the most status, how has the best kill to death ratio, who has died the most, who has died the least, who has created the most tradeskill items for crying out loud! and about 30 other things and you can search either globally or by server.
So I agree, keeping laddeers by difficulty level is a good starting point. Why not also track by the type of victory condition, map size, and anything else anyone can think of?
Personally I am a big fan of being scored, even if I don't use the metaverse. I like when it breaks the score down and tells me WHY I got the score I did : 20 points for how long it took to win, 50 points for my healthy economy, 10 points for my (lackluster) research, etc etc.
I love the ability to track my relative strength to all of the other civs throughout the entire timeline of the game. Being able to go back and do a "post-mortem" on my game, win or lose, is a MUST for me in a TBS, (and very enjoyable in a RTS also, C&C3 was awesome).
Well, the whole therapy thing was largely tongue in cheek, but not entirely... The fact that you consider playing on the highest difficulty level at which the AI doesn't cheat neutering yourself is absolutely ridiculous. Elemental is (or will be, anyway) a game. It's purpose is to be used by people as a method of entertainment - to have fun. Different people obviously have fun in different ways, and maybe in your case you have fun by playing every game trying to do better than you did the last time; always trying to improve. That's fine. The mere presence of cheating difficulties should affect that, though. You can still always try to improve on the non-cheating difficulty. But if somehow their presence is stressful to you, and makes you all of a sudden feel like you're being presented with two bad choices (even though one of those choices is exactly what you want), then you have a problem...
And I realize this is not your only argument, but it is clearly part of it. Otherwise you shouldn't have written:
"Cheating AI will result in me -having- to play against it at one point or another, to get a comparable relative difficulty level to my peers. It becomes a pissing match who's playing against the most unfair odds, rather than who's actually the better player."
"For the sake of argument, it's important that you realize the internal conflict of interest. One one hand, you want to play against "the hardest possible", but on the other hand you don't want to play against a cheating AI - you want a level playing field against the AI, just like you want a level playing field when comparing your relative "worth" to that of your peers, opponents, rivals, enemies or whatever."
And with regards to the whole thing about in order to really compete with other players you'd have to play on the cheating difficulties, there's this: if people get to the point where they can consistently beat cheating AIs, then chances are if said AIs weren't even there, most/all of the top players' scores would pretty much be maxed out. They would utterly destroy all opposition every game. What's the point in competing at something that isn't even a challenge?
There are plenty of people who enjoy strategy games because of the challenge. Unfortunately, after a fairly short period of time, many of those people get so good that non-cheating AIs don't pose any challenge for them anymore. If there's no way for them to make the game harder, they'll lose interest and move on to something else. If letting them up the difficulty by giving the AI all sorts of advantages will keep the interested and a part of the community, then I think it's more than worthwhile.
Pretty much all your arguments have to do with competition and ladders and the like. The regular singleplayer experience should never be limited based on ladders or multiplayer considerations, especially in a game like Elemental, where, let's face it, 92.35% of all games will be played outside of any ladder or competition.
There are many great features available to you once you register, including:
Sign in or Create Account
