Reliability or unpredictability?

Hi,

I have a question which I'm trying to find out for a while and I'd like to ask opinions about it.
So, in game design we utilize both the predictability and unpredictability for a certain degree.

I mean if you do A, B happens, usually when you do A it's a good idea to let B happen, otherwise the player will be confused for a little and if you introduce more of this discrepancy, the player will be totally confused since they can't know when to do what to achieve their goals.

On the other hand, there is this interaction among humans. Humans are not reliable, they aren't predictable, especially when it comes to skills and ability.

The question is, do you think it's a good idea to build it in to game play?

Let me add some example.
Player is hiding behind the corner, the others are looking for they.
If the player is leaning out to see what's happening, the chance of their discover is rising a bit.
But, the others have some kind of perception ability which is different for everyone.
So sometimes, when the enemy is looking in the direction of the player will discover them, sometimes not.

What do you think, is it a good idea to utilize such RPG-like elements?

Obviously I have done some basic tests, but I'm very biased, I play(ed) a lot of RPGs both pen&paper and computerized, so I think I have a giant tolerance to unpredictable events.

And of course, it's applicable to many other nuances of the game play, but the question is if it's a good idea to implement and if it is to what degree?

 

Unpredictability is a tricky topic in game design. It's nice to add variance to the game, but you don't want to annoy players by arbitrarily changing how things work.
A stealth game is a great example as you have pointed out.
In stealth games, there is a learning curve where you start to understand how guards move, what they can see, how far they can see, how effective darkness and cover conceals you.
Learning how this all works is part of the game, but if you start changing things on a per guard basis, you turn your game from a game of skill and practice, into a game of chance, and your learning curve just turns into spaghetti.

Imagine this scenario: You've made it past 5 guards in a level, slipping past them in medium darkness at about 10m range, you're almost at the end of the level and you find yourself in the same conditions for the last guard, so you make your move and - BLEEEP! the guard saw you because his perception stat was 2 points higher than the other guards!

To me, that's just frustrating, and I would not want to play a game like that.
However, that's not to say you have to do away with variation completely.

Let say that guard in particular was wearing clearly visible glowing night vision goggles which cast out a beam of green light which shows his extended range. Now we're adding variance, without the element of chance.

I think it's fine to have variance, as long as it's learnable and understandable.

 

I will second what @Serinx mentioned.

Certain variance is ok, in certain game types.
All is fine, as long you can weight chances, for minimum and maximum chance of happening.

Lets say you fight enemy, and you can hit it with 15 to 20 damage, with some % of critical hit.
You probably take an average of 15-20, before attacking, to evaluate your chances. Extra critics hits are bonus, but shouldn't be relay on it. In worse case scenario, you should have option of retreat, or relevant escape action.

In RTS for example, you compensate uncertainty, by sending more troops. Probably much more than minimum requirement, to ensure win condition.

But in platform 2d arcade type game, if you got unpredictability of some enemy jumping on you, or maybe not, then your chances to win shrinks dramatically. Which may increase frustration and which reduces enjoyment.

 

It's like asking, "should I salt food?"

Some people might say, "cavemen didn't need salt, and neither do you!"

Others will say, " what kind of barbarian wouldn't put salt on their food?"


The smart person ask each individual, "how much salt would you like?"

 

Wonder how that affected their life span
Hence a game ...

 

Salt War's BCE


An extremely unpredictable action RPG in which tribes who don't have salt attempt to massacre tribes who have too much salt all the while everybody agrees salt is bad for their health.

 

I'm inclined to agree with @BIGTIMEMASTER. There is no absolute correct answer to this. Players enjoy both Frozen Synapse and Dwarf Fortress.

Random musings to consider:

How often is your game intended to be replayed? Single play troughs should have very low levels of randomness. Multi play through games can gain a lot of replay value from random elements.

Make the randomness big. Lots of randomness in small areas doesn't give that much value, and may frustrate the player. Go for big elements. Game defining enemy encounters. Key story moments. Make all options roughly equivalent in difficulty/enjoyment, but make them very different.

Give the player control over the amount of randomness. Diablo has a great example of this with lightening vs fire damage. The average DPS of both was fairly equivalent. But the RNG was wildly different. With fire the player knew "it will take me 6 hits to kill the enemy". With lightening the player knew "I could kill them on the first hit, or it could take 30". Shotgun vs sniper rifles is another good example.

 

However Diablo (specially Diablo 2 from my memory) was linear as hell (pun intended), I loved it for randomized loot (including names) and maps.
Restarting, or even loading last saved game, has regenerated whole region.
So even playing Normal, Hard, Hell with colleagues online, didn't made it so boring.
But that was decades ago of course.

 

Correct. Diablo 2's quest/story system was completely linear. Drops and encounters were randomized. Maps were a hybrid, there were four maps for each region, and the game loaded a different reason each time you played. It gave the impression of randomness, but after enough playing one tended to learn all of the region combinations.

 

Yep. It happened that I played once or twice alone D2.
I was able to finish first difficult level. But don't remember passing second difficulty level, as I get typically bored alone, in mid game.

I hated jungle.
But liked desert.
And of course famous holy Secret Cow Level with Tristram.

 

Great point.

Yeah, little things changing randomly is whatever. As in, things that don't make enough of an effect on the gameplay to cause player to have to consider it carefully. So it's probably a waste of time then to design some method to make two types of low-level enemies randomly spawn if there is little difference in the practical smiting of the varieties.

But a horde of little enemies, or a chance encounter with a powerful enemy -- if player is guessing when this might come and always has to be ready for it, that is a total change of the game dynamic versus if they come at set time/places. If player has to stop and think, "eh, should I go foward as is, or should i grind a bit to get some gold so I can stock potions before I head out, just in case a mini-boss catches my scent."

That's the kind of thing that makes player get deeper into the game, and spend more time playing it.

 

I also really like how Stellaris does random event chains (at least in the early game). Players will randomly encounter starting events. These starting events give the players choices, which each lead down a random branching chain of further events. The whole thing is a massive tree.

The results are often game defining. I had a random encounter with a sentient black hole (random). My empire decided to send scientists to investigate (player choice). The black hole devoured the scientists (random). My empire decided to capture the black hole (player choice). The black hole was caught (random). My empire brought it back to the home planet and built a religion around it (player choice). The black whole escaped and shattered my home world into four desolate asteroids (random). As a result I spent most of the mid game scouring the galaxy to find a race that could reinhabbit my home world. Eventually I uplifted a primitive cockroach species that had a preference for desolate worlds, and resettled them on my home world.

That's the type of situation where unpredictability shines.

 

I haven't played that game, but that sounds like fantastic gaming right there.

Similarly, in The Long Dark, which I always use as an example because not only is it a fantastic game but it's also a small scale Unity indie title, so I think it's especially relevant here -- but anyway, you can edit random parameters but in the vanilla survival game, where you start and what you start with is random. Sometimes you get an easy start, but one time I started in a blizzard at night with a F***ing bear on my trail. My initial reaction was "screw this!" but I decided to just go with it and see if I could survive. After all, that's what the game is all about. Well, after probably about of an hour of desperate struggling, the Long Dark finally took me, but man what an intense hour of white knuckle gameplay. What makes it so great is that doing your absolute best is mandatory, but it's not enough. You need some luck as well. So when you somehow just keep going a little longer and a little further, it's just magic.

 

Sounds interesting. I trust you it was / still is executed correctly, to bring most of enjoyment.

Is that one, where Paradox has put its fingers and start mangling with a gameplay?

 

I think that positing this as a question of predictability or non-predictability is not the best way to go about it. Generally speaking, I see games as an abstraction of mastery over an environment (just like pretty much every 'game' of life). The only difference with RPG games being that the player is (usually) given an identity within a meaningful story while mastering the game world, making it a much more personal and characteristic experience.

The crucial thing is that a game must be able to be mastered, at least to a level of great satisfaction. For me, unpredictability is not so much a tool of generating new gameplay content (as it seems to be often used for, almost always to the detriment of a game) as it is a reminder to the player that they must pay attention or else stumble a bit. It's a way to say "you can master this game but you will never be 100% comfortable". It's a small symbol of the eternal unknown. It's much more subtle than what it's generally used for in practice.

It follows that if a game can be mastered, it must be consistent. And this leads to something that I think is almost a rule: If a player is ever faced with a situation where something they thought was true was not true, then there must be another truth that supports every outcome they have faced leading up to and including the current point. This means that underneath everything, there must be structure and logic, and consistency. There must be a truth about the game world that the player can arrive at, eventually, that basically explains everything they have and will encounter in the game and how to succeed with it.

The popular idea (not saying this is what you're doing, but I think it's worth mentioning) of springing unpredictability - or worse, sheer randomness - on a player as a sort of pseudo-meaningful experience is in my opinion lazy, unless it's something fairly irrelevant like varying the colors of the flowers on the field. The reason why is that reality is not random, or if it can be (at the level of our human experience) we haven't encountered it yet. And our entire consciousness is dedicated to formulating structure and meaning out of what we don't know and understand, such that throwing a bucket of randomness into a game just because one cannot be bothered to structure it is almost criminal.

 

Many of the play throughs aren't quite as spectacular. But sometimes all the stars align and the story comes out brilliantly.

Mastery in games is one reason to play. But its not the only reason to play. I doubt it its even the dominant reason why people play games today.

 

I do wonder, have you ever play civilization games for example?
No single game play is the same. That doe to randomness and unpredictability, specially at map generation, which has critical effect on game play.
Hence great replayability.
Does it mean civ devs over number of years were lazy, with each civ iteration?
Unless you mean something different, or I misunderstood you, your argument will fall a part in this case.
Not that randomness is good everywhere. But is in number of cases, when executed well.

Multiplier RTS games, introduce high volume of "randomness". Is hard to predict other players actions. There are things you can learn about players, but still "surprises" are not uncommon.

 

Wut? Reality IS random. Crash course on entropy and reality:

I think you are arguing with someone else. I'm not talking about laziness, I'm talking about deliberate implantation of random. But if you reread my initial post and take the time to understand, you would know.

BTW thank you for your opinion in the first three paragraphs. That (about the mastering the gameplay) will be taken into consideration and obviously I will reread everyone's posts and distill a standpoint afterwards.

 

I disagree, I think people play games pretty much always for the purpose of mastery. And I don't mean in an absolute sense, just a relative sense of "I'm good at this, I'm winning, this is fun".

Random map generation has an impact on gameplay, but it does not invert the rules randomly. It does not lead to a different outcome in the same scenario. It creates a different scenario, but not one which contradicts previously learned truths about the game and how to win it.

Bad randomness is when outcomes vary for no apparent reason. For example if you're playing a stealth game and you have to shift across a doorway, you would expect an enemy to have a consistent probability of seeing you at a particular angle, distance, and amount of light. Otherwise you can't formulate a good idea of when to make your move. If they are sometimes looking right at you but don't see you, and other times see you out the backs of their heads, it's not something consistent enough to master, and it's just confusing. And it's not good enough, in my opinion, to simply say that in reality, outcomes vary in unknowable ways. For one thing, this level of attendance to reality is in all likelihood inconsistent with pretty much everything else in the game. For another, in reality, there are always signs and signals too subtle for games to reproduce. For another, randomness in games is very easy (and tempting) to abuse. Games are not representations of reality, they are representations of very limited, distilled parts of reality that, like any stylistic artwork, must be carefully balanced in terms of what they focus on and filter in and out.

Lastly, multiplayer games are different, because there is a human being on the other end of the line. That changes everything. Even if human beings were random (which they aren't) the exercise in futility would still be worth risking, to try to prove that wrong and master a game played with them, since they are the most complicated and advanced thing produced by the universe and what they do instinctively means so much to us as fellows. Playing against a random number generator on the other hand is not particularly worthwhile.

Entropy is negligible at the level of conscious human experience, and in the context of the human lifetime (let alone a 1 hour gameplay session).

I didn't want to sound argumentative, and (like I said before) it's not that I think you're being lazy. But I think that it's easy to believe that randomness adds something when it really doesn't. It's something I've faced and thought about a lot when tempted to pursue simple procedural algorithms for generating game content. Maybe I was a bit too dramatic, but I really do believe that to formulate an idea of what a good game is, you have to know what a bad game is (and have to believe there's such a thing as a bad game). To know what a meaningful game is, you have to know what one is that's not worth playing. And I've come to the conclusion that a game that cannot be mastered is fundamentally a bad game. It's a prodding of our instinctive desire to know something new through play, but without reward. Worse, it might also be an attempt to escape the consequences of players judging the quality of any structure we can conceive of and implement.

I think that a game exists for two possible reasons 1) To teach a player something new and/or 2) to strengthen the system that we use to encounter things we don't know and transform them into things we do understand. Anything that does not do any of these things (or worse, confuses and fragments our perceptions) is not worth playing.

 

If you look at human psychology, you will realize, people play games, to satisfy they need for a reward for doing things. There is also significant element of the challenge. Taking any randomization, reduces challenge significantly.
For which different players, have different demand on challenge and satisfaction. From very trivial, to highly intellectual.
Game which is linear, gets boring relatively quickly, if repeating over. Often once you won, is no point playing again.
However, dev. can design levels, in pseudo random manner, to keep players occupied for long time.
See for example original mario maps.

Imagine now Tetris is linear. You will get the point, if you ever play Tetris.
Player would memorise movements, while progressing and repeating game.
How many times you would enjoy playing Tetris otherwise?

Seams you trying picking bits, which suites you best. We talking about randomness in general for game design. Not randomness of rules only. Otherwise, we can argue that when I click a mouse, my player do action. Thats the rule. Despite the fact, what cause me to click mouse in first place? It appears you generalize, not really responding based on experience, playing in this case any civ games. Otherwise, you wouldn't be replaying in this format.

Civ series are very random based game. It has rules true, but player will never know the outcome. Player may loose, or win, depending on range of events, of which many are random. In Civ IV for example, there is actually random seed tick box, where if you check it, every time you load a game, for example when you loose, the next upcoming events are randomized. Otherwise, events happens in the same order, as per seed.

How that is different?

If I play for example fps every time same, against my colleagues, they will learn my moves. Or vice versa.
So I am not taking stairs this time, because MAYBE, my opponent will camp there. Hence I will take one of two remaining routes. Lets say left, or right. I can use to the extent my judgment, experience and prediction, to estimate my chances of shooting an opponent. However, at this point I do it RANDOMLY, hoping for best outcome. For each route I may know where these will lead me, but I don't know what is behind the corner. From my perspective, this will be random event, for which I need to be always ready. Then sometimes got afterthoughts like "maybe I should take other route?". But that after fact.

In my understanding from this statement, you may believe, that Notch has predefined each minecraft map generation outcome? Or let algorithm to do it.
Yes there are certain rules, like biomes, but driven by randomized number. And yes you can use seed number, to regenerate same map. In vanilla players tend to choose options, for unknown landscapes. Which brings a new challenge. Which brings back to the first point.

 

Something something, no accounting for taste, sure. I always feel like that's an escape hatch- the designer shrugging and saying, "I dunno, you decide!" but by giving up that power, they can hamstring their own ability to bring the best out of a game. I feel like you know this, but I felt compelled to say it anyway.

I wanted to point out that you've actually got the choice to not take a risk at all by remaining in cover. That's significant as long as not peeking != not playing. You can choose a reliable option or accept a measured un/less-predictable one- player autonomy is preserved because they are choosing to accept risk or not, and they have a reasonable idea about the scale of risk they have accepted.

In high level terms, both are tools. You add randomness to something when you want to spread out the result across a 1 dimensional spectrum. You did a sneaky thing- how much were you caught? And then you weight the risk according to some factor so that the player can make good judgements about how likely certain outcomes might be. Note that if you don't communicate the factor that weights the outcome, the player is always just flipping a coin- good for creating tension over a decision, bad for preserving player autonomy. A choice perceived to have equal or unknown chances of outcome isn't really a choice. Not taking the risk == not playing the game. My fellow americans will recall the 2016 elections. (excuse my digression into snark)

The result should always have meaningful design implications- How much you were caught directly influences how much the landscape of the situation changes. You can take the percentage of how much you failed and scale the response accordingly because its distributed- you can funnel values into 'result ranges' to make them easier to work with (0%-30% == concrete result #1, 31%-50% == concrete result #2... ect) OR you can just apply the value. (How much is it raining? -> 65% of as hard as it can possibly rain.) Either way, it gives flexibility you don't have otherwise, at perhaps the cost of making more complexity.

Mark of the Ninja would not benefit from spreading out how much people see you in this way. The AI only has, what like 4 states? Unaware, Alert, Sighted, Terrified. The system isn't rolling dice to see what reality is, and you're able to learn how to manipulate the AI as a system because you can expect certain reactions from them when you do certain things. You're able to gain mastery of the AI because you can see and understand how it works.

I guess if you're pretty flexible about the meaning of 'mastery' I can kind of go along with that. I can't have fun at something I perceive myself as having 0% chance of winning/doing well at. If skillful play in XCOM means I know how to make decisions that result in the opportunity of taking actions with higher chances of success while limiting how successful the actions my opponent can take are, I would consider that 'mastery' of that game, in spite of the fact that I can miss 3, 4, 5 90% shots in a row because rngesus decided he hated me for a hot minute.

A multiplayer experience is different because you can't use characters to set up situations. You can't predict what a player will do, but you can play with what their available options are, and you know what the rules that everybody has to abide by are. (ie. granting newly-respawned players a short invulnerability shield so that they can play more aggressively than they would otherwise. Or, In a round of hearthstone, both players naturally generate mana at a fixed rate so that the match usually escalates over time.)

TLDR: Use predictability and randomness as tools instead of 'ways of life.' always have a good reason for why something is random as opposed to deterministic, communicate how this stuff works to players and give them the information they need to have autonomy when they take risks.

 

It's not to be taken literally. The point is, know your audience.

You don't ask your audience who they are and what they want. They don't know. You have to know them better than they know themselves. How else will you sell somebody something they don't need?

How do you catch an animal in a trap? You watch it until you know it's patterns. Don't trust what people are saying -- words are damn near meaningless most of the time -- watch what they are doing.

The other point is, don't try to simplify complex situations into rules. This is a flawed way of thinking in which a person tries to categorize everything they experience into little boxes they can put away neatly. It's an attempt to say, "I understand," so that you can then stop thinking about it or feel secure or whatever your subconscious motivation may be. I mean, I don't think the OP was trying to do anything more than start a discussion about the topic, but the way the questions are worded suggest they were looking for an easy answer. Like, "use 40% unpredictability in this case, 60% in this case, and voila, you'll make a masterpiece."

 

You're right, I am asking for opinion on the topic, I'm not looking for guidance, I'm looking for who thinks what about the topic. It is a possibility that the fact that English isn't my first language may give you the impression that I'm looking for exact answers what I should do, but I assure you, this is not the case, I merely tried to initiate a conversation on it.

 

I like games that are deterministic. I think also from a psychological point of view it’s very encouraging. You remember encounters so it makes the second playthroigh easier. It requires that the game design be legitimately difficult too and not rely on difficulty by obfuscation.

I’m a speed runner though and we’re a small percentage of players. But lots of us grew up on games that you could completely master by repetition. There was a time that I could make it through the entirety of Contra without getting hit once. That’s simply not possible with a lot of games.

 

At least in board gaming people are steering away from pure randomness nowadays when going for some level of realism. It's pretty awful when a combat system relies on dice hitting some increasing target number as the game progresses, but the only solution to increase your chances is more dice. That often means you'll have a ton of failures just more often than a ton of successes at even hitting something big and nasty, which doesn't feel right when you're playing a trained master swordsman or whatever.

Randomness is great for loot determination and varying how much damage you do though. If you're good with firearms you should hit even moving targets fairly frequently, and the target's protection/toughness should make little difference in the damage if you hit the right spot. No soaking up the full effect of a .45 to the head just because of hit points. That's oldskool gaming

Randomly determined defences are also great. Instead of combat becoming a long series of rolls which do nothing, you can at least whittle away at the health of something that rolled max health. If you also randomly generate resistances it can become a puzzle of sorts - an extreme case of this is found in Shadow of Mordor. Most of the enemies gain strengths and weaknesses over time, for instance surviving being defeated by the player and coming back a little angrier and tougher, but afraid of fire because they were pushed into a campfire.

The puzzle is using the environment to disable whatever strengths they have which make them hard or impossible to kill (some nemesis-level orcs have perfect counters to special moves and block everything else - unless panicked). Sometimes the only solution is to ride a big beast and have it eat them, but if they come back at least there's a chance they'll have a new phobia.

 

I don't play games to master them; I play them to escape reality for a while, to see something new, or to watch a story unfold (maybe the story of how my little settlement grew into a huge trading empire, warding off evil rivals). I often don't bother to play games to completion (mastery?), because the rest of the game is just more of the same.

Hmmm, I should specify that I'm thinking of strategy games, particularly tbs ones. Other types of games, such as platformers or arcade games are based on a goal of mastery. Really, all game discussions should have a little checklist of what genre and subgenre of game is being discussed. For tbs games, for example, the design goals of single-player are completely different from multiplayer.

As for random variations, I like some variation for replayablity and to keep me from making the same choices every time. Changing the rules is okay as long as the player isn't basing his decisions on the rules remaining constant. If the player is required to make a choice, he should know what the choices are at that time, and that they'll remain constant at least for the time specified. Random changes just to screw up the player are not fun.

 

Unfortunately the variations in civ are superficial. It doesn't really make a difference to the development of your empire if you get luxury 'a' or luxury 'b' near your starting city, or if your nearest neighbour has a beard or wears a dress. It's just a race to fill your bucket (with gold, research points, culture, or military units) faster than your rivals. It's so shallow that I can't be bothered to start a new game. I'd like a game where random variations make an actual difference in my strategic decisions and the outcome of the game.

 

Surely no difference, if you end up on small island or without strategic resources in a reach.

 

Your example is a fairly rare case of extremes. My comment was for the more common starts. Also, it doesn't really change your overall strategy; it just creates an unexpected obstacle which sets back your timeline. If you were planning on a domination victory, you're probably still going to do that, and what changes is which turn you defeat the blue rival or the red one, unless you fall far enough behind to lose.

 

In not uncommon to end up wihtouth one, or other resource in reach. Specially when you set uneven strategic resources generation at start. That means, for example if plan playing passive, for diplomacy victory, with economical lead, and nearest critical strategic resource was generated on enemy land, you know, you need capture it, or trade it, or you will loose. And AI, or player can be pain in back side, in regards of trading. So you may need to attack in this case, as an option. Which requires to change a strategy. Even if it is temporally.

So what defines in your opinion "unexpected" obstacle?
That can be random event, of the pool of possible options, if you want to put it that way.

Which may be result of lack of resources. Or settler eaten by barbarians, if not protected.
Or do to many other "unexpected" events. Which are likely being randomized at some point.

Throwing a dice is pretty good randomization event, as it is near impossible to replicate exact conditions, to throw it exactly the same way. Card and paper RPGs utilised different variances of dice, to make game unique as much as possible, for every play. Players knew, even playing same character, having genera strategy, could not lead always to same outcome. And even then, if predicted outcome did happen, is least likely to happen in exact same sequence and iteration, as previously, giving enough permutation options.

 

I just find Civ5 lacking in true variation. Starting on a tiny island makes your rate of progress different, but you're still following basically the same strategy: outbuild/outfight your rivals, following the same tech tree path and building path. Even the different victory conditions aren't much different. The differences are just too shallow to make the game interesting for me. I'd like the game to surprise me, and force me to reevaluate my strategy and maybe make a drastic change. I'd like my empire to be memorably different from previous games. At least in Civ1 I had distinct gold-producing cities and hammer-producing cities (producing caraqvans to help build wonders elesewhere). Civ5 does offer specialization, but somehow it seems less distinct than in Civ1. Overall, I feel that Civ5 is shallower than Civ1, which is really sad given the 20 or so years of game developing experience and several orders of magnitude of hardware capability.

 

Civ is a really good game series to look at on this topic. It actually includes both a lot of randomness and predictable reliability at the same time, both arguably in just the right doses.

You have randomness of map generation, but within certain non-random rules the player can control. You can have random leader/civ choices for AI or manually select them at the player's option. You have random placement, and don't even discover the shape of the entire world until relatively late in the game. The AI can have somewhat random appearing behavior, where you end up fighting multiple enemies across several fronts out of nowhere, unexpectedly turning a long time friend against you, when your strength should well have deterred attack by your known foes.

At the same time the AI's behavior isn't completely random, as the AI won't generally attack if it would expect to be overwhelmed, and is less likely to attack a long time ally on their own than someone with poor relations. The AI is tweaked based on personality traits of the various historical leaders (or by a nod to a Civ 1 bug in the case of Gandhi the maniacal nuke thrower). AI battle tactics are generally fairly predictable, with few instances of anything remotely sneaky like you will see from human players. Once you learn how the AI fights in each iteration in the Civ series, it is fairly easy to come up with your own tactics, unit compositions, and advantageous land and sea positions, to maximize your ability to defend from an attack or push forward your own with good repeatability. Once you learn how to best defeat the AI in combat, you generally won't lose a war unless facing overwhelming numbers or overwhelming technology.

All in all Civ is a good mix of both randomness and reliability, allowing for learning to master the game at the same time as having good replayability. It also allows for the chance to win using completely different strategies, where the game is completely different if you choose to be an aggressive conqueror from the start, or someone who turtles up until your tech advantage allows no one to stand in your way.

 

Definitely + for that, famous (Osama) Ghandi issue ;p

Yep, I like its balance.
And dose of predictability is also good, so player don't get completely frustrated every time, with rage quit
While a bit annoying, I liked mechanics, where loosing too much units, could turn your best friend into foe.

In civ IV recalling also aspect of climate change. Where certain tails change their role. I.e. from tundra to desert, from jungle to forest, or something like that. Which again, changes economical aspects.

 

I first started playing the series with the original DOS version of Civ 1 when it was a new game, and I still clearly remember the first time I was doing well and had made it to late game, Gandhi to my north was big but I wasn't concerned because he had been nice with me for thousands of years, and then it was like a switch flipped where he was threatening nuclear war. It wasn't long before the nukes started raining down, I had to throw them back at his armies and cities, millions of virtual people were dying every turn, and I was yelling "How the **** is this something Gandhi would ever do? He's a ******** pacifist!"

It wasn't until at least a decade later that I found out it was a very simple integer overflow bug when Gandhi changes to a new government type that becomes available around the time you can get nukes. I got a huge laugh when I found out.

 

Not sure in how many civ, but in civ IV, I was dropping rain of nukes on my enemies. Every second city was building nukes, during long continuous war, against big faction. Then I have received notice, from global government, or scientist, don't remember, informing me that if I continue the rain, the earth my crack. Well me educated, of course I ignored their warning

And guess what, few nukes later, Puff ... game over. Earth crack due to immense strain of nuclear explosions
Had to reload game and manage without more nukes

But you see, that when you starting understand what is behind a scene
Even more interesting would be, what made Ghandi to change.