What is the depth/complexity of Civ 5 compared to other games?

[MyopicCat]

An interesting way of defining the depth or complexity of a game is the maximum number of "distinct" skill levels between a typical beginner and a world champion (or more theoretically between random and perfect play). Here "distinct" could mean that the stronger player has >60% chance of winning (corresponding to roughly a 100 point ELO rating difference in chess).

Here are some examples of the depth of games (and criticism of the measure) according to this paper:

Go|40
Chess|16
Shogi|11
Checkers|8
Backgammon|4
Magic the Gathering|3
Poker|1

(Not sure what variant and play length of poker is intended here.)

Where would you place Civ 5 on this scale? Would the results be same in a pure duel compared to the relative results of two players in a multiplayer free-for-all?

Personally I think that the skill/luck ratio of Civ 5 is similar to that of Backgammon or Bridge, but I feel that Civ 5 still has a bit more "depth".

 

[Moriarte]

Depth, to me, is existence of interconnected layers, where each layer represents a field of knowledge.

Layers of multiplayer civ: diplomacy, warfare strategy & tactics, economics.

Layers of chess: warfare strategy & tactics.

Clearly, civ goes far beyond aforementioned adversaries, if you accept my definition of the word "deep".

 

[dexters]

IIRC , no one's been able to do a good computer AI to handle GO but there are plenty of very good Chess ones.

As a 4X game, I'd say Civ and other 4X games are nearer to Go in complexity than Chess, as evidenced by the general difficulty in making competent AI in any 4X game without giving AI massive bonuses or hiding the AI's weakness through gameplay limitations

 

[Memoryjar]

Against a screen is not like against a human in real life. It's another source of complexity and depth.

 

[MyopicCat]

IIRC , no one's been able to do a good computer AI to handle GO but there are plenty of very good Chess ones.

True, but Go engines have recently made enormous progress. As a club-level Go player, I could easily beat the best AI in the world 5-10 years ago even if I gave it a handicap of 5 stones. In chess terms, I was several hundred ELO points better. Today, to make the game even I need to accept 3-4 stones. If the AI is given the same 3-4 stone handicap, it can now occasionally beat high-level professionals.

Fascinatingly, the AI breakthrough is due to implementing a Monte Carlo algorithm. Basically, each turn the AI plays out a very large number of games where all moves are completely random, and then chooses the move that gives the highest probability of winning. The selected move is quite robust and often feels surprisingly intuitive. Of course there are all sorts of clever tricks involved to make this work well, but this is the gist of it.

As a 4X game, I'd say Civ and other 4X games are nearer to Go in complexity than Chess, as evidenced by the general difficulty in making competent AI in any 4X game without giving AI massive bonuses or hiding the AI's weakness through gameplay limitations

I suspect this is more an indication of the limited effort put into the AI of typical 4X games than of the complexity of the game itself. If Firaxis abandoned (or limited) the crappy heuristics that is the core of the current AI and implemented a quick random Civ game simulator, Monte Carlo may work very well for Civ as well.

 

[Alphons Rodulfo]

Isn't there a data set of multiplayer results available somewhere? That could be used to calculate the number of levels of depth, if the data set contains results form both the best and the worst players.

 

[paralistalon]

Aside from having an obvious hard-on for Go, the author makes an interesting assumption about game complexity that I really wouldn't want to agree with (as he/she points out, "who was born first" would clearly be more of a complex game than Chess if you're comparing ELO ratings). I also find it completely wrong to assume that Checkers is more of a complex game than Magic the Gathering from a face validity standpoint.

Now, I do think that comparing ELO ratings can reveal information on how much luck/randomness is involved in a game. The less random a game, the more of a chance that one person can "master" every possible move of a game, which is why Go and Chess are so much higher on the list. According to the author's proposal, a game is most complex/deep if you can sit down the best player in the world and they will win with almost 100% certainty over their opponent. I would argue that this doesn't make the game any more complex, just more consistent.

This is why the range of ELO ratings in Magic is so comparitively low: if you can lose even one match through bad luck alone (which is true), while infrequent, this will completely devastate your ELO rating.

In the Civ community, players are generally ranked by what difficulty level they can consistently win on. This progression, I reason, relies mostly on learning what to prioritize (science, growth, wonders). If you prioritize everything well, then you get a snowball effect, take the tech lead, and dominate the game in any way of your chosing. If you take an expert Civ player and put them against a newbie, the expert will almost always win because they will have the optimal strategy. But once the newbie learns this optimal strategy and becomes familiar with how many units it takes to conquer a given city, a game between them will come down almost entirely to who has the luckier starting landscape. Because of this, I would place Civ pretty low, like at a 2.

And this is just for 1VS1 play. If you're going to introduce multiplayer games, then the "complexity" becomes even lower because two or more players can form an alliance to easily beat the most experienced opponent.

 

[Arcaian]

And this is just for 1VS1 play. If you're going to introduce multiplayer games, then the "complexity" becomes even lower because two or more players can form an alliance to easily beat the most experienced opponent.

I have to disagree with this. If we're talking a 6 player free for all in mp, I'd say it's more complex not less complex than a 1 v 1 duel, because even if you can form an alliance to beat the most experienced player, it's more likely for sakes of winning that the more experienced players will work to destroy the weak players early on, meaning experienced players still win.

And if we're talking 1 player 5 AI free for all I'd say it's more complex again, in that an experienced player is more likely to be able to 'abuse' AI.

 

[Alphons Rodulfo]

I think counting the number of distinctive skill levels is a reasonable way of expressing the depth of a game. One can sort of picture it. Yes, the amount of luck/randomness in a game decreases its depth. That is also something that is easy to accept, isn't it? If a game is purely based on luck, like poker, then a beginner that understands all the rules has a fair chance of beating the world champion. But a game without randomness can still be a game without much depth. Take tic tac toe for example.

 

[MyopicCat]

Maybe the word "complex" is misleading for this concept, let's try sticking with "depth". I agree that MtG is obviously more complex than checkers in the conventional sense of the word, but perhaps checkers is still more "deep" in terms of emerging strategic content that you discover as you study the game.

I also agree with the line in the paper that suggests that the depth measure should be normalized over some time of play. Then shorter games with a large degree of luck (e.g a hand of poker) become more skill-based affairs (e.g. a poker tournament) that may have considerable depth.

Paralistalon, your point on consistency is taken, but I still feel that there is something intuitively reasonable about this depth measure if we can modify the definition appropriately. If there is a non-trivial game where the world champion A almost certainly (say 99%) beats B, B beats C, C beats D, all the way down to the beginner Z, then that game certainly has "more to it" in some sense than a game that only has 3 of these levels.

The problem for Civ is that it has long play times and a lot of randomness, so it becomes hard to distinguish the actual depth of the game. But I think it's there!

 

[Biz_]

the 2010 version of civ 5 was far deeper than the version that exists today

the tradeoffs involved much more strategic decision-making and there was a much wider range of viable ways to grow your empire

since release, firaxis has systematically removed depth+complexity from the game while adding miscellaneous mechanics that add breadth and make the game less balanced

ultimately, civ 5 is broken because the challenge is about exploiting an idiotic AI and essentially finding weaknesses/patterns in the code. the decision-making at this point has very little to do with the game mechanics / game design

the game actually isn't that bad from a design standpoint. it's just that the depth only exists in the theoretical realm of human vs human (which is basically non-existent because of the broken multiplayer implementation)

 

[Horizons]

Civ 5 appears complex at first but you soon learn that most choices are sup-optimal and you have to play the same style every time in order to win.

 

[4N4C0ND4]

Magic the Gathering 3

I just have hard time believing that.
MTG has some much possibilities. I have played for years and there is still loads of example where neither player was sure about how some effects could interacts.
I have to look out of this was calculated.

Interesting thread btw.

 

[Memoryjar]

Maybe because, the random factor is important, instead chess or go.

 

[Callonia]

Not sure on how to rate civ5.

Random factor is Map generation, Difficulty setting.
Then heuristics that AI uses to smack your face around with.

Most of the times they will just waffle around and fall on their faces.

But once in a blue moon, they will do great.

This is one of these times.
http://forums.civfanatics.com/showthread.php?t=523229

 

[civ54lyfe]

trying instead to compare apples apples, take Civ5 vs its peer EU4. in this case imo EU4 has quite a bit more complex gameplay because it's a game kind of obsessed with procedural diplomacy and rule sets. but Civ5 actually has more depth? i guess i say that b/c Civ5 has a lot more different aspects like real technology/religion/SP

 

[joncnunn]

I just have hard time believing that.
MTG has some much possibilities. I have played for years and there is still loads of example where neither player was sure about how some effects could interacts.
I have to look out of this was calculated.

Interesting thread btw.

Lots of interaction rules is actually easier to teach a computer than for a human to remember them all.

I'm not sure what exact number Civ V would be assigned, but it's safe to assume it would be higher than Go.

 

[Funky]

Though I'm dubious in regard to the numbers in the table, I find it rather appealing to define complexity as the range between novices and masters. I consider myself an experienced chess player and trainer, and when I play against my students it is virtually impossible for me to lose. An international master however would beat me every time. And in return he would be chanceless against a top grand master. And even top grand masters have plenty of room for improvement. One might add that a regular game of chess allows for far more possibilities than the number of atoms in the universe. Needless to say, chess is far more complex than any computer game ever invented, if we go by the definition of complexity suggested in the OP.

 

[joncnunn]

Though I'm dubious in regard to the numbers in the table, I find it rather appealing to define complexity as the range between novices and masters. I consider myself an experienced chess player and trainer, and when I play against my students it is virtually impossible for me to lose. An international master however would beat me every time. And in return he would be chanceless against a top grand master. And even top grand masters have plenty of room for improvement. One might add that a regular game of chess allows for far more possibilities than the number of atoms in the universe. Needless to say, chess is far more complex than any computer game ever invented, if we go by the definition of complexity suggested in the OP.

It's actually Go that's been quoted as having more possibilties than the number of atoms in the obsevable universe:


"In combinatorial game theory terms, Go is a zero-sum, perfect-information, partisan, deterministic strategy game, putting it in the same class as chess, checkers (draughts) and Reversi (Othello); however it differs from these in its game play. Although the rules are simple, the practical strategy is extremely complex.
The game emphasizes the importance of balance on multiple levels and has internal tensions. To secure an area of the board, it is good to play moves close together; however, to cover the largest area, one needs to spread out, perhaps leaving weaknesses that can be exploited. Playing too low (close to the edge) secures insufficient territory and influence, yet playing too high (far from the edge) allows the opponent to invade.
It has been claimed that Go is the most complex game in the world due to its vast number of variations in individual games.[101] Its large board and lack of restrictions allow great scope in strategy and expression of players' individuality. Decisions in one part of the board may be influenced by an apparently unrelated situation in a distant part of the board. Plays made early in the game can shape the nature of conflict a hundred moves later.
The game complexity of Go is such that describing even elementary strategy fills many introductory books. In fact, numerical estimates show that the number of possible games of Go far exceeds the number of atoms in the observable universe.[nb 15]"

Also note that none of the Go AIs are at the "professional level", while some Chess AIs are at "Grandmaster level"

 

[Funky]

It's actually Go that's been quoted as having more possibilties than the number of atoms in the obsevable universe

I don't know much about Go, other than that it is also a tremendously complex game. But the atom comparison applies to chess as well. There are 1079 atoms in the universe, while a chess game of average length has 10100 to 10120 possibilities.