4921

No need to wonder, we have Steam statistics for that. Civ5 has been continually in the Top 10 most played games since release, with 10k-20k active players at any given time (peak today says 26.5k). I’m going to assume that’s mostly single-player since that’s the most popular mode for Civ, and multiplayer is still so unreliable anyway.

Now if we look at the global Steam achievements for Civ5, we get the following percentages of players who won at least once on any given level:

25.7% on Chieftain
16.6% on Settler
14.8% on Warlord
12.9% on Prince

4.6% on King
2.2% on Emperor
1.1% on Immortal
0.9% on Deity

So a couple of observations:

  1. Up to 3/4 of players never finished even a single game. That sounds terrible but is actually par for the course – Valve found the same situation with Half-Life 2, as I recall, and it’s a reason why games keep getting shorter and easier.

  2. Strangely, Settlers appears below Chieftain; my guess is that’s because Settlers is the tutorial level which many player didn’t bother finishing before moving on to the first “real” level.

  3. So with that out of the way: Victories drop by half from Chieftain to Prince, and then by two thirds from Prince to King! Less than a fifth of all players who beat Chieftain managed to beat King.

  4. From then on the progression gets shallower again, until there’s almost no difference between Immortal and Deity – but we’re already down to 1% of the player base now.

Conclusion: The challenge level is in fact perfectly adequate for most players, with the slight cheating on King already putting victory out of their reach. The small minority that does walk over the AI on King then proceeds to beat Civ5 on any level, right up to Deity.

(That list mirror my own experience, by the way: I can win most games on Prince but lose most games on King because I’m getting overwhelmed by the AI’s production bonuses.)

4922

Losing because the computer is cheating at a crazy amount isn’t that fun though.

I lost on higher difficulties because I couldn’t penetrate the AI’s sea of troops on his continent and ran out of time. It was pretty far from fun. I lost all the time on the fair setting in Civ4 and didn’t mind cause it meant I got out played or did something wrong. When I lose because the computer can make 20 times as many units as I can I don’t have the same feeling.

4923

I’m enjoying the game a lot after this latest patch. I beat it at Prince once (Science victory) with minimal conflict, and now I’m in the middle of a conflict-heavy game as Siam. I don’t know if it’s the rules changes or wishful thinking, but the tactical AI does seem better than before. It’s done a pretty good job of screening ranged units, picking on my weaker units, and retreating in order to save its own. Honestly, if there’s a better tactical AI out there in some other turn- and hex-based game, I’d like to see it. Given what it has to juggle, imo Civ5 is doing a fine job at the moment.

4924

I agree, and theres another factor making it even less likely players will finish Civ. With something like Half Life theres a story that (hopefully) gathers momentum and so people feel more driven to finish it.

Civ games are the opposite. The game gets drawn out at the end when you know you’ve won, but have to manage a big empire and wrap the game up. I’ve declared myself the winner in many games of Civ 4, Civ 2 and SMAC, without officially winning. (I never played enough of the other Civ games for it to be an issue).

The problems even worse in Total War. I’ve never finished a Total War game since Shogun, and I love that series. Heres the stats for the Empire campaign, which has famously bad AI.

Hard 0.7%
Medium 7.7%
Easy 6.3%

Tony

4925

I’m not sure you can draw that conclusion from those stats because it doesn’t tell you how many people attempted each of the difficulties nor how many tries it took for them to reach that achievement. For example, we don’t know how many people played on Deity or how often. Perhaps 0.9% of the total players played on Deity and every one of them beat it or it could be 50% of the players attempted it and only 0.9% won. Perhaps it took every one of those players one try or perhaps it took them 20 tries each.

Unfortunately, the stats don’t help in coming to any conclusion about the challenge level of the game.

4926

How does any of that matter? I’m basing my conclusion only on the games the players won, not on the games the players attempted to win.

4927

Another possible issue: many Civ V players are new to Civ- and don’t understand the game as well as vets. Also, some players despite AI cheats, and won’t play above a certain level due to them. It could be that the cheating on King is so blatant it makes people bored/frustrated with that level.

4928

Of course it matters. It doesn’t tell you how many attempts or how many people played on that level. It tells you the total percentage of all players that played and won on that level. If 100% of the people that played on Deity, making up 0.9% of the total players, won on Deity in one attempt, would you still say that the challenge level is perfectly adequate? If so, why?

Too, saying “Less than a fifth of all players who beat Chieftain managed to beat King.” is another poor conclusion. For one, perhaps all those players that played on Chieftain remained on Chieftain (or Warlord, or Prince) and never even attempted King. Perhaps those that played on King only ever played on King and never played on any lower difficulty level. You’re making an assumption that I don’t agree with: that everyone started out on a low difficulty and then continued to move up the ladder as they beat the previous difficulty, seeking the point at which they started losing. That’s not an assumption you can make based on your stats. You can say, however, that the number of players that managed to beat King were 1/5 of the number that managed to beat Chieftain. Whether they were the same people, no one knows. You already noticed this with your point about Settler difficulty victories compared to the higher difficulties.

4929

There would be a cost-effectiveness argument- why spend resources on 1% of the playerbase.

4930

Over my gaming life I’ve given up on more games because they were tediously easy, rather than because they were punishingly hard. (But mostly I just give up because I’m bored with the game design.)

Making a game shorter makes me more likely to finish it, making the game easier does not.

If the player feels awesome/clever/skilled playing your game they are more likely to enjoy it. But thats not the same thing as easy. Then again, Farmville is very popular, so maybe most people don’t care…

Tony

4931

Sure, but those are ultimately still players who don’t find the game too easy…

Also, some players despite AI cheats, and won’t play above a certain level due to them. It could be that the cheating on King is so blatant it makes people bored/frustrated with that level.

The cheating is actually not blatant on King, but regardless – wins on Prince are still only half of the wins on Chieftain. If many people refused to play on King or above merely because they dislike cheating then you’d see little difference between Chieftain and Prince, or even most victories on Prince. But you don’t.

We’re talking about an adequate challenge level for most players, so what 0.9% of players think is really not relevant. Someone always thinks any given game is too easy (or too hard).

You’re making an assumption that I don’t agree with: that everyone started out on a low difficulty and then continued to move up the ladder as they beat the previous difficulty, seeking the point at which they started losing.

Of course I assume that’s what most players do. Trying various difficulty levels until you find your comfort zone is the usual and sensible behavior, I don’t see what’s contentious about that.

However, I don’t even need that assumption. Observe that win percentages are monotonically decreasing as the difficulty rises, from Chieftain up to Deity. If all levels were equally easy for everyone, and all players picked a level at random, then you’d see equal win percentages for all levels which is not the case. So even if you assume that players pick levels at random, the distribution still shows that Prince is pretty challenging for most players.

You can say, however, that the number of players that managed to beat King were 1/5 of the number that managed to beat Chieftain. Whether they were the same people, no one knows. You already noticed this with your point about Settler difficulty victories compared to the higher difficulties.

And I also noted that Settler is special because it’s the tutorial difficulty level. The other levels are not tied to tutorials. Even so, Settler is still between Warlord and Chieftain so we’re not talking about some gigantic deviation here.

4932

FWIW, I detest the concept of AI cheating and would much prefer a stronger AI to make up the difference in its stead. The challenge level seems adequate for most as far as I can tell, I just wish the way they got to it was different because it would simply make the gameplay better.

That said, I’m off to play as the Polynesians.

4933

I’ve played the Civ series since the original, and I’ve played the various incarnations of the series a lot (it’s one of the few that never gets uninstalled, barring 5), but I can probably count on my two hands how many games I’ve actually finished. The mop-up phase of Civ just isn’t entertaining, I’d rather start a new game once I know I’ve won (or lost).

The higher the cheating threshold, the longer an more drawn out that (read: tedious) it is.

4934

Actually, I don’t think it’s unplayable, or even too easy. It’s easier than Civ IV, I think, but the game is still very challenging on higher difficulty settings. And to be honest, I don’t even mind when a game is difficult because the AI gets lots of extra bonuses that I don’t get. That’s just not something that bothers me. If you have to make a game asymmetrical in order to provide a good challenge, then so be it.

But the tactical AI does bother me (or did–I haven’t played in the past few months, or tried the game with the new patch) because it sort of pushes all of the AI’s shortcomings into the spotlight and constantly reminds you that they are there. A lot of the fun of this style of wargame-y combat is trying to predict what the enemy is likely to do–generally by figuring out what I would do in that situation–and then devising counter-strategies. Civ V fails to scratch that itch, because it often feels like the enemy is selecting moves at random. Resource and production boosts can help compensate for the inability of an AI opponent to execute sound long-term strategies, and they do it behind the scenes, so to speak. The tactical foibles in Civ V all take place right in front of you, and are a lot harder to ignore.

As I said in my last post, I really like all of the other changes and new systems in Civ V. I’ve just had trouble getting past the tactical AI issues. Maybe I should give it another go, though, now that this new patch is out.

4935

That’s a strange thing to say, because your conclusion is stating that the difficulty level is adequate for most players. By stating that, you’re indicating you know how the players are thinking, that they are choosing to play and win on a particular level because of the difficulty. What they are thinking is absolutely key to your conclusion and that’s why I have a problem with your conclusion based on those stats. You don’t know that players are choosing to up the difficulty to a point of reaching a particular level of challenge.

Of course I assume that’s what most players do. Trying various difficulty levels until you find your comfort zone is the usual and sensible behavior, I don’t see what’s contentious about that.

And I disagree with your assumption. I play on Prince, not because it offers a particular level of challenge but because I enjoy playing on Prince. I haven’t even tried a higher difficulty level to see how it would turn out so I can’t even make a determination on whether it would be too difficult for me. If there are a significant number of people that just stop at the difficulty level that they have fun playing, instead of seeing whether they are capable of winning on a higher difficulty, then that would skew your stats to the lower difficulties and make your conclusion incorrect. In other words, it’s not the difficulty level that is adequate, but the “fun factor” that is adequate. Perhaps people simply enjoy the casual building of a civilization and don’t have any interest in pushing the limits of strategy.

If all levels were equally easy for everyone, and all players picked a level at random, then you’d see equal win percentages for all levels which is not the case. So even if you assume that players pick levels at random, the distribution still shows that Prince is pretty challenging for most players.

I disagree. People understand what the difficulties are supposed to mean and so wouldn’t pick them at random even if they were all equally easy. Generally, unless you are someone seeking a challenge, you are going to start at the lower difficulties. If you enjoy playing at those difficulty levels, whether it’s because of the difficulty itself or not, you aren’t necessarily going to increase the level. People choose the lower difficulties because they are told the higher difficulties are more difficult. Nothing surprising about that. But it doesn’t mean that the difficulty is adequate. Maybe they don’t care about that aspect of the game; I don’t find Prince level particularly challenging at all, for instance. Perhaps the majority of people who started Civ5 on a difficulty level of Prince or lower never even attempted a higher difficulty, yet if they had maybe all of them would have won on King. Maybe none of them would have won on King. We don’t know because they didn’t try (or did they?).

My point is that your conclusion doesn’t follow from the stats you’ve used. The theory could be valid, but it requires the right data to back it up and that data requires you to understand the motivation for choosing a particular level.

4936

If you enjoy a difficulty level then the challenge on that level is adequate for you, whatever that means given your particular style of play. So your distinction is meaningless. Of course many people play Civ as a SimEmpire, and that’s why they’re willing to settle for a lower level of challenge than in purer wargames. That doesn’t contradict anything I said.

4937

This is exactly what I do. What an interesting way to put it.

4938

I understand the critique of Christoph’s reasoning, but I think one key point remains. A lot of people have been saying the problem with Civ V is that it’s just way too easy because of the poor tactical AI. That the only way to get a challenge out of the game is to crank up the difficulty until the AI gets so many cheats that the game isn’t fun or realistic anymore, and it’s therefore not really worth playing.

What the Steam stats show, though, is that for the overwhelming majority of players, that is not true. Eighty-seven percent of the people who play the game have never beaten it on Prince (the “fair” level at which neither player nor AI gets any bonuses). Maybe some of them are TBS geniuses who started on King and never looked back (although the fractional numbers of people winning at King and above tends to say that’s not the case). But even if that’s true for some number of people, when compared to an install base probably in the high tens of thousands, it’s statistically insignificant. The vast majority of people probably started on Prince or lower, and of those people, 87% of them have not won the game at Prince yet. It’s hard to square that with the view that the tactical AI is so bad that the game isn’t challenging. The stats track with my own experience, which is that the game is challenging on middle difficulty levels even if you’re a longtime gamer and longtime player of the Civ series.

Of course, that leaves the argument Ben is making, which is that the game is challenging but the AI is still so bad that the game’s not worth playing. That hasn’t been true from my perspective at all and I have a little bit of a difficult time understanding the argument (it moves at random and yet is challenging enough to beat a human? Eighty-seven percent of the time even when it has no production bonuses? Huh?), but at some point I guess we just have to say “OK, it works for me but not for you, it takes diff’rnt strokes to move the world” and move on.

(KevinC makes the argument that maybe the stats are off because people don’t play their winning games through to the end. I don’t buy that at all. Nearly everyone plays through to the end at least the first time, to see the ending sequences if nothing else. I mean, are we really to believe that the AI is super-easy, but the stats show 87% AI wins because tens of thousands of people are quitting out every win, even their very first one, without ever having laid eyes on the victory screen?)

4939

Me too. I quit almost all my games at this point after it’s clear I have started rolling over everyone else.

4940

While Equisilus’ critiques of the Steam stats are technically correct, to rush to dismiss them entirely would be to miss the forest for the trees. The extremely low percentages (in absolute terms) do compellingly hint that interpreting the numbers to indicate the vast majority (roughly 100 to 1) of those who own Civ V (at least to the point of installing it and allowing it to communicate with Steam) have not in fact beat the higher AI settings of Civ V.

Anecdotally, reading these forums, one could come to a very different conclusion, which is understandable as a self-selection bias of the membership in this board.