The Final Analysis?

[JohnnyW]

I was going to respond to your post with more direct quotes, but I think just one will suffice:

Just like pineapple in a pizza, somebody likes long building times.. err I mean pineapple in a pizza and someone hates pineapple in pizza. It’s a matter of taste.

Some people like cutting themselves; it's their taste.

But hey, if someone made a game where I had to cut myself to win I wouldn't play it.

In regards to your analogy of the chess computer being hard for one and easier for another: well yeah, people have different skill levels and I thought that was obvious without being mentioned or taking an analogy to describe. The thing you aren't realizing is that this game is incredibly easier than previous titles because the AI cannot handle the 1upt combat even remotely close to how well it could handle SoD combat. Even on Deity it acts moronically but it's difficult simply due to pure numbers advantage. You even admit the AI sucks later in your post:

At the end of the day I must say (and I really have said this a million times) that civ5 has its flaws, it is not perfect and it needs a better AI, not only for the combat but also for the rest of the game. Critics is good, but fighting over things that are simply "just a matter of taste" is stupid.

 

[DLL]

Archers cannot shoot over mountaisn nor hills in civ5. Well they can shoot over hills with ''Indirect fire'' -promotion. That is if that promotion is still available for archers after the patch, at least i havent seen it for a while..

Yeah, it was with that promotion. Like I joked though, I was willing to just smile at the idea of it all for the sake of the tactical elements they were trying to add to the game.

Sometimes, the tactical elements were fun. It was one of the things that made me look forward to Civ 5 prior to its release, too.

 

[Atwork]

chess is 1upt

First, I love to play chess. Second, I mostly support the change to 1UPT.

the best chess AI would most likely beat most of the worlds people in a chess game, and as a result of that we could say that the AI did indeed handle 1upt very nicely.

This is true. But, it would not be true to say that C5's combat AI would beat most of the worlds best CIV players -- even those playing at the highest difficulty settings. The combat AI in C5 can't even slightly compete with a good CIV player even at the highest settings. And that is the difference between your chess program analogy and the current state of combat AI in C5.

give that same "fairly easy for an average adult who just happens to know chess rules" -chess game to a six year old kid who just happens to know chess rules. It happens that the kid loosed the game, in fact the kid wouldnt win it no matter how hard he tried.

By analogy, the point that your argument seems to be making, is that the AI in C5 is not actually flawed, but rather just appears weak to those who know the rules. On the other hand, some kids out there find the AI impossibly hard. In short, you’re saying it’s all subjective. Well I’m sorry, but I don’t buy that argument for one second.

What separates a winning chess player from a losing chess player isn’t that one player “happens to know chess rules” and the other player doesn’t. In vague terms, what separates players is the ability to more fully appreciate the variables of the game-situation by analyzing the range of choices available, by appreciating the chain of consequences, by factoring the advantages/disadvantages to a given choice and associated consequences, and then opting for the choice that will achieve the most advantageous outcome accounting for both long-term and short-term goals.

A well designed AI should account for the situational variables successfully, even if the extent to which it aggressively pursues advantageous outcomes, and to what ends, might be nerfed at lower difficulties (for the 6 year-olds and causal fans) but seriously buffed at the highest levels.

However, this isn’t the situation encountered in C5. It is not the case that the combat AI plays better at higher difficulties. And that’s the problem. The problem is inadequate coding. The coding that should exist to make combat more challenging at higher difficulties is missing, is in error, or is in conflict with other code. At this point, it's a one-size-fits-all mess of a combat AI.

To counter the mess, the patch gave us ICS and a bunch of other collateral fixes that doesn't address the fundamental problem: that the AI doesn't utilize the full range of battlefield options, poorly executes the options it does pursue, and is completely dependent on the ridiculously huge advantages it gets in terms of happiness etc. just to compete at all.

When the AI fails to bring its Great Generals in to action and instead leaves them in distant cities, this is not the AI failing to handle 1UPT according to my subjective standard of 'not good enough.' This is a failure of coding.

When the AI fails to maneuver injured units out of harms way, and it sits turn after turn taking bombardment damage until it finally dies -- again, this is a failure of coding.

When the AI embarks units into ocean hexes within sight of my cities and navy, and proceeds to be sunk, this again is a failure of programming.

I could go on and on with further examples.

These aren’t issues of subjective expectations. These are problems with the coding, which may have a lot to do with the fundamental design making too high of demands upon the devs.

 

[lietkynes]

When the AI fails to bring its Great Generals in to action and instead leaves them in distant cities, this is not the AI failing to handle 1UPT according to my subjective standard of 'good enough.' This is a failure of coding.

When the AI fails to maneuver injured units out of harms way, and it sits turn after turn taking bombardment damage until it finally dies -- again, this is a failure of coding.

When the AI embarks units into ocean hexes within sight of my cities and navy, and proceeds to be sunk, this again is a failure of programming.

Agreed.

I am not expecting tactical genius from Civ5 AI (just look at how poor is Civ4 AI despite 5 years of expansions and patches), but the problem here is that most of the Civ5 AI combat behaviour can only be classified as a bug.

 

[aziantuntija]

@JohnnyW, @Atwork ,you understood me wrong or you miss read something from my post (or my post was not well written wich is also very likely scenario since I don’t need English language so much). Mainly my topic was about perspective & opinion and not at all about civ5’s AI. But please dont tell me you didnt read my warning before you read my example? I writed it with BIG letters and bolded the whole warning so that I wouldn’t have to go through this. In case you still missed it, here is a copy of that warning: -WARNING- THE FOLLOWING TEXT WAS NOT MEANT TO HAVE ABSOLUTELY NOTHING TO DO WITH CIV5. IT IS JUST AN EXAMPLE AND I BEG THAT IT SHOULD BE TREATED AS SUCH. So that chess game+kid -writing worked as an example of that there can be (and almost always is) more than just one perspective to everything, even to some games difficulty level. However, it was NOT an example of how civ5's AI would be IMO, somehow flawless.


Oh and btw Atwork, the question about that first chess example, you know the "the best chess AI would most likely beat most of the worlds people in a chess game" –thing did not even mention civ5. I just answered to this "The AI can not handle 1upt." -claim. So the question was not specificly pointed of how civ5 handles 1upt, but about how can AI handle 1upt system. The claim was that it cant so i answered to that.

Btw, here is another example about the thing that I was talking about in my earlier post. This time the example is from civ game: Somebody might say that King –difficulty level is hard but someone else might say that King difficulty level is easy. Wich one is right? Answer: They are both wrong and they are both right, it is an opinion wich breeds from perspective. This is why games have different difficulty levels.

Once again: You (and possibly everyone else) must understand that with that writing about that kid playing chess game, I did NOT mean that I would NOT want to see a better AI for civ5, because in my opinion, civ5 really needs a better AI. So thats why it had a warning attached to it. My post was mainly about opinion&perspective and NOT really about civ5's AI. Even though I emphasized my view about the current -not so good- state of the civ5 AI at the end of my post but apparently, it wasn’t enough. Well I hope this post clarifies my view.


EDIT: Talking about SoD vs. 1upt, I agree that the AI is much better at pumping huge amounts of units and stacking them to a one huge stack than it is in using 1upt, i mean its much less harder for AI. But imo, this doesnt make the SoD's more fun than 1upt.

 

[Iranon]

Chess is a perfect information, 2-player, zero-sum game with no inherent randomness. None of the issues that make 1upt problematic for an AI in a civ-like game apply.

 

[Atwork]

Somebody might say that King –difficulty level is hard but someone else might say that King difficulty level is easy. Wich one is right? Answer: They are both wrong and they are both right, it is an opinion wich breeds from perspective.

Both statements may be subjectively true. However, it would not be objectively true to say that the player who found King difficult found it to be difficult because the combat AI played better at King difficulty as opposed to Prince difficulty. More likely, the player found King difficult because of other collateral advantages given to the AI. If it were just an issue of combat AI, the outcome would be the same at any level.

Chess is a perfect information, 2-player, zero-sum game with no inherent randomness. None of the issues that make 1upt problematic for an AI in a civ-like game apply.

Agreed. Chess is a bad analogy to C5 1UPT coding.

 

[aziantuntija]

Chess is a perfect information, 2-player, zero-sum game with no inherent randomness. None of the issues that make 1upt problematic for an AI in a civ-like game apply.

I must admit that I don’t know whats the case here but if you are referring to my first chess example in my answer to "The AI can not handle 1upt." –claim. Then I will once again highlight the fact that the question was about 1upt system, not civ5 combat system.


Also i must say that nothing is perfect, therefore chess cannot be a perfect information game. If it were game of perfect information, then the player would have to know what has happened in the past, what happens now and what happens in the future, that would be ´perfect´ information. Chess player can guess what happens in the future when the game advances but he cannot know that for sure.


There also is inherent randomness in chess game. Imagine a situation where the player number 1 has two (or possibly even more) moves that he can make that are, in his opinion, equally good, in this case the decision will be done randomly. Also the player number 2 might of thought that when he himself made his move before the the player number 1 got his two equal options, he might of thought that the player number 1 “will most likely” do something that he himself (the player number 2) sees right, but instead player number 1 does something completely different. But of course the better example of randomisness in a chess game is propably the first one i gave, wich was the player number 1 facing, in his opinion, two or more equally good moves. So yes there is inherent randomness in chess game, like there is in so many other games that I can quickly think of, you just have to see it.

More likely, the player found King difficult because of other collateral advantages given to the AI. If it were just an issue of combat AI, the outcome would be the same at any level.

Yes but once again: I-wasn’t-talking-about-civ5-combat-AI. This was just an example, it read it on the front of this example that this is an example "about the thing that I was talking about in my earlier post" –wich is: perspective & opinion. This example (nor any other example that ive given so far) does not even mention civ5’s combat AI.

 

[JLoZeppeli]

I must admit that I don’t know whats the case here but if you are referring to my first chess example in my answer to "The AI can not handle 1upt." –claim. Then I will once again highlight the fact that the question was about 1upt system, not civ5 combat system.


Also i must say that nothing is perfect, therefore chess cannot be a perfect information game. If it were game of perfect information, then the player would have to know what has happened in the past, what happens now and what happens in the future, that would be ´perfect´ information. Chess player can guess what happens in the future when the game advances but he cannot know that for sure.


There also is inherent randomness in chess game. Imagine a situation where the player number 1 has two (or possibly even more) moves that he can make that are, in his opinion, equally good, in this case the decision will be done randomly. Also the player number 2 might of thought that when he himself made his move before the the player number 1 got his two equal options, he might of thought that the player number 1 “will most likely” do something that he himself (the player number 2) sees right, but instead player number 1 does something completely different. But of course the better example of randomisness in a chess game is propably the first one i gave, wich was the player number 1 facing, in his opinion, two or more equally good moves. So yes there is inherent randomness in chess game, like there is in so many other games that I can quickly think of, you just have to see it.





Yes but once again: I-wasn’t-talking-about-civ5-combat-AI. This was just an example, it read it on the front of this example that this is an example "about the thing that I was talking about in my earlier post" –wich is: perspective & opinion. This example (nor any other example that ive given so far) does not even mention civ5’s combat AI.

You don't understand in the first place that 1upt of chess is based on a fixed board, fixed pieces and movements: something that an Ai can understand better, and using heuristic "less complex" (and it's quite the euphemism) than one needed for a game like Civ (and when i say needed i can say impossible to do, there are too many heuristics needed to work properly and with an elaborating time very long...)

 

[ohioastronomy]

If everything is just opinion, with no facts, then we don't have anything to base a discussion on.

We have to start with a set of things that we know - for example, that the computer struggles with tactical combat in Civ 5, or that turns in Civ 5 are slower than in prior versions even on fast computers, or that tile yields are lower in Civ 5, or that production times are longer in Civ 5. The good news about this approach is that it can yield common ground and allow us to understand how we get to our differences on items which really are matters of subjective judgment.

For example, an unskilled chess player might well not see the consequences of his actions, but a chess master could analyze the game and explain precisely when things went well (or, more likely, badly.) By comparison, it can be much more difficult to trace cause and effect if there are random numbers being rolled which dictate the outcome.

 

[Iranon]

aziantuntija... I'm trying very hard to remain civil. There are facts and opinions, and sometimes the difference is obvious.

'Perfect Information' is an unambiguous term. If you challenge that chess is a game with perfect information, you pit your opinion against the definition everyone who deals with game theory uses. Most classic board games are perfect information games, as opposed to many card games (you know your own cards, but not those of your fellow players).

You also don't seem to use the same definition of 'random' as everyone else.

You may as well define up to mean down.

 

[Kerosene31]

Why does this have to be "final"?

 

[aziantuntija]

@JLoZeppeli. You know what? You could just read couple of first sentences from my previous post and that will answer very well to your whole post.

@ohioastronomy. Yes we have plenty of facts. Here is what I replied to builer680:

So as you can propably allready see, im not going to focus on answering to all your statements on civ5, unless of course they are written as flaws in wich case, most of them would be just your opinions. Why do tey change from statements to opinions if they are listed as flaws instead of just statements? Heres why: For example this: "Production times are high because of the need to keep armies small." is just a comment. BUT, if you say it as a civ5 flaw though, then its -again- just your opinion and not a fact. So as a comment something like this: "Production times are longer in civ5 than in civ4" might be- and pretty much is a valid statement, but if you list it as a flaw, then that same comment isn’t anymore a true as a flaw, then it becomes –yes you guessed it- just an opinion. Just like pineapple in a pizza, somebody likes long building times.. err I mean pineapple in a pizza and someone hates pineapple in pizza. It’s a matter of taste.

So once again, if you say that for example: "Civilization V uses 1upt system." then you are absolutely right and that is considered as a fact. However, if you say that its bad thing that "Civilization V uses 1upt system." then its just your opinion and is not a considered as a fact. Its simple really.

@Iranon. You might say that the game of chess gives all the information to the player, except the information that what your opponent is thinking. So the information that chess gives to its players, is by no means perfect, nothing is perfect. You may guess what are your opponents intentions, but you cant know it for sure. To design a fun game you need to hide at least something and in a game of chess this hidden thing are the opponents intentions. If ALL the information is available to a player I mean literally EVERYTHING, then that would be a dull game, but then again, at least I have never heard of a game that would give a all the information to its player. Heck im not even sure its possible to do that.

I don’t seem to use the same definition of 'random' as everyone else? I suppose that goes for all of its synonymes as well. Well in this case I most likely managed to great a sentence wich sound very normal but nobody just cant fill the empty spot with any word that would make the whole sentence seem like a sensible sentence.

Heres that magical sentence: "Imagine a situation where the chess player number 1 has two (or possibly even more) moves that he can make that are, in his opinion, equally good, in this case the decision will be done *insert your word here (the word cannot be ''random'' nor "randomly" or any of their synonymes)*."

Please do not change the rest of the sentence!

According to Iranon, there should be at least one word other than ''random'' or "randomly" or any of their synonymes that will make that sentence sound sensible.

aziantuntija... I'm trying very hard to remain civil.

Good, at least you are trying. If you start to feel like it is becoming very, VERY hard for you to stay civil, then I would suggest that you would take a nap or something. Now if you will excuse me, ive got some weight lifting to do.

 

[Celevin]

@Iranon. You might say that the game of chess gives all the information to the player, except the information that what your opponent is thinking. So the information that chess gives to its players, is by no means perfect, nothing is perfect. You may guess what are your opponents intentions, but you cant know it for sure. To design a fun game you need to hide at least something and in a game of chess this hidden thing are the opponents intentions. If ALL the information is available to a player I mean literally EVERYTHING, then that would be a dull game, but then again, at least I have never heard of a game that would give a all the information to its player. Heck im not even sure its possible to do that.

http://en.wikipedia.org/wiki/Perfect_information

This is the standard definition used. You can't argue with anyone if you don't even use the standard definitions.

Heres that magical sentence: "Imagine a situation where the chess player number 1 has two (or possibly even more) moves that he can make that are, in his opinion, equally good, in this case the decision will be done *insert your word here (the word cannot be ''random'' nor "randomly" or any of their synonymes)*."

People use the same word for different things all the time.

 

[aziantuntija]

http://en.wikipedia.org/wiki/Perfect_information

This is the standard definition used. You can't argue with anyone if you don't even use the standard definitions.


People use the same word for different things all the time.

Ok I will answer to this but then I must start doing some serious weight lifting

Ok so it seems that finally, not writing in my native language has totally got me confused. I have successfully mixed the words complete (<---- I belive this is the correct word) and perfect. Well sh** happens. Still I consider the rest of my post a valid.

People use the same word for different things all the time.

I dont understand what are you trying to say.

 

[Celevin]

Ok I will answer to this but then I must start doing some serious weight lifting

Ok so it seems that finally, not writing in my native language has totally got me confused. I have successfully mixed the words complete (<---- I belive this is the correct word) and perfect. Well sh** happens. Still I consider the rest of my post a valid.

Civ5 also does not have complete information. Chess has complete information.

 

[DaviddesJ]

First, I love to play chess. Second, I mostly support the change to 1UPT. ....

These aren’t issues of subjective expectations. These are problems with the coding, which may have a lot to do with the fundamental design making too high of demands upon the devs.

If you accept that the Civ 5 1upt combat system makes demands on the developers that are too high, so that they can't deliver an AI that can put up a good fight, then why do you support that design anyway?

 

[Aesculapius]

The Final Analysis? (see question mark) Doesn't at all Have to Be Final.

First of all, thanks Brau for initiating a post that clearly pertains to an ongoing and profound point of passion for many people (372 posts in 9 days) and probably many times as many readers. While for many years I have followed Civ Fanatics' streams for their frequently valuable insight and utility, I have not posted until now. And like many contributors to these posts, I have a deep passion for Civ that begins, well, in the very beginning. But the addition of Civ 5 to the series and the substantial controversy it has wrought has led me to post.

Certainly, there are no shortages of sound analyses pertaining to Civ 5 among the many and varied contributors. But in reading Sullla's analysis I could not help but recognise a sad truth. Here is somebody who obviously has extensive Civ experience and acumen who has roundly (and for the most part seemingly quite restrainedly) articulated the shortcomings of Civ 5 not as minor foibles to be tweaked but core-penetrating faults that offer little other than a very bleak prognosis for the 5th iteration of a truly hallmark and ground-breaking series. Sad, because as much I love the series and wanted it to continue to thrive, I can't help sensing (perhaps more in my heart than in my brain) that Sullla is essentially right.

For my part, Civ 5 was extremely-highly anticipated, excitedly installed, earnestly initially explored, and soon ambivalently ignored. In the beginning, it wasn't clear why the game wasn't delivering the captivating-virtually-addictive traction of its predecessors. Perhaps I was missing some of the game mechanics or not appreciating the finer novel concepts. But either way, the game did not meet what for me is the crucial litmus test - is it worth my time playing? The answer evolved to be a steadfast and consistent 'no.'

The game appears to have many issues that could be addressed, but in his exceedingly astute review, Sullla identifies the 1UPT attribute as the principal fault of the game. I agree whole-heartedly and in discussion with a deeply-Civ-entrenched colleague, this element was such a significant departure from the series mechanics when it was announced early last year as to cause us both concern as to how well the game would play. Not very at all.

I also agree that the 1UPT was very likely well-intended. The idea of more realistically portraying manoeuvre units on the battlefield remains an excellent one. Clearly, in real warfare, missle troops (whether they be archers or self-propelled artillery) cannot effectively function without their fraternal front-line combatants (or static defensive element) protecting them. Thus, the tactical consideration of positioning and defending ranged forces was an appeal in the design of Civ 5 and I believe remains an excellent element to be developed.

So, to Sullla or anybody else possessing the where-with-all (and the time) to effectively translate Civ ideas to reality (regrettably, it's certainly not me!), I offer the following thought:

1UPT might still be salvageable by incorporating it in a sub-map layer of the game. Perhaps there could be two map layers (as in the Total War series): one strategic, same as is now but without 1UPT, therefore no absurd traffic jamming for military or civil units in a hex-limited environment; the other tactical, which blows up the strategic hex to a multitude of 1UPT tactical hexes (30x30, 50x50, 100x100, whatever works), with likewise magnification of salient terrain features as the tactical battle is often won or lost by defensive elements such as rivers, swamps, ridges, and hills.

Being attacked or initiating one? Well, no problem, switch to the tactical map, deploy your forces, and may the battle commence. And now as deployment and protection of various troop types will be critical, each player will certainly want to make make sure he didn't bring a knife to a gunfight.

At the end of the day, Civ 5 for me has neither lived up to my expectations nor my perception of the series' reputation (and obviously in the end it's all about perception). Can it overcome its shortcomings? Regrettably, I'm inclined to think not, and certainly not without some profound restructuring. Perhaps an abandonment of 1UPT and necessary follow-on feature modification, perhaps a dual-map structure, who knows? But one thing remains clear: the interest in this game, its series, and the genre robustly persists. As long as players and contributors like Sullla and Dale and the innumerable others who have yet to post are out there, the Civ series will continue.

My deep thanks to you all.

 

[Atwork]

If you accept that the Civ 5 1upt combat system makes demands on the developers that are too high, so that they can't deliver an AI that can put up a good fight, then why do you support that design anyway?

I'm admittedly not an expert when it comes to computer programming. So, I accept that the inadequate combat coding MAY have a lot to with the fundamental 1UPT design. However, I personally don't have enough information to draw any absolute conclusions. If you know better with a high degree of certainty, then please educate.

On the other hand, I do know that I prefer the possible tactical challenge that 1UTP could (should) offer that SOD didn't. But, I don't think 1UPT is worth the trade-offs in terms of how the rest of the game had to be designed around the 1UPT concept (tile yields, production times, etc.). Thus, I would prefer a non-absolute 1UPT system. I've previously posted that I think ranged units should be able to stack with a limited number of melee/gunpowder units that could act as escorts. I've also come to support other peoples' desire to allow for unit bundling for the limited purpose of troop transport (similar rules to Panzer General apparently).

Maybe these moderate changes would allow other aspects of the game to be improved. In the final analysis, if the devs can't get this 1UPT system to work well in combat and they can't strike a better balance elsewhere, then SOD would be preferrable to me -- or at least some variation in which the number of units per stack is tied to tech, resource, and tile yield.

 

[Gre_Magus]

Sulla's arguments are pretty insightful. I hadn't thought about the connection between 1-unit-per-hex and slower late-game production times, but it makes sense. I'm actually still enjoying Civ 5 (since I haven't had time to play it to death like some other people not expert enough to just rip apart it's AI), but I think he has some very valid points.