Authorial influence and fiat.

The goal of this blog has been, and will be, to explore believability. This means that it looks at ways that a fictional world can be made more "real", in the sense that it's tangible and logical to the audience. However, there is one major interfering factor when it comes to the issue of authorship. No matter how logically or illogically a story is developed, there is one key rule that absolutely cannot be forgotten: what the writer says happens is what happens. It doesn't matter how little sense it makes, or how much: if there is logic, it is because the writer has arranged the story in a way that things make sense. If there isn't, it's because they didn't bother to arrange the story in such a way. There is no actual system of logic in play: it's all set up by the writer, regardless of the end result. Of course, the audience is free to make their own conclusions about the morals and lessons taught by the events, but the events themselves are orchestrated entirely by the author.

All of the updates on this blog - past and future - have the goal of making it so that the author has more tools at their disposal to create the illusion of reality. This is done to make that illusion more tangible and palpable so that the audience can maintain their suspension of disbelief. The less things that draw their attention to the fictitious nature of the enterprise, the less likely that this major issue (the fact that it is fiction, and hence it's all made up anyways) will be breached. But this is an optional objective - there are many writers more concerned with telling "the story they want" than worrying about consistency or logic. Can you stop them? Of course not. It's their story. If Karen Traviss (pictured left) wants to say that all Mandalorians are super-awesome warriors who are way better than every Jedi, then technically she's right - insomuch as she's making it all up anyways, and one made-up thing is just as true as another.

A writer is essentially on their own in terms of logic. There is no system governing what they can and cannot write, except in a meta-sense (censors, legal issues, etcetera). If they want something to happen, it is going to happen. This brings us to the issue of fiat (Latin for "let it be done"), a general term referring to things like deus ex machina or anything where something happens because "the author wills it". This is most obvious when there is a logical leap or gap and it's not bridged by anything other than the author saying it happened. However, if we examine the initial statement, then every single thing that happens in a story is fiat. The difference is separating the fiat that's logical from the fiat that's not - what's believable and what's implausible.

In an earlier post I compared the idea of enemies being plausible threats and/or beings in their own right to enemies being static non-entities who exist to be trampled by the hero. As you might guess, both things in this case come down to fiat: will the author portray it plausibly, or will they just go the easy route and make all the enemies stand still to get blown up? In the end, it comes down to how the author wants to show it. You can talk about plausibility and believability all you want, but the actual events are only influenced by how the author chooses to depict things.

No example shows this more clearly, in my mind, than Berserk's hundred-man battle. This is combat that is simultaneously difficult and trivial. Both of the protagonists have to block enemy strikes and counter in weak spots, rather than just blasting through them with no effort. There's a real effort to show that they're outnumbered, they're getting tired, and they're on their last legs. But at the same time, the enemies never bring their superior numbers to bear in a logical way, the protagonists get away with a lot of stuff that they shouldn't (at one point Guts waves his cape and this is enough to somehow knock crossbow bolts away), and ultimately the moral of the story is "The good guys kill a lot of guys who attack in a way that allows the good guys to kill them". The consistency of the situation is there for atmosphere, rather than to shape events.

This example is important to me because there's an attempt to have a logical, grounded system for the whole affair, but ultimately it comes down to "we need the good guys to survive and beat the odds, and the bad guys to all die messily". The plausibility of the scene is a speed bump on the road to Guts winning. There's a similar feel around the series "Gundam: 08th MS Team". The three main pilots are meant to be fairly normal, but none of them die even when the odds are against them. This is a good clip to illustrate what I mean: Shiro is a normal human who is forced to resort to a desperate ambush against his opponents - something that wouldn't happen in any of the other series - but at the same time he dodges gunfire from a mech, and then later he shoots down some tiny anti-personnel mines. He's superhuman, but not; he's really whatever the story needs him to be in order to survive. The "realism" of the show is part of the illusion, but in actuality he's going to win no matter what it takes.

This brings us to another key issue: the capabilities of a character will always be whatever they need to be. I touched on this briefly when discussing PvP/PvE, but here's the bottom line: measuring strength is always going to be ridiculous because the strength of a character is derived solely from the writer. If I write a story where I say "this character is a normal human but he can eat the sun", should you be impressed with this character? Would you be drawn to learn more about him and how his totally insane logical inconsistencies can exist? Well, there's no logic behind it. He can eat the sun because I, the author, said he can. That's it.

Now let's compare this to some actual shows and movies. I've touched on the later Gundam series before, but they really are a pretty good example because they rely on the audience being impressed by "this mech is powerful". What is there to be in awe of? The weapon is powerful because the designer made it powerful. Without context or drama, what is that "power" appealing to? Even with that one clip, they won't leave well enough alone and power it up even more. What does this accomplish? There's no tactile or experience-based viewpoint to actually approach the gun from, so it might as well be a sentence reading "this gun can blow up everything". Its power exists in a vacuum. Giant mecha can do everything from destroy an army to throw a galaxy like a shuriken, and what does it mean? It means that the writer said "and then the giant mecha threw the galaxy like a shuriken" - no more, no less.

This problem can almost be made worse if there's an attempt to connect to some sort of realism (although arguably all things try to appeal to realism at some level - more about this in a future update). For example, in the game "Halo: ODST", the player takes the role of an Orbital Drop Ship Trooper (essentially a futuristic paratrooper). The difference between normal Halo games and ODST is meant to be justified by the fact that in most Halo games you play as a shield-equipped super-soldier, whereas in ODST you are a normal guy. However, the grand list of things that changes about the game is this: you can no longer dual wield. Your health regenerates, your movement is light and quick, your guns are bolted to your frame rather than swinging wildly around, and you can punch through tanks. The aspect of "being normal" has absolutely no connection to the gameplay because it all exists as author fiat, and "author fiat" only wanted the player to be an ODST so they could expand on an element they thought was cool. The ODST, as a soldier, is exactly as strong as the game needs him to be. And yet there is a trope praising such characters on the basis of "it's more impressive because they don't have any powers", forgetting that the only power they need is the power of "the author wants me to win". Nothing else matters except that.

However, there is one method of storytelling that actually has the potential to overcome this hurdle: luck-based or skill-based gaming with a solid rule system to ensure fairness. This primarily takes the form of RPGs, but actual gameplay in videogames may also count. Games in general have a few distinct traits that can free them of the quagmire of authorial directorship:

1: The Game System Provides The Rules. One of the issues with authorship is the issue of consistency: an attack might not be enough to harm an enemy in one scene, but it is in another. The consistency of the story is up to the author to maintain, and if they want to ignore it there's nothing that can be done. Games, on the other hand, establish a solid system, and with the exception of cutscenes generally try to maintain that system at all times. If a character is low level, they're going to have these capabilities. If they level up, their capabilities will improve. In this way, the playing field can be kept consistent, if not necessarily level. A character will consistently be the same strength unless they take an action in-universe to change it.

An issue related to this is "real logic" versus "game logic" - that is to say, a game meant to represent some sort of "reality" (humans are humans, physics are physics) inaccurately representing that reality. One could also point to "higher-level" stuff, which suffers from the same "higher numbers = better than" mentality seen in DBZ or any other high-powered anime series. The difference here, to me, is that there's a concrete method of improvement and a concrete level of power. If you want to be high level, you have to earn your way there. If you're a low-level person and you attack a high-level person, you're probably going to lose. There's no arbitrary "well he's tougher than me in every way but I'm a good guy so I win" stuff unless, say, the player rolls a critical hit. This leads into the second point.

2: Events Change Based On Luck or Skill. Earlier with regards to Berserk, I mentioned that, in essence, the way combat was portrayed was unimportant because the end result had to be "Guts wins". In a game, it can all come down to either a roll of the dice (for most role-playing games) or the player's capabilities (for more direct action games). The gaming comic Knights of the Dinner Table actually has a lot of characters dying and campaigns failing as the result of bad rolls and bad decisions, which is perfectly logical and happens all the time in real D&D campaigns. The success or failure of the characters actually has a lasting impression, and this makes it far more tense because the story might actually change. You don't just get a do-over, you could totally screw up and all your characters could die and that's it, the story is over for them.

It's important to note, however, that the way an adventure is tailored and challenges are shown affects this. Oftentimes, encounters are "balanced" to provide an optimal combat threat. I don't support this method; I prefer a system where enemies are whatever type it would make sense for that area, and it's up to the players to analyze their chances. If they're out in the woods and they hear a hill giant approaching, it should be up to them to do something about it - either take it head-on (if they're confident), set a trap, run away, or try to hide. If the encounter's balanced, there's no real reason to do anything other than "fight normally"; if it's unbalanced, the players have to think on their feet and react to the monster as an event, rather than an obstacle. A hard encounter should be overcome with guile and cunning, instead of just not happening until the party's ready for it.

3: The Players Make Decisions And Direct The Course Of Their Characters' Actions. This is the point that's least universal - there are plenty of railroading DMs who want to put their players through "my story", and far more videogames that do the same thing. Still, in most RPGs the onus of activity lies with the players, often to the detriment of the DM who has to keep up with their pace. This freedom is shared with a lot of wide-open sandbox games, although video games are inherently limited by things like resources and design.

The importance of this freedom is that it makes the story much more malleable: it's the result of a logical (well, not always logical, but at least player-decided) train of thought leading to a rule-moderated conclusion based on planning, strategy, and luck. The more that the DM or developers say "no you can't do that", the less it's going to feel like the player has any influence on what's happening. In this case, the DM and developers represent "fiat": you can't do that because I said you can't, deal with it. To be a truly interactive, developing story, these restrictions must be minimized.

RPGs are naturally equipped to deal with this - unlike a computer, a human DM can adapt new material "on the fly" (or at least relatively quickly) based on consulting rules and stats in the book. If the players want to hire mercenaries, the books will probably have rules and tables for that. If the players want to build a castle, the books will probably have rules and tables for that. In fact, this aspect is why I personally prefer AD&D to newer editions - the lighter, less-intensive combat system means there's more time devoted to establishing background data and materials for the DM to call upon.

---

I appreciate the stories generated by pen-and-paper RPGs and sandbox games because they're naturally occurring; a system was established, and the players let events happen. To me, it's much more of a "big deal" to hear about an expedition that nearly failed but managed to survive at the last second, or an X-COM recruit who managed to get off that lucky shot. These stories are more real to me because they operate under similar rules to reality, most importantly "success isn't guaranteed". In contrast, authored stories are, well, authored: what the author says happens, happens. If they're a good author, I'll be drawn in enough that it's not predictable and I can try to figure out what's going to happen, but in essence it comes down to what the author wanted to happen.

There's a term in fandom that I think is fairly important for this: in a series where many characters die, it is the author who "kills their characters". Even if a show or book or movie depicts war in an accurate sense, the characters who died are, in fact, killed by the author, because it is the author's decision whether they live or die. In the last, climactic book of the Harry Potter series, many characters died - but those characters lived and died because J.K. Rowling wrote them living or dying. She could have spared any one of them, and every one that died did so because she desired it to happen. Gundam director Yoshiyuki Tomino (pictured left) has actually been criticized for making series where characters are killed off frequently, despite the fact that most of these series are war-oriented. Structured rules take the decision out of the "author's" hands by making it up to the dice. War is fatal, and it may end up being random whether you live or die - but there's nothing "random" about an author, because the author's role is to govern the story, and they have absolute control over every aspect of it.

This is the base concept of what I'm trying to get at: no matter what, it's all going to come down to the author's decision. An author can kill characters off to try to achieve the "randomness" of war, and logically their choice will be justified, but essentially the characters who live and die are still chosen by the author. This is why internal consistency is so important: because the decision an author makes should be backed up by logic for the audience to become invested in them. Without internal consistency, each decision sits alone in a void, and the audience is essentially listening to a dictated list of events without any connection to why any of it is happening. The more rules the author abides by, the more "fair" their decisions will seem.

I constantly bring up Band of Brothers and Saving Private Ryan on this blog for various reasons, and to me there is one major difference between them: Band of Brothers is based on true events, and Saving Private Ryan is not. When a character dies in Band of Brothers, it is because it actually happened based on a logical system of events in real life. When a character dies in Saving Private Ryan, it is because the script called for them to die. Even if the death is done well and feels natural, it's going to be underscored by the fact that the death happened as part of the script, not "naturally". Perhaps, then, scripts should be built off of role-playing sessions? It's worth a gamble.

Accessible realism.

In creating "Red Orchestra", the WW2-themed mod for Unreal Tournament, the game's designers referred to the game's theme as one of "accessible realism". What does this mean? Many highly realistic, detail-oriented games are relegated to the status of "simulation", and are enjoyed primarily by enthusiasts of that field, rather than the general public. As such, they divert resources from things like graphics, sound, and other aspects that would appeal to the general public in favor of focusing on making the experience detailed and realistic. The goal of Red Orchestra, on the other hand, was to present a realistic form of gameplay that would also appeal to gamers as a larger group based on its merits and style - to balance these two issues.

In my analysis of Company of Heroes, I compared the game to "Men of War", a more realistic, but less accessible, real-time strategy game. Company of Heroes is by far the more well-known of the two, because it's a combination of realistic elements and a high-quality presentation. Men of War, on the other hand, has poorer animations, graphics, and sound elements, which makes it less exciting for the player. In short, a person interested in hardcore realism will likely prefer Men of War, but the average player will respond more positively to Company of Heroes, which is reflected in its higher sales and critical reception.

The Silent Hunter series is about as "real" as a submarine game can get, and as such it had a very loyal following of submarine and history enthusiasts. However, by Silent Hunter 4, the developers wanted to branch out their audience. They did this by improving the graphics (adding a lot of background details and "eye-candy") as well as making the difficulty system more dynamic and flexible, going from "really unrealistic" to "incredibly realistic". The idea was to attract gamers who wouldn't normally consider playing by eliminating some of the issues with the difficulty curve and mechanics that kept them away, while also making the game nicer to look at.

Still, one of the potential problems for this approach is that Silent Hunter's basic gameplay (a wide-open sandbox where ships and convoys patrol back and forth, leaving the player to decide their own goals) is enjoyable primarily if you like submarines. There's no advancement, at least not in the direct RPG sense. It's a game that's played if you enjoy playing games about submarines. There's nothing else to it. This is an aspect that has been criticized about the games, in the same way that Full Spectrum Warrior was criticized because you "don't learn any new maneuvers".

In fact, Full Spectrum Warrior is a good game to bring into this discussion. There are essentially two versions of the original FSW: the more detailed, but less "prettied up" Army version, and the more consumer-friendly Release verson. The Army version is actually included in the Release version and can be accessed through a code - it's got less visual quality, but more detailed gameplay that includes civilians and improved building mechanics. This can be identified as the major tradeoff and the intended balance of "accessible realism": detail versus presentation.

Detail refers to the amount of systems in play in a given situation. For example, in a flight simulator, a "highly detailed" game would include things like wind resistance, proper aerodynamics, and stalling. This would make the game more realistic and complex, but also give it a higher learning curve. In contrast, a more "arcade-like" simulator would start with simple concepts (pitch, yaw, loops, and so on) and then work with that. Another example would be the physics system in Dwarf Fortress. There are many things constantly being influenced, which makes it complex and realistic, and yet intimidating as well - since you can't just sort of dig haphazardly through the rock, the gameplay becomes more complex.

The dynamic difficulty mentioned with regards to Silent Hunter is by no means unique. There are many simulators, including IL-2 and Wings of Prey, that have a dynamic difficulty system. The concept is this: the more things you have to pay attention to, the harder the game will be. If you are playing Silent Hunter with infinite oxygen, you can focus on major tasks without worrying about oxygen levels. If you are playing Wings of Prey with arcade rules, you can shoot down enemy planes without worrying about running out of ammo or fuel. The basic concept may be the same, but there's less things for the player to worry about.

For example, with regards to Full Spectrum Warrior, it's easy to pick up on the basic gameplay: keep your teams behind cover, use smoke or suppression to allow your teams to move forward. The unrealistic parts of the game, such as the simplified ammo system and the fact that "behind cover = invincibility", allow the player to focus more solidly on those prior elements. If they were accurately included, then the player would have to juggle all those different elements. It's the difference between a tutorial mission and a trial by fire - you have to space out concepts and allow the player time to learn them at a leisurely pace, or else they're going to feel overwhelmed.

Men of War tends to provoke this sort of response: it gives you a lot of tools to use, but its presentation makes it so that rather than those tools being valid options for different situations, they're making things aggressively complicated by throwing a billion things at the player and saying "okay, sort them out while you're being shot at". In contrast, Company of Heroes only gives you a few things to worry about (mostly your larger system of resources), meaning that you don't have to interrupt your orchestration of an entire platoon to have one guy run out and get some more ammo for his gun. The question a developer must ask themselves is: what is the focus of my game? Is this low-level stuff worth exploring, or should I assume that the player would rather worry about higher-level tactics?

Having a complex game isn't bad, but it's definitely intimidating. In real life, the things depicted by most simulators are actually jobs - you have to go to school or train for weeks to learn how to do it, and the reason you would do that is because it is a profession. Learning how to play a "realistic" game can, therefore, be a huge commitment, and is anathema to the idea of "cheap thrills". However, the more things that are unrealistic, the less sense a game makes when examined under the light of its connection to real life. The more realistic a game, the more authentic an experience it provides, and the more it is capable of teaching the player in mechanical terms. Detail is, therefore, the logical arm of the game's believability.

Presentation refers to the quality of the game in visual and audio terms. Good design should allow the player to locate and use resources, and they should also make the player feel immersed in the game. Good graphics and sound design can mean the difference between a basic mechanical understanding ("I am being shot at") and an automatic emotional response ("Get down!"). It is the job of presentation not just to "look good" but also to convince the player that the world they are seeing is real, rather than simply being markers on a board or cutouts at a range.

Better graphics allow for better identification on the part of the player. For example, in Company of Heroes the brightly colored stripes on vehicles aren't realistic or even sensible, but their role is to help the player locate them and identify them. The same thing is also why there's a mini-map, and why a highlighted squad has circles under all its members. One of Men of War's problems, on the other hand, is that it's difficult to keep track of your troops because of the graphics design and quality. Red Orchestra tries to make the realistic difficulties of identification into a gameplay element. However, if its graphics weren't as good, things like draw distance and poor models would adversely affect the player's ability to identify targets at "realistic" distances.

Even in highly detailed games, a good production value can help with immersion. Steel Battalion is an XBOX mech game famous for having a custom controller that represents literally the entire dashboard, down to the windshield wipers. In such a game, something as simple as the startup sequence is part of the immersion. In Silent Hunter, the graphics are part of convincing the player that they're really in a submarine - everything from the crewmen to the detailing on the sub itself. Without convincing graphics, a "simulator" is just a training tool: it teaches the system, but it doesn't really put you there. Presentation is the sensory arm of the game's believability.

The conflict between detail and presentation is one of resources, rather than ideology. This manifests in two ways: hardware resources and developer resources. Detail and presentation vie for position in these two fields, and it is these fields that causes them to be at odds in the first place. Dwarf Fortress can help us illustrate these issues.

In Dwarf Fortress, the complex mechanisms covering every single aspect of physics in the entire virtual world are capable of slowing down powerful modern computers even though the game itself is done with ASCII characters. You could not replicate the experience with better graphics, at least not without improving computer processing technology. You could simplify the game to make it more feasible, but since a lot of the game's systems (not to mention charm) derives from these complex operations, it wouldn't be the same game. This is a hardware resources issue.

The one-man dev team behind Dwarf Fortress updates the game fairly frequently, but each update consists of more complexities, rather than, say, something to make the game more accessible. For example, he'll happily make lava physics more realistic when the game is still basically a mess in terms of controls. His work and coding makes the game more complex and realistic, but it does not make it better in terms of accessibility. This is a developer resources issue: time spent doing one thing can't be spent doing another, and budget spent on one aspect can't be spent on another. If the developers want to appeal to a mass audience, they're going to prioritize graphics. If they want to appeal to enthusiasts, they're going to prioritize detail.

Both detail and presentation are important to create a sense of immersion. The former is necessary for logical reasons, and the latter is necessary for sensory ones. However, developers often choose between one or the other, resulting in highly-accurate, highly-detailed games that don't draw in the user, or very attractive and visually stimulating games that don't have any basis in reality. Movies like Saving Private Ryan work because they actually have both bases covered. If Saving Private Ryan wasn't so gritty and accurate (except for the entire movie after Omaha Beach) then it wouldn't have been praised as highly as it was. The improved presentation was used to reinforce the details by making them more tangible and visible to the audience.

Therefore, I would say that instead of "niche games" only working towards total accuracy, they should try a little more to make it visually immersive, as has been done by Silent Hunter, Full Spectrum Warrior, and so on. If a game is totally accurate but has poor presentation, it will only be accepted by people who like the subject enough to ignore the presentation. If a game is totally inaccurate but has great presentation, it will be picked up by people who like the graphics enough to ignore the unrealistic elements. Presumably, the best option is to find a balance somewhere in the middle.

Analysis: Company of Heroes

I've referenced Company of Heroes a lot before, primarily due to the fact that it strikes a strange balance between "depicting reality" and "being fake and/or gamey". However, there hasn't really been a specific point-by-point analysis of it, and how its smaller systems lead to a larger game, as of yet. In short, I'll be trying to relate it to the points and topics I've established thus far.

Company of Heroes is, in terms of realism, leaps and bounds ahead of most other Real-Time Strategy games on the market. In comparison to some of the more "hardcore" turn-based strategies, it pales in its accuracy. Nonetheless, it remains one of the most accessibly realistic games: it uses the basic concepts of real warfare in a way that doesn't drive off potential players. Its combination of easily understood gameplay, eye-catching graphics, and general balance of play-flow is more easily understood than, say, "Men of War", which is more realistic but also far less playable for the majority of gamers.

Nonetheless, it's important to understand what, exactly, is "unrealistic" about it, and how some of the game's touches encourage the kind of background information that we've discussed. Most of the "realistic" things are done based on copying real things, and most of the "fake" things are based on the necessity of gameplay. With that in mind, let's move on to the topics.

Supply

 As mentioned far earlier in this blog, the way Company of Heroes handles supply is abstracted on a huge number of levels. In fact, it's probably the most abstracted thing about the game. In short, players capture "points" on the map, which gives them "manpower", "munitions", and "fuel", as well as securing territory for them in which they can build emplacements, upgrade squads, and so on. The main issue here is that this works on a very conceptual level - if, for example, it was a hex-based map and a unit controlled the whole hex, and supply from the main base to the outer bases was implied. The further out the camera is zoomed, and the longer a stretch of time is meant to represent, the more acceptable it is when time is compressed for purposes of resupply or movement. If you're looking at a hex, you don't know exactly what's going on in that hex. If a turn represents one day, then it's assumed the orders you give take a full day to carry out, which leaves plenty of time for rest and resupply. Therefore, a territory being "under control" could indicate that there are supplies being shipped back and forth - the player just doesn't see them.

However, Company of Heroes is down at the lowest level - squads and platoons. You, the player, are seeing everything that's going on; there's no resupply trucks going back and forth when you're not looking. When your troops get those guns, they teleport into their hands. If you're in friendly territory, you're in range of the Magic Appearing Guns upgrade. The resources are another weird thing similar to this - you can pay 25 munitions to throw a grenade, or pay 125 to launch an artillery barrage. The munitions are almost a universal resource that is converted on the spot to whatever ammunition type you need. It's simplified to a point where it's hard to connect it to reality, because there's no source for any of this. Your troops are standing out in the middle of a field and BAM suddenly they have sub-machine guns.

The reason for the "munitions issue", at least, is the difficulty in giving each squad their own inventory. In fact, "Men of War" tries to do this very thing (each soldier has their own inventory), and it doesn't work particularly well due to the fact that the game is real-time. In some cases it costs munitions to use normal ammo, which is otherwise infinite; for example, the Browning Automatic Rifle's "suppression" ability and the scoped Lee-Enfield's "marksman" ability both cost munition to use, even though it's really just "fire more shots" and "fire more accurately" respectively. Munitions make sense within the rules of the game, but they're difficult to explain in-universe. Yes, it makes sense for units to try to capture an ammo dump - but ammo in real life would work in a different way than "supplying more munitions points", which brings it back to being hard to explain.

In short, few things in Company of Heroes' supply system have a "source". Some special units come from off-map, which seems reasonable, but they're paid for with the "manpower" resource, which seems weird again (why do you need to build a barracks for riflemen, but not for rangers?). There's an infinite supply of soldiers as long as your resources build up enough to pay for them, and you won't get any reinforcements unless you've got enough of those resources. Essentially it's all an excuse for capturing territory, but it just feels kind of silly when you think about it. In real life, there's things like "securing territory", "establishing lines", and "resupplying", but it's done in a more sensible way that reflects the long-term nature of the war.

Company of Heroes wants to have those things, but the constraints of its "video game" nature means that it does it in a way similar to a lot of other RTS games: simplify the economy to the point of being unrecognizable. The question is, what else can it do? How do you arbitrate the inclusion of elements that rely on the whole war being included - because how else are you going to set up the amount of supplies available? Still, there's so many liberties taken that the supply situation might as well be from an entirely different game - there's nothing "WW2" about it.

Soldiers

CoH's squad system means that, unlike a lot of games where every individual soldier must be maneuevered into place, there's a lot more room for "limited intelligence" on the parts of the units. Soldiers will dive into cover, move around during firefights, and generally give indications that they're actually people rather than machines. They've got enough quirks in their behavior that they feel more natural than a perfectly robotic soldier who just moves from point to point without responses.

In addition, the context-based content of the game, such as voice acting that changes based on whether or not a unit is in combat, helps to make them feel more natural. The "retreat" option helps as well, because it allows the player to try to preserve their soldiers rather than having to watch them die. These touches aren't really "necessary" - at least in the gameplay sense - but they do help a lot when it comes to making the soldiers feel real. Even reinforcing a squad, which is kind of a "gamey" thing (since it falls under the unrealistic supply system detailed above), makes some sense: you can't just conjure reinforcements out of thin air, they have to come from your base or a transport.

However, there is one major exception to this: "Garrison units" (not just soldiers in buildings, but soldiers specifically attached to buildings) do not behave in any of the ways specified above. If you build an MG nest, the soldier manning that gun will fire at enemies in his forward arc forever. He won't leave the nest, he won't change the direction he's firing in. He can't be sniped, either - he lives and dies with his MG nest. Also, when engineers build an MG nest, he just sort of appears with it. He doesn't behave like a human being; it could just as easily be a robotic sentry gun and it would behave the same. The garrison units aren't totally devoid of characterization, but they don't operate under the rules a human being would.

Vehicles are less problematic, but still somewhat so. A vehicle's crew won't leave their vehicle, even if it's clearly about to be destroyed. If a vehicle's immobilized, they have no choice but to die even if they're a veteran crew who've got invaluable combat experience. Even weapon teams will refuse to abandon their weapons - if a two-man weapon gets down to one man, the last man will literally drop dead rather than retreat. Again, it's hard to induce sympathy when your guys won't just run for their lives.

The final issue with soldiers involves healing and medical supplies. Allied forces and Panzer Elite heal in proximity to a medical station or forward base, while Wehrmacht squads can be issued medical kits to heal themselves for a small cost in munitions. Additionally, medical evacuation stations can be built by most factions, rescuing "downed" (but not dead) troops and reforming them into a new squad. The issue is that there's no way to actually get a soldier out of the fight. If they're wounded lightly, they either stay wounded or heal at a medical station. If they're wounded heavily, they either bleed out or get reformed into a new squad. It's hard to care for one's troops when there's no way to actually get them out of the fight. The fact that you can try to keep units alive is endearing, but the fact that you can't actually get them out makes it more like some weird gladiatorial arena.

Uniforms

In CoH, each faction has a standard infantry type, used for most things that require some generic infantryman. This includes acting as infantry, garrisoning emplacements, and manning vehicle weapons like pintle-mounted MGs. In addition, if a medic station is set up by that faction, they will eventually convert injured soldiers into a squad of that kind of infantry. For the American and Commonwealth factions, this unit is their basic rifleman. For the Wehrmacht, it's the grenadier. For the Panzer Elite, their two main infantry types (panzergrenadiers and assault grenadiers) alternate between the role; the former (with grey jackets) generally man vehicles, while the latter (with camo jackets) man emplacement-type weapons.

The deviations from this "basic model" vary by faction and unit type. The standard US rifleman serves as the basic unit for the American faction, and many other infantry types are a variation on the rifleman. For example, within the rifleman squad there is always one soldier with an M1 carbine and a radio pack, rather than an M1 Garand. Mortar teams use the rifleman's uniform, but with a different helmet, an ammunition backpack, and slightly different gear. Rangers have the riflemen jacket, but it's worn underneath a combat vest, and the other parts of the uniform have some modifications as well (snipers use the same combat vest). Artillery crewmen just use the normal rifleman uniform with no modification. Most vehicle gunners use the standard rifleman appearance (even when it would make more sense to have a tank crewman gunning), but the one different unit is the Jeep's gunner, who has belts of ammunition draped over him.

The fact that there are so many units that are similar make units that are different sort of visually jarring. For example, on the German side, the fictional "Volksgrenadiers" (conscripts who are meant to represent the "Volkssturm" militia formed near the end of the war, but are named after a fairly elite unit instead) wear a shade of grey that's dissimilar from the grey worn by full-fledged Grenadiers. MG teams for the Americans and Wehrmacht also wear a different jacket from the normal troops, perhaps to make them more identifiable; the American MG team wears green jackets, while the German MG team wears blue (in comparison to khaki and grey jackets normally).

One aspect of CoH that can affect this is the fact that when a weapon's crew is killed, the weapon is generally dropped and can be picked up by anyone. Therefore, it is possible to make, say, an entire army of paratroopers, including MG teams, mortars, and anti-tank guns. When a unit is reinforced at the HQ, they are reinforced with the same type of unit (i.e. if you have paratroopers pick up a heavy weapon and one of them is killed, the replacement will also be a paratrooper). This can be noted most with both American and British airborne units, since both have the option to resupply in the field (the Americans through air-dropped heavy weapons and equipment, the British through gliders set up to bring in heavy weapons teams as well as commando squads). In short, judging by the established scale, units in CoH range from a Type 3 (same uniform, minor differences) to a Type 6 (separate branches, but consistent uniforms within each branch). I specifically liked the ability to make a whole group of commandos or airborne because it made them feel more like a united unit (Type 2/3) and less like "some paratroopers mixed in with the normal soldiers" (Type 6).

In terms of gear, soldiers carry all the things that they would be expected to, and most have big backpacks or at least a belt with ammo pouches and so on. However, this brings up some problems. For example, the "munition" thing mentioned above becomes strange when a unit has grenades visible on its uniform, and it still costs munitions to throw them. The Panzer Elite tankbusters and assault grenadiers even have panzerfaust disposable rocket launchers, but no way to actually use them. While the backpacks look neat and make things feel authentic, they're disconnected from the actual gameplay because of the strange supply system.

Tactics

There are a lot of things that make CoH's tactics seem closer to real life, such as the presence of cover and suppression, and the use of barricades and obstacles to create "kill zones" for crew-served weapons like mortars and machine guns. However, in a lot of ways the system is similar, the specifics aren't. I can't blame CoH's developers for this - they're creating a game, and in creating a game things have to be adjusted to fit the environment. For example, all of CoH's maps are so small that if tank guns had realistic ranges, most of them could fire clear across the map. One thing that's internally inconsistent, though, is the nature of artillery. It's possible either to use off-map artillery (which costs munitions) or build/set up an artillery piece on the field (which costs manpower and sometimes fuel). What's weird about this is that the former represents actual accurate artillery ranges (i.e. they can fire anywhere on the map, as long as you can see it), and the latter can't actually fire across the map, even if it's the same kind of artillery you're calling from off-map. It's not just unrealistic, it's actively inconsistent. It also calls the nature of "munitions" into question, since you can fire the artillery you built at no munitions cost - so clearly the munitions aren't paying for the shells.

The healing issue is not just a "soldier" issue, but also a tactical issue. In real life, if someone is hit, they're most likely going to go down. In CoH, death can be quick and instantaneous if it's a mortar round or a sniper, but if it's just enemies shooting at them it's going to take a few hits for them to die even if they're standing in the open. Notice that I said hits, not shots. A unit can be hit and just sort of ignore it until they heal up in the ways mentioned before. If someone's hit enough to be critically injured, they writhe around on the ground waiting for a medic, but there's no impediment prior to that. This is pretty clearly for reasons of simplification (it's easier to keep track of squads and soldiers like that) but it's also kind of weird if you're thinking about things in-universe.

It also changes the tactical layout, since firefights aren't just about hitting the enemy - they're about hitting the enemy a lot. This means that automatic weapons are superior in every way to bolt-action or semiautomatic ones, even if they're just spraying wildly at close range. Yes, there's some range differences, but the role that riflemen play in maintaining a perimeter is minimal compared to the importance of suppressive machine gun fire. Again, this isn't bad, it's just a way that these comparatively small changes (HP instead of real injury effects, smaller ranges) changes the larger battlefield.

One final thing that bothers me has to do with the use of equipment. The majority of abilities in CoH are equipment-related, whether that equipment is for building things (shovels and sandbags) or destroying things (grenades, explosive charges). The fact that, for example, only British troops can dig slit trenches is kind of weird because it doesn't make rational sense. It's not just that sandbags and materials come from nowhere when an engineer builds an emplacement, it's also that it's inconsistent. British infantry can summon up sandbags to build an MG nest or a mortar pit, but only sappers can actually make a wall of sandbags - like, just put them out in the open to take cover behind, instead of using it to make a structure. It would have felt a little more realistic to establish that certain units carry certain tools, and those certain tools allow them to dig trenches or cut barbed wire or blow up bridges. If a unit obtains those tools, they can do those things, but if they don't have them then they can't.

Conclusion

I want to state that I like Company of Heroes a lot, and it's a pretty good game all things considered. However, it's also the kind of game where the little differences add up and become noticeable, especially because of how "accurate" things can be otherwise. For example, the unforgiving nature of the game, with soldiers being destroyed by artillery, mortars, and airstrikes, is undone by the unrealistic healing and repair systems. The openness of the game, allowing for flanking and terrain deformation, is undermined by the fact that you can never really "prepare" a unit for that - you just have to hope you have enough munitions to throw a satchel charge or something. The human elements of the game, such as the movement and banter of the soldiers, are undermined by inconsistent behavior, as only infantry really behave like "people".

Company of Heroes, in short, is realistic, but not quite enough. It opens the door to a lot of new things that would be interesting if developed more, but trying to use those things for any larger meaning, morals, or personal attachment is undone by the "game-like" qualities of the whole thing. I've mentioned Men of War a few times and in general Men of War is a much more accurate game. However, it's also important to note that Company of Heroes' production values, including its graphics and voice acting, make it a lot more tangible than Men of War's less immersive fare. Company of Heroes is a fun, enjoyable game, with a complex tactical layout, but it's not really a great learning tool, and it doesn't recreate World War 2 as well as it could.

Comparisons of power: "PvP" versus "PvE".

In the gaming world, and especially MMOs, there are two kinds of combat: Player versus Player (PvP), and Player versus Environment (PvE). The former refers to player combat, and the second refers to combat against an AI-controlled entity. PvP combat is generally more tricky, because the other player possesses just as much adaptability, cunning, and strategy as you do. In contrast, the AI may be effective, but it is not usually as adaptive. In addition, in a PvP match, if someone wins, someone loses. In PvE, the computer doesn't care - if the player wins, then the player wins.

This is the foundation for a key difference in perspective, not just for games but for all media. In general, an enemy soldier, especially a faceless one, is perceived as a non-person (as noted before). They may as well be an automaton - they don't get to care about whether or not they live or die, they just shoot their gun ineffectively and carry out their orders. However, in real life, combat of any sort is a match between two entities, both of whom want to live.

In a "PvP sense", both sides are active, adapting participants in the combat process. They may have varying skill levels, but the fact that they are both motivated by a drive to survive, and a drive to win, changes the dynamic. In "PvE", the enemy exists as an obstacle, rather than another human. One example that can be used to illustrate this is Company of Heroes. When fighting against higher-level AI, the computer receives more resources than the player. However, the computer is not an "adaptive" opponent, and in many cases this is meant to be its weakness. If the player builds, say, an entrenched position on a bridge, the computer will throw away its resources charging blindly into that position if it cannot find a way around. It will not, however, use things like artillery or other long-range bombardments. In fact, there are some mods that change this, and they make the computer a lot more plausibly threatening. The normal enemy AI is overwhelming in numbers, but simplistic in tactics.

In short, fictional universes like to make characters seem powerful by throwing hundreds of weak enemies at them and letting those enemies get mowed down with hardly a struggle. However, this often stretches the limits of disbelief, usually under the premise that, firstly, a character fighting people who are objectively weaker than themselves is not particularly heroic, and secondly, the enemies in question don't do the kinds of things that would make sense for an individual attempting to stay alive. Would you watch a series where the heroes blindly rush to their deaths with no attempt to use tactics or strategy? Why should it be any different for the enemies?

Let's compare two examples. In the original "Mobile Suit Gundam", the main character was an ace pilot, but he was still relatively balanced against his foes. He piloted a new, high-performance prototype mech against the enemy's mass-produced mechs, but not only were those mechs still fairly capable, his own mech wasn't that much better. It was comparable to, say, a Panther tank versus a Sherman tank - yes, the Panther has better armor and weapons, but the Sherman could still theoretically take it out despite its disadvantages.

In addition, the skill levels of the pilots were much closer, reflecting the fact that while the protagonist was gifted, his enemies were professional soldiers with years of experience. Defeating even a few enemies was a big deal, and fights lasted for quite a while. There are scenes where the loss of a few "Zakus" (the standard enemy mech) is treated about the same as losing a few tanks or planes in real life.

Mobile Suit Gundam even had a reversal in its spinoff, the 08th MS Team, where the protagonists (a squad of regular pilots, rather than aces) are fighting as a unit - and are outclassed by an enemy ace. In short, the enemies in these shows are depicted as being, on the whole, just as capable as the "good guys", and while the protagonists may be skilled, they're not invulnerable. This is a solid PvP situation: both sides are reasonably human, but the protagonist's equipment and reflexes just happen to be better in some cases.

The later series were not so balanced, however. The escalating need for constantly higher levels of power to maintain the same level of viewer attention meant that in series like Gundam SEED, Wing, and 00, entire armies are being destroyed by lone mechs. Enemy soldiers are little more than speedbumps. The show becomes a contest to see who can destroy the most enemies with the least visible effort. It had turned into a fully PvE affair: enemies are weaklings who still charge in huge numbers to be destroyed by a high-powered laser or swords or whatever the protagonist happens to be using. The style of these shows is best noted by the fact that it has a Dynasty Warriors spinoff adaptation.

Where does the difference come from? The original Mobile Suit Gundam (and its related timeline, the Universal Century) was clearly meant as a war story, perhaps (like Star Wars) as a sort of WW2 in space. In real life, the Allies were not "guaranteed" to be superior; in fact, most German units had better training and equipment, and would win in a 1 on 1 fight. The same thing is reflected in Mobile Suit Gundam; the titular mech is better than normal mechs, but there's only one and it's not invulnerable by any means. However, the later series' threw away the "war" concept and went with the concepts that were more popular, namely "giant mechs fighting with laser swords". However, the fact that the victories were nowhere near as hard-won ought to have made them less important; judging by the fanbase that the Gundam franchise still has, this is not the case.

This establishes an important concept: the idea that "more kills = more skills". The plain fact is that defeating a bunch of unskilled enemies isn't that huge of a deal, at least not when compared with defeating enemies who actually possess skills. In real life, a fighter ace is any pilot who has achieved at least five air-to-air kills during their career. In a game, this would seem almost preposterously low; by the end of the Ace Combat games, for example, your kill count can and will number in the thousands. This discrepancy is created because of the numerous advantages that the Ace Combat player has - respawning missiles, weaker enemies, and maneuvering that ignores many of the harsher elements of real life aeronautics. Still, all the player sees is the number, in the same way that a Dynasty Warriors player can kill hundreds of foes with ease because those foes barely fight back.

The discrepancy of "ace kills" occurs in real life, as well - compare the kill counts of the Luftwaffe on the Ostfront with aces around the rest of the world. The simple number does not reflect that many of the German pilots' kills were done against Soviet aircraft. In the words of Erich Hartmann, the highest-scoring ace of the war, "In the early days, incredible as it may seem, there was no reason for you to feel fear if the Russian fighter was behind you. With their hand-painted 'gunsights' they couldn't pull the lead properly or hit you." The ground war can have similar gaps: killing a conscript is not the same as killing a well-trained, properly-equipped soldier. The focus on "kill counts" means that the context and difficulty of the situation is not included.

Let's step back to fiction. In fiction, there has to be continued action (rising kill counts), but the protagonists have to be "safe" (because death means the loss of a character). The obvious solution is to make it so that the enemies are pitifully weak, but highly numerous, so that the main characters can kill a bunch of them but not die themselves. This is true of everything from zombies to enemy soldiers in FPS games: there has to be a threat, but the player has to be able to deal with it, which naturally means that things are unfair. Upping the numbers increases the tension, but in most cases the universe simply adjusts to keep it "fair". The basic idea is "defeating a large number of enemies indicates prowess", but that idea doesn't reflect the skill or odds involved.

There are plenty of examples of battles that would have been balanced actively being made unbalanced because of the expectation that the protagonists should win. For example, in "Halo", the Covenant forces are superior to the human forces in every way - they have shields, superior weaponry, etcetera. In the beginning of the first Halo game, a lone human ship is cornered by a fleet of enemies, and while it is eventually brought down, it is in a very controlled manner after defeating several enemy ships. Prior to that, there are (canonically) many examples of human ships being destroyed even with superior numbers, but the need for the ship in Halo to be superior means that it has to defeat enemies before being taken down.

The above image comes from Wing Commander 1, which is an odd example because it's possible to lose wingmen, but even if you lost all of your allies you'd still be maintaining a huge kill/death ratio. What's notable is that dead pilots are removed from the board, so by default nobody dies (except in cutscenes or if they're flying with the player), but they do get more and more kills. Hence, the war against humanoid cats in space-fighters is less "a war" and more "pest removal" - we're not taking any casualties, but it's going to take a long time for us to kill all of them. Even more oddly is the existence of enemy aces, who have themselves killed hundreds of human pilots. This seems a tad implausible given the scoreboard above; presumably, the pilots on the Tiger's Claw are the cream of the crop, and those aces were killing all the "regular" pilots.

"Weak enemy" situations also don't really give the enemies a lot of credit. Let's say there's a scenario where you're holding a bridge or gap, like in "300". It would be really great if, say, the enemies just charged in and allowed themselves to be machine-gunned by the thousand. Then you could say you'd just defeated a huge army with almost no effort. But, realistically, why would they continue to attack in a way that makes no sense? Let's say that they have artillery, or air support, or any other form of long-range weapon where you can't effectively attack back. Suddenly the dynamic has changed. Yes, they'll still be attacking, and you still have a chokepoint to hold, but it's clearly a lot more difficult. But why aren't you killing just as many enemies? Surely if you were just as powerful, you'd do just as well. A "PvE" enemy wouldn't use that support, because their job is to make the player feel good. Therefore, they would keep running into the guns no matter what.

In the Battle of Wizna during the early stages of World War 2, the Polish forces were outnumbered 59:1 by the German army. Not only that, but the Germans also had the aforementioned support - artillery, tanks, bombers, and so on. 720 men held on for 3 days against a force of 42,200, and managed to inflict 900 casualties in the process. Wouldn't that have been more impressive if that was a higher ratio - like, say, if they'd managed to defeat all 42,200 because they didn't have that support? The law of "more kills = more skills" would indicate so, and if there was an anime series about the Battle of Wizna that's probably what would happen.

However, the actual battle is notable specifically because of that aversion - even with these overwhelming odds, the Polish forces hung on and managed to deal a comparatively sizable blow. Their enemies were highly trained and well equipped, and the Polish forces still managed to hold them off for at least a little while. This is the kind of context that's omitted by the simplistic "more kills" model. There's no reason to care if a character defeats a hundred enemies who don't actually do anything sensible to try to stop him, but if he's defeating a hundred enemies who are actually trying to survive, then it's far more impressive. If the enemies are just there to be beaten up with no effort (like the Battle Droids in the Star Wars prequels), then there's no tension.

There's a second aspect to the "PvP"/"PvE" issue, as well. The perception of "PvE", in general, makes it so that the enemies aren't people. This is why they don't, for example, use tactics or common sense - because they are cannon fodder, not human beings. In general this is going to connect to a lot of cause-and-effect stuff that gets ignored by most anime. For example, if a young hotshot protagonist with minimal training goes up against an experienced veteran who's not a protagonist, the protagonist is going to win - because the show is about them, and the show must go on. The results of the battle are unrelated to the actual skill levels of the combatants. It's unfair not for in-universe reasons, but for meta-reasons. The "redshirt" or "mook" dies because they are a "redshirt" or "mook", not because they trained less, or were less competent. They die because it is their role on the show to die.

What this does is undermine the sense that every person on a show is, themselves, a person. Many of the other concepts explored previously are used to try to make a character seem like a rounded, sensible human being. In this case, the need for an enemy to be weak and harmless (while still ostensibly a threat) means that they are no longer a human being. At the very least, they are not playing by the same rules as their enemies - they can barely hit them, while in return they can die quite easily. Is this fair? If the camera were following these characters instead of the protagonist, and they were just as weak, we'd expect them to come up with some way to overcome the gap in ability with intelligence or cunning. The natural assumption is that a protagonist will automatically overcome the enemy no matter what. This means that (a) the enemy is no longer subject to the same rules as the protagonist, and (b) there's really no tension - we know the protagonists will win.

"When I was a boy I was taught to think of Englishmen as the five-meal people. They ate more frequently than the poor but sober Italians. Jews are rich and help each other through a secret web of mutual assistance. However, the followers of Ur-Fascism must also be convinced that they can overwhelm the enemies. Thus, by a continuous shifting of rhetorical focus, the enemies are at the same time too strong and too weak." - Umberto Eco

In this quote lies the crux of "enemy difficulty". Because the protagonist must win, the enemy must be weak - weak enough to be guaranteed to lose. Because the protagonist must take pride in their victory, the enemy must be strong - strong enough to put up a good show before succumbing to the inevitable. If you reversed the circumstances, the "protagonist", being invulnerable but demanding sport, would seem like the villain. Instead, the protagonist is lauded under the assumption that the enemy posed some potential threat, even though they clearly did not. All it does in fiction is create a "threat" that the hero can feel good about defeating, but it doesn't actually make it legitimate - it just justifies their perceived superiority. The efforts of the "enemy" party never comes into it; they are an obstacle to be overcome, not a foe to be fought. This is the essence of PvP versus PvE: In PvP, there is a winner and a loser - for every kill, a death. In PvE, there is a winner and a non-entity.

So, to sum up:

1: If you want to make enemies seem like "people", then you need to have them play by the same rules as the protagonist. If there's a difference in power, justify it, but if you have that difference then you need to play it straight the whole time. Also, remember that for every character who kills twenty or thirty guys without getting hit, there are twenty or thirty guys who were killed without even hitting the guy who got them.

2: Don't be so quick to take pride in easy victories. Killing mooks is like beating the AI on super-easy: it's not a huge accomplishment, don't pretend that it is. The harder the victory, the more meaningful it should be - it shouldn't just be "whoever can curbstomp the enemy the most will be the best". A character being skilled or powerful should be seen as a result of their fighting adversity, not just slicing their way through enemies who can't even hurt them. Would you consider it impressive for a high-level RPG character to go around beating up low-level enemies? The odds of a fight should be based on more than "number of enemies involved".

Facilities, cities, and the "lived-in" feeling.

Man-made facilities are a common part of games, whether it's a crowded city, an isolated research station, a spaceship, or even a dungeon. However, in games these places are often levels first and places second. As with many other issues, this comes down to sensibility: games need to guide you through an area to accomplish an objective, and real places have real roles and real uses they need to fulfill. The problem arises because of the differing needs of a real facility and a fake one.

For purposes of this analysis, a facility can be seen as a combination of human needs, purpose-based needs, and structural/security measures. "Human needs" refers to anything that falls within the necessity of the people that work there. This includes anything from sleeping areas to food preparation areas to rest and relaxation. Any part of the facility that exists for the sake of employees, rather than for the building's purpose itself, would fall under this category. "Purpose-based needs" would basically be anything the facility is designed for. This includes areas like  A factory's production areas would fall under this, as would a lab's research areas. "Structural security" is basically everything else: something that exists to protect the assets of the building, like security cameras or checkpoints. Of course, this is a very simplistic analysis, but its purpose will become clear.

In real life, the prioritization of these features would probably be "purpose-based", "human needs", and then "structural security" with the least amount of space. This varies from building to building, but in general most facilities are designed to be accessible, because otherwise people are going to get lost, have long walks to reach places, and so on. An environment must be open enough for people to get where they need to go and to do what they're supposed to do. Also, buildings must conform to codes and standards meant to protect individuals by preventing environments from being unsafe or hazardous, because in real life a person getting hurt is a pretty big deal.

In games, security measures are usually bumped up to the highest priority. This is because games are about conflict, and nothing causes conflict like a thing that gets in your way. In general, a character will interact with security measures (have to find some way around), but purpose-based and human-need areas are irrelevant except in terms of the game items they have. Dungeons in fantasy are basically the perfect evolution of this, since they're underground areas often designed with no purpose other than "have a bunch of traps and stuff", usually justified by them being some mad wizard's lair or a trap-laced tomb. Corridors are long and meandering, and the area may be decentralized in a way that forces you to go far out of your way to get to a room that may not even have a real purpose other than "holding treasure".

Here we reach the obvious divide: a real-life building is designed to do something, a game building is an obstacle painted over to look like a real thing. It's a lot like World War 2 in games - the outer shell is the same, but the inner workings are entirely different. Most games could replace their levels with blank white facilities and the only real difference would be in the player's perception of it; in terms of its layout, it would be just as alien from a "real place" as a normal level. All the linear requirements of most action games make it impossible to comprehend a location as an actual place where work gets done.

Cities are, naturally, even more complex. While there are many examples of twisting, turning city streets, the general idea of a city is that people should be able to get around. However, in games, cities are often the same as any other level, but more grey. The biggest offender, in my mind, is Metal Gear Solid 4, where the entire urban landscape only really exists as places to sneak through and, as such, is incredibly unrealistic. There's few to no doors, there's ramps and ladders in places that make no sense, and the streets are crowded or choked off. There's plenty of ways to justify this as a level, but not as a city you can imagine people living in. There's too much stuff that just exists to facilitate Snake creeping around.

This certainly isn't helped by the fact that the game spawns enemies and allies, rather than giving them a real logical way to enter the battle - in one area, they climb over a wall to enter the area, and if you go to the other side of the wall they still climb over the wall, but now from the other direction (as in, there's some infinite dimension where they're coming from - there's no actual way into the area for them). There's some semblance of streets and rooms, but they're incredibly shallow - some crates and barrels here and there, but nothing that actually suggests that people ever had a use for the building.

Another major offender is the Resident Evil series, both in urban and facility-based fields. In Resident Evil 1, the mansion is meant to be a combination of a highly-secure laboratory and a "decoy house" on top of it. However, attempting to get any work done in a facility like that would prove implausible, due to the number of puzzles and traps that serve as part of the game's "survival horror" element. This includes jewel retrieval, block-pushing, and the assembling of various "emblems" in increasingly unlikely ways. While it might serve to illustrate how paranoid the Umbrella Corporation is about its secrets being revealed, it also indicates that a given worker has to go through a huge number of hurdles just to get to his secret zombie virus medical lab. What if there was an emergency? Oh, wait, there was and everybody turned into zombies.

On the other hand, at least the mansion was specifically designed by Umbrella, no matter how implausible it was. In Resident Evil 2, the game takes place in Raccoon City, which - conveniently for the gameplay - happens to be just as tight and constrained as an underground laboratory. Cars, especially, serve as blockades, often jammed into alleys they have no business attempting to drive through. It also manages to include the same puzzle/key gameplay, meaning that the protagonists are running through alleys and warehouses to find jewels and emblems and keys to open doors in office buildings and police departments. The end result is that "Raccoon City" is basically just another labyrinth, but with a different paint scheme.

However, there are some games that manage to have more believable building layouts. These include the police simulator "Swat 4", the assassination game "Hitman", and the mercenary-based "Jagged Alliance" series. The reason for the difference tends to be that these three games are generally open, and use their levels as areas to be explored rather than to be progressed through. Hitman will generally place the character in an area and then set limits on the area in a way that allows the building itself to be mapped out more naturally. In action games, the fact that you're simply fighting your way through the building means that it's a setpiece, rather than a real element. In SWAT 4, things like doors and corners are an important part of the game, and the gameplay takes advantage of the "real" buildings rather than them being an impediment.

Another way to make areas feel more realistic is to make it so that the same requirements apply to them as to real buildings. That is, workers have to move through them quickly and have all the same areas to allow them to do their jobs. One game that managed to do this was "Evil Genius", which tasked the player with creating a secret island base like the ones found in old James Bond movies and so on. While there were a lot of unrealistic things in the game, one more plausible element is the fact that workers are constantly moving through your base, either to go to a work area or to get some rest or downtime. On the "human needs" end, there's barracks', mess halls, and rec rooms, while on the "purpose-based" end there's armories, control rooms, and laboratories (among others). Basically every room in the base has a purpose - the only room that it's notably lacking is a bathroom.

While it's possible to put traps in your base to ward off invaders, your unwary workers may trip on these traps if they are overworked and not alert. The "security measures" aspect is undermined by the need for people to actually do their jobs in a timely fashion, and hopefully to not die while trying to do it. Evil Genius certainly wasn't perfect, but it depicted an environment where the requirement for employees to get work done required that (a) they have facilities for all the needs such an organization would have, and (b) that the design of the base allow for workers to get to the areas they need to be at in a timely fashion. Other games that require building a base with sensible features, such as Dwarf Fortress, create a similar dynamic.

The reason that most game buildings aren't realistic is because it's detrimental to gameplay for them to be. In "Max Payne", shown above, the hospital is unrealistically open because Max needs to be able to dive and flip around. In a lot of games, the direction of the game means that allowing too much exploration is unhelpful for what the developers are trying to do. However, facilities and cities are best designed around a roughly squarish pattern, so that all parts are fairly accessible, rather than a long, winding path with only one way to reach each new area. Imagine if your house was built like that - you'd have to go through four or five rooms to reach the bathroom. It's a practical model for a video game level, because they're trying to make you take things one at a time, but not for an actual building that people live in, because having to go through the rest of the house to reach an area is not a good design.

The reason, I think, that games like SWAT 4, Jagged Alliance, and so on can occasionally subvert this is that their levels are more open, but also more closed off in a larger sense. There's no transition to other areas, at least not within the area itself (Jagged Alliance had area transitions by walking to the edge of the area). The actual building exists only within the confines of the area itself; they don't have to worry about connecting it to the rest of the game. One level of "Hitman: Blood Money" takes place during Mardi Gras in New Orleans, and the level's limits immediately show themselves - there's a lot of locked doors, a lot of too-small buildings, and so on. The fact that a larger area has to be simultaneously established and cordoned off influences the development of the buildings that make up the whole thing.

This hypothesis can also be connected to the level design in MGS3, where the buildings often stood in a larger area, with a lot fewer constraints. This meant that the constraints of linearity were often on the larger area (forests and cliffs serving as impenetrable barriers) while the building itself could be more sensibly designed. In MGS3's online mode, the design of urban areas was generally a lot more similar to the MGS4 example, with unlikely ramps and ladders and alleys everywhere. This is because in such an environment the city itself provides the constraints and scaling, which is less realistic than having a smaller structure in the middle of a larger open area.

Furthermore, those "open" games exist to provide areas for the player to scope out and observe things like patrol routes, enemy locations, and so on. The map needs to be open because the gameplay supports it - identifying enemies is just as important as neutralizing or avoiding them. The game's flow supports an open map because the player's job is to move cautiously through an area, paying attention to the surroundings to identify enemies. In contrast, an action game is always moving the player forward, because the goal of the game's design is to advance through contact. Therefore, a game with a loose objective like "move through the town" would be more likely to have an open, realistic design than a game that had to close off alternate paths to force the player through a specific area. The area itself may still be an obstacle, but it is an open one where the potential for ambushes or the memorization of guard patrols is part of the gameplay. In a more linear game, "getting past blockades preventing forward progress" is generally the main focus of the gameplay, and this is reflected in the level design.

So, to sum up:
1) Like anything else, real buildings have a wide variety of roles they need to fulfill, and to make a building feel "real" it should have all of those same role-filling areas. The best way to make a plausible building is to make it fulfill all the roles it would have to do in real life, and introduce the same factors that would affect it in real life like travel time and safety.
2) An open building requires accessibility, just as it would in real life, while a linear building needs to convey a sense that at some point it was an open building, but now it no longer is. This is generally harder than it sounds, because the ability of the player to influence their environment often has to be disconnected from the fact that they're carrying around rocket launchers. In general, though, an open building will be designed for ease of access (with a few entry/exit points), while linear buildings are designed to slow the progress of the player (and, by association, everyone else who works there).
3) If you're designing a city, remember that cities aren't scaled like normal video game levels - they have alleys and so on, but for the most part there need to be roads and streets that actually suggest the ability to allow cars to pass and so on. Also, try to avoid the use of things like ramps and ladders where they don't belong - keep things like accessibility in mind so that it actually makes sense that people who live there would be able to get around.

If worst comes to worst, you can always refer to the actual OSHA regulations. They're a pretty solid guideline for how a building should be designed, allowing for the safety of the workers with minimum fuss. Other than that, just think about everything that a building would need to cover and try to fit it in.