Recently we wrote about computer hardware industries stuck in the stone age. That article examines hardware industries relevant to PC gaming. Today, we will look at the various areas of game design that are stuck in the past and not where they should be.
Some areas of technical graphics fidelity, namely shaders and textures, combine to make what is probably the only area of game design that improves constantly, generally at a fairly steady and predictable rate. Scale (the size of game worlds/levels) is something that also continues to increase, although the artistic aspects of level design as well as attention to detail aren’t so quick to improve. But many aspects of game design, some of the most important ones even, have either not improved or have gotten worse compared to games from 10-15 years ago, believe it or not, and these are the things we discuss starting on the next page.
Of all the game design aspects we discuss in this article, this is the most important one. Let’s start with the most obvious: Combat AI in action games has not improved within the last ten years. AI in Turok 2: Seeds of Evil, older Serious Sam games, some of the older DOOM games, and Half-Life are not worse than your average shooter today. They respond to the same stimuli and perform mostly the same actions. Rainbow Six 3 AI is better than newer Rainbow Six games. Half-Life 2, Call of Duty, Return to Castle Wolfenstein, and especially S.T.A.L.K.E.R. combat AI are steps above most action games old or new, and then F.E.A.R. has by far the best combat AI (at least for any amount of enemies versus a single target, that being the player) of any action game. Although S.T.A.L.K.E.R. AI handles large scale battles better, perhaps better than any action game, best showcased by Call of Chernobyl.
Deus Ex: Human Revolution – Director’s Cut (2013)
Thief II: The Metal Age (2000)
Turok: Dinosaur Hunter Remaster (2015)
Shadow Warrior (2013)
Fallout 4 (2015)
The AI in almost all action games, and even tactical RPGs, is still quite dumb and no match for a seasoned player even when the player has no advantages. This is unlike modern strategy games (but not older ones) such as Total War: ATTILA, Total War: WARHAMMER, XCOM: Enemy Unknown, and XCOM 2 all on harder difficulty modes. The AI in these games are a major improvement over past strategy games, and they can actually compete with moderately skilled players. They demonstrate impressive coordination and teamwork. We need combat AI this good in non-strategy games too.
An issue most gamers make is mistaking good AI for “an enemy AI’s ability to kill you very quickly in a shooter.” The only explanation for this is gamers being oblivious to how easy it is from a programming perspective to simply make AI never miss a shot, but what is more baffling is how such gamers fail to realize how unrealistic and inhuman this is.
Gamers also struggle to differentiate between scripted and dynamic/procedural events and AI. The latter is infinitely superior, more realistic and more impressive programming, and it leads to diverse gameplay.
How about simulation AI for open world and other large scale games? Bethesda Game Studios tried to make a serious advancement here in 2006 with their Radiant AI system in The Elder Scrolls IV: Oblivion, and then GSC Game World elevated it tremendously with their 2007 game S.T.A.L.K.E.R. Shadow of Chernobyl and continued to improve in their sequels. Nobody has tried to evolve simulation AI since. Bethesda has made some slight, marginal improvements with The Elder Scrolls V: Skyrim compared to Oblivion, and GSC unlocked their simulation AI a bit in S.T.A.L.K.E.R. Clear Sky (2008) but again, all marginal improvements.
The only major improvement since then has come from a mod for S.T.A.L.K.E.R. Call of Pripyat called Call of Chernobyl, which obliterates the simulation AI from all other games. But aside from this mod, simulation AI still has a large way to go before it starts to resemble something truly impressive, something worthy of being called a virtual world or truly “next gen” games.
Sadly, great AI doesn’t sell games (excluding certain strategy games), so developers do not bother. It will probably be decades before we see significant evolution in video game AI. Until then, it remains one of gaming’s biggest obstacles. Until then, Call of Chernobyl is the only game (and it’s actually a mod) that appears to be on the right path, and it is already perhaps a decade ahead of its time just as Shadow of Chernobyl (released in 2007) was, and just as F.E.A.R. (2005) was. The proof is in the pudding; those games are over a decade old now, and still superior to every other game outside their franchises.
Hardware and game programming have been somewhat of a barrier for AI as well, but now with the widespread adoption of multicore processors and DX12 and Vulkan APIs, this shouldn’t be an issue. Ideally, AI would be GPU accelerated, preferably via OpenCL.
It’s a big problem when modern physics intensive games like Battlefield 4 and Battlefield 1 use the same physics as Red Faction (2001) albeit on a larger scale. Then again, Red Faction’s physics are applied on a larger scale than most modern games. Half-Life 2 (2004) has greatly superior physics to just about all modern games, and in many ways. Remember that physics doesn’t mean destruction, destruction is just a part of it.
Destruction in games like Battlefield 4 isn’t even procedural, it’s pre-determined. Everything breaks the same way, in the same places every time. In Battlefield: Bad Company 2 you can even see the lines that show where they would break, but at least that is gone now.
With that knowledge, now look at these demonstrations of NVIDIA PhysX (formerly Ageia PhysX), by far the best and most advanced game physics engine in existence. Credit given to the video uploaders and creators.
In order for such advanced physics to be effectively used on a large scale, it needs to be GPU accelerated. CPUs cannot handle it. To illustrate this, a NVIDIA GeForce 8800GTX delivers very similar performance in the Cryostasis Tech Demo to a 4.6 GHz Intel Core i7 6700k, although that’s only using one thread from the 6700k.
Now that PhysX has been ported to DirectCompute, that means any modern video card can run it. Still, an OpenCL alternative is preferable since it offers potentially better performance due to parallelism. Sadly nobody has attempted to compete with the realism and advancement of PhysX. Crytek attempts to drive physics forward in their engine, but to our knowledge it is still CPU limited not to mention it is integrated into CryEngine which is very limiting. And at the end of the day, it is still inferior compared to PhysX.
Physics aren’t just for fun and to look pretty though. Gameplay can be designed around it, as Half-Life series, F.E.A.R., Dark Messiah: Of Might and Magic, Crysis, and various horror games wonderfully demonstrate in different ways. Dark Messiah is probably the most appealing example to most people; it is the best example of an action game that lets you use the environment to kill your foes. Collapsing things onto or from below enemies, using an ice spell to make the floor near a deadly cliff slippery, using a telekinesis spell to hurl objects at enemies, and far more as this video demonstrates.
Other games/genres can find their own way to incorporate physics into gameplay, such as in exploration. New gameplay concepts would undoubtedly be born of such advanced physics. Ballistic physics, for example, is one area of physics that could use an improvement as well. The Arma games are on the right track, but still don’t have simulation of windage plus they are still rather inaccurate especially for high caliber sniper rifles, and bullet penetration is primitive in games still.
Left = Creative Sound Blaster X-Fi Titanium HD, Right = Creative Sound Blaster X-Fi Titanium Fatal1ty Pro
We return to sound once again. Sound effects design in games has largely gone backwards, has gotten worse, since 10-15+ years ago. Hard to believe, but it’s true.
The issue isn’t purely technological. Today’s Steam Audio far surpasses DirectSound3D and OpenAL. With the release of the Dolby Atmos standard, which places speakers above you to directly replicate the missing Y coordinate from normal surround setups, we have all the potential in the world now to create amazing, truly three dimensional sound. Sadly, Dolby Atmos in games is niche with only a few games supporting it (licensing anything Dolby is expensive so indie games won’t ever use it). DTS:X is a less costly equivalent, but not at all used in games. Besides, we already had these capabilities in DirectSound3D and OpenAL, we don’t need these standards for vertical audio.
Games today are just now catching up to what games accomplished in the late 1990s and 2000s with DirectSound3D and OpenAL, but the potential of Steam Audio and AMD TrueAudio Next and features like physics based sound propagation with ray tracing are not being explored.
Multiplayer game design has regressed so much, shooters most of all. The multiplayer shooter genre has been degraded more than any other. What we have today are objectively lesser games, with things like free customizable dedicated servers, extensive modding, and more original concepts being rare. As far as AAA multiplayer shooters go, only Valve still stands strong even though they fall short of the “original concepts” part. Aside from AAA multiplayer shooters, which (aside from Valve) are generally inferior, we do have a fair amount of multiplayer games that have not deteriorated at least.
There is no better illustration of an objectively lesser game than this.
But this is a topic we have written about extensively. Read all about that right here. But in a nutshell, we are referring to the fact that 2000s PC multiplayer games usually have these strengths compared to today’s:
- More official content.
- Free DLC was common.
- Almost limitless mod support with official mod tools/SDKs, auto mod downloading for server side mods. These mods often led to some heavily modified servers running custom game modes. Prime examples of this include the Counter-Strike mod for Half-Life, Natural Selection mod for Half-Life, Day of Defeat mod for Half-Life, Team Fortress mod for Quake, Killing Floor mod for Unreal Tournament 2004, Red Orchestra mod for Unreal Tournament 2003 and Unreal Tournament 2004, DotA mod for Warcraft III, Garry’s Mod mod for Half-Life 2, Insurgency mod for Half-Life 2, DayZ mod for Arma 2, and PLAYERUNKNOWN’s Battle Royal mod for Arma 3. You might recognize all of these as global bestselling games.
- Free dedicated servers, via a dedicated server application included with the game was the standard practice. Whereas now they are often paid and very restrictive, as shown above.
- More configurable servers, with parameters such as ping limit, friendly fire, weapon restrictions, spawn items, and much more, just without mods. Mods would obviously add far more.
- Deeper, more complex and tactical gameplay was present in some but not all games.
- No annoying forced external programs like Battlelog or uPlay.
- You never even had to question whether or not VOIP was included. It was a given, as was text chat.
Almost all of today’s AAA multiplayer games are only objectively lesser games because many 2000s PC games offered the same things and infinitely more.
Of course we had to bring this up. It’s a bit more specific than the previous topics, but it is a noteworthy area of game design. Like sound, it has regressed. So many RPGs today are like past ones, but lesser. Less role-playing (thus less dialogue), not to mention less gameplay depth and diversity and usually less originality and style, and much more shallow world building and writing in general.
This is illustrated by our RPG tier list. Take a look at every game and their release date on every tier. Only one game released in the 2010s made it to tier 3 or above, which represents a massive leap in objective role-playing capability compared to tiers 4 and above.
Fallout: New Vegas. Isometric point of view and turn-based or real-time with pause gameplay are not required for high levels of role-playing, as this game demonstrates.
No RPG has reached the heights of gameplay depth and diversity shown by Neverwinter Nights 2, nor Neverwinter Nights and The Temple of Elemental Evil. This means all RPGs since then have fewer and simpler character builds, far less skills (which are used wonderfully in those games), and less ways to play the game. Sure, not every RPG requires such elaborate gameplay design, but those that have tried fall gravely short, like Pillars of Eternity and Divinity: Original Sin, which are not bad or borderline bad in these regards like Tyranny, Dragon Age: Inquisition, and most modern RPGs are.
But most of all, RPGs of the 2010s except for Fallout: New Vegas have far less role-playing. Greatly reduced ability to define your character’s persona during character creation and through dialogue throughout the game and through your own actions. Reduced role-playing and character agency, and the game world and it’s characters do not respond to your character nearly as much nor in as many ways.
Personally, I would not be surprised if we never see another “tier 3” role-playing video game as defined by this article. I had already completely given up on tier 1 long ago, or RPGs with as much role-playing as Fallout 2.