How to innovate the First person shooter (FPS) genre?

GeneralJist said:
what would you cut from the conventional FPS?

I would cut anything that has been added over time: Reloading weapons, and extended movement (e.g. dash on pressing the left way button two times).
The perfect example seems Quake → Unreal. Epic has added such things, either while aiming to innovate, or just to be a little bit different from the original.
Now Unreal players may disagree, but following the argument ‘easy to learn, hard to master’ to conclusion is obvious: Adding complexity bloat does not make it easier, and so it does not make it better either.

GeneralJist said:
What would you want and or expect from a project that tries to innovate this genre?

Sadly, the more you'll try to innovate classical core mechanics, the less it will work. That's really demotivating for all of us i think, but it's the hard truth.

Still, there were some innovations that worked:
Zoom in for sniping weapons, e.g. Quake 3 but mostly MDK. That's great, but only for single player.
Jetpacks, e.g. Tribes. Also great, mostly for multi player i would say. It's a way to make playing on boring hight maps interesting.
Vehicles or Portals.
Various game modes. Capture the Flag… Battle Royale…. and finally now ‘Extraction Shooters’? (I do not even know what that means.)

Coming up with something new seems hard. If you want to innovate, i would not touch the core mechanics too much, but rather focus on level design and enemy behavior. Here innovation is possible and welcome.
For example, modern games tend to heavily underuse dynamic level design, speaking of moving platforms or skull rafts from Super Mario. The jump pads in quake 3 were also great. But modern games are way too static, making them boring.

However, if we are really ambitious and not willing to accept FPS is done and can't be improved anymore, it's easy to see the main limitation of FPS:
Jumping lacks accuracy and we can't get back what we've lost from 2D platformers. You can't see your feet and the ground, so it's hard to time your jump from the edge. And worse: FP lacks oversight, hindering not only tactics but also distance estimation needed for advanced platforming.
I think i know how to solve this btw. My idea involves to sacrifice FP immersion but using some ‘robust’ TP camera instead, which also works for interiors. What i have in mind would force players to get comfortable with some new things, likely feeling pretty unusual at first. Similar to back then, when people had to learn mouse look to play post Doom FPS games. Not sure how this works out, and it's hard to try since traditional rendering is not ready for my idea either.

But yeah, imo the main problem of FPS is bad oversight at the cost of immersion. Back then the immersion was so impressive it was worth the sacrifice, but today there is no more wow effect of ‘being in the game’, and that's why FPS is pretty dead. (Beside shitty console gamepads ofc., since you can not even do a proper FPS on consoles.)

To say it differently, you worked on RTS the most afaik. The question from that perspective is: How can we make the global connections and actions of an RTS observable in a FPS?
If we can answer this, potential to innovate opens up in all directions. We could finally merge genres ‘for real’, and probably much more than just that.

I don’t know if modern FPS games already have this feature but controlling bots using speech recognition is an idea. I think this is plausible.
Another idea that also has to do with communicating with bots is to have them keep the human player/players up to date by spoken word about the things they are currently doing. I know bots in Unreal will talk to you once in a While but they provide little information. I was once thinking you could have a viewport in your hud displaying what the bots are seeing. The problem with that is that your mind is submerged in the level based on your own camera view switching to a different view means that you need to step out of your current view and step into the bots view. Basically you need time for adaptation. You can’t be on two maps(views at the same time). With sound it’s a different story but it still requires some degree of multi threading on your minds behalf. Who knows one day maybe technology will allow us to be in two places at the same time.

I personally hate jumping in computer games. But I doubt that many players feel that way.

I would love if there is a story. I liked Halflife because there was a story.

This could be an interesting thread to keep going. Lets try to break down what an FPS consists of:

• The character controlled by the player
• Movement and possibly special-abilities for the player-character
• Gunplay, including tactically reloading and conserving ammo.
• Interactable props such as collectible weapons and powerups, or vehicles or teleporters
• NPC enemies, and their abilities and behavior.
• NPCs that are neutral or friendly.
• Characters controlled by other players.
• Interactions between characters
• The level itself
• The UI, including HUDs that are always visible, and maps and other secondary screens.
• The overall gameplay model (such as story missions, openworld, going from A to B, corridor-shooter, team-shooter etc)
• Changes to each of the above, either as a result of player or other character actions, or the progression of time or the game world.

Did I leave out any? Every FPS can be broken down into how they set themselves apart in one or more of those categories. I'm sure if we all worked on brainstorming small ideas for each one, it could get interesting in aggregate.

All the recent hero shooters have been working on the first three points, with unique characters abilities and weapons, so that would be an uninspired direction to continue in.

Brian Sandberg said:
Did I leave out any?

I think of a few:

• Interactions of the player with the environment. That's usually just pressing buttons to call elevators, or mounting a static turret. Things like that. Not too interesting. I always think we could come up with more interesting dynamic environments. I think of complex machinary, raising bridges or opening up new pathways. Puzzle stuff, but ideally not too constructed. Something that enables optional, creative gameplay. But i fail to come up with somethign concrete.
• The quick save function. In early FPS this was used as an actual mechanic. This could be extended. Imagine some time travelling backstory, and you can do some character build. Skill trees and such crap. And now you can use your saved games to travel to the past, but you can combine the saved states, and take one advanced character build from the later game to an earlier section. Being more powered up you can now do some things you had failed on early, and maybe this enables some other progress at a later time. But not sure if i would like this. Maybe i would feel the save game management is just complexity bloat and this should not be needed at all.

Brian Sandberg said:
NPCs that are neutral or friendly.

That's where i personally see most potential to innovate.
I think of the scientists in Half Life. I would like to get such behavior, but not from scripting, rather entirely procedurally.
It's fun to predict enemy behavior. The same should work for friendly NPCs. Maybe i could affect them indirectly, so they do things which help me. If i can make them to walk in front of me, they could ignite landmines or catch some bullets, etc. I would like that. >: )

JoeJ: I would like to get such behavior, but not from scripting, rather entirely proceduraly

You need to tie a chat bot to a problem solving algorithm that works in a 3d environment I think. Here is an idea: a bot A in a room monitors all objects inside the room ( to see if interactions between objects take place ) if for example another friendly bot B appears in the doorway that’s an interaction between an object and a character ( which is just another object ) hence bot A which is observing the environment for changes will use a greeting phrase. If an object is about to fall on bot B bot A will use a phrase to signal danger. Its not that complicated. Some scripting needs to exist IMO, open ended, MMO type games are boring. The world needs to be going somewhere.

The reason I’m saying you need a chat bot is that if the environment is changing that affects what the characters are saying and doing. It’s difficult for a robot to scan the environment in a game you don’t have such worries. Once a bot enters the room it’s easy to pass a list of objects that are found inside the room.

JoeJ said:

I think of a few:

Oh I thought I had the interactable vehicles/structures, but yeah, quicksave and all that kind of meta-level stuff. That would also include level editors and similar customization of the game, along with matchmaking and server/game hosting and anything that goes into joining a game.

Matchmaking feels like a problem that hasn't been solved. I used to play a few rounds of Overwatch regularly, but these days 90% of random matches result in extreme one-sided defeats or victories which can be called within a minute, instead of the balanced brawls that are fun. It might not be a sexy feature to work on, but I can think of few things that would do more for the genre.

We should also try to bound the range of ideas that would actually classify as “FPS innovations”, as opposed to turning it into another genre. It's easy to throw ideas at the wall and end up with a first-person puzzle game, or a vehicle sim where you also occasionally shoot a gun.

Calin said:
You need to tie a chat bot to a problem solving algorithm that works in a 3d environment I think. Here is an idea: a bot A in a room monitors all objects inside the room ( to see if interactions between objects take place ) if for example another friendly bot B opens the door in the room that’s an interaction between an object and a character ( which is just another object ) hence bot A which is observing the environment for changes will use a greeting phrase. If an object is about to fall on bot B bot A will use a phrase to signal danger. Its not that complicated. Some scripting needs to exist IMO, open ended, MMO type games are boring. The world needs to be going somewhere.

Regarding AI chatbots i'm doubtful. I expect a new all time high of low brow, AI generated content is coming.
People will only accept this crap if their content addiction is already that bad, so they indeed listen to a machine talking to itself. Or even worse, they will talk with those machines themselves, more than to their (probably non existing) family.

Will a future society indeed accept an AI entertainer?
Idk. But probably not all people will. And i'll certainly join the resistance, serving them with ‘handmade’ games if i can.

Calin said:
Who knows one day maybe technology will allow us to be in two places at the same time.

What technology should this be? Neural brain implants or other transhumanism ideas? /:O\

Omg that's even worse. Facing a rising AI god is one thing, but becoming one with it? Fuck, no.

I really think there is a point where we have to stop developing useless tech further and further, at latest once that tech imposes a threat to our society and economy, health or even biology.
And maybe that point is now.
Altman and Jensen can take their AI crap and shove it up their asses, so it goes back to where it comes from.
I don't want it, and i will not use it.

So i will design around a need of dynamic conversation in my game. All this blah blah and cutscene nonsense is just boring anyway. It's a tool movie makers work with, but it's not for us game devs imo. Games are not about talking NPCs. If i need that, i'll do a multiplayer game, letting real people play a role and talk. And if i need intelligence in a game, i will use my own to fake it well enough.

But maybe that's just me, and AI will give us those new and awesome games we desire.
Maybe our lacking creativity can indeed benefit from such AI crutches. We will see, but i hope not.

Otherwise i agree about your examples.
I think friendly NPCs need a routine. Something simple and predictable.
Maybe they are workers in a factory. They work on a conveyor belt. It's all part of some larger mechanical machinary i've mentioned before.
They produce bullets. And the conveyor belts are actually supply lines, transporting ammo to some distant outpost of enemy guards.

Notice: Instead proposing brain implants so you can ‘imagine two different places at the same time’,
i rather attempt to build up motivation for the player, so he uses imagination to think about the option of manipulating the supply line. The player may not actually see those larger scaled relations visually, like he would in a top down RTS. But it is enough if he has reason to think about such large scaled relations. Then he will just do it: Sabotaging the supply line, eventually by manipulating worker NPC behavior, and then enjoying to beat out of ammo guards easily. It will be fun to the player, and we come closer to the goal of merging FPS and RTS mechanics.

Ofc. it's some work to figure out all the details, but i see no need for dynamic NPC AI conversation. And expecting ChatGPT NPCs won't be practical on consumer hardware, i have no interest in related cloud services anyway.

Brian Sandberg said:
We should also try to bound the range of ideas that would actually classify as “FPS innovations”, as opposed to turning it into another genre. It's easy to throw ideas at the wall and end up with a first-person puzzle game, or a vehicle sim where you also occasionally shoot a gun.

I'm afraid we are done with that, and thus the goal of inventing a new genre is our only option to innovate at all.

I'm still primarily a FPS player. But the genre is dead. Not because modern AAA studios fail on it, but because it's just done. It's over, like Rock'n'Roll.

The best games i've played last year were the new Amid Evil DLC, and the Quake II remaster.
Both those games are FPS masterpieces.
But from the current day perspective, neither game tries to innovate. No new ideas. Just perfect level design, basic movement and combat.

You can't do it any better. No way.
If you want to do better, you have to do something different.

Question:

Is nostalgia as big in the FPS community than it is in other communities?

I wonder…

GeneralJist said:
Question: Is nostalgia as big in the FPS community than it is in other communities?

Sadly yes.

Arena Shooter: gone.

Narrative FPS: gone.

Retro / Boomer shooter: gone.

All that's left is Call Of Duty, Battlefield, Fortnite. Which is all multiplayer only, basically.

So nostalgia is all we have, and only indies serve the niche. Games like Dusk, Amid Avil, Prodeus. They are very good, but it could be more of it… ; )

Notice Dusk is really amateurish in some ways. Really poor graphics. But it's probably the most known / successful FPS game of modern time. I did not want to play it, because it looks so bad. But then i've tried it anyway due to lack of alternatives, and it's super fun. I'm happy the dev is already working on another game.

Then we have a whole lot of indie FPS games with 2D sprite enemies currently. At least the devs seemingly think that's popular. Prodeus for example uses sprites as default, but you can change it to 3D models in gfx options. The option is hard to find, but makes it a much better game. (It's pretty perfect.)

3D Realms has some yearly show about their games in development. (Better than Game Awards). Some use the original Quake, Doom, or Build engines. Ion Fury or the upcoming Selaco for example. Or look at the recent Bolt Gun, which also uses 2D characters.
For me that goes to far. I don't want to go back to such extreme technical limitations, and usually i reject such games. But due to lack of alternatives… some times i have to accept even that. : /

It's maybe worth to look at newer AAA FPS too, because it tells us what true innovations have added to the table:

Titanfall (i know only 2): Wallruns and parkour movement.

Doom Eternal: Added basic Melee combat (glory kills - basically just a gory execution animation on a button press, ‘pausing’ the rest of the game for a second.)

Both games are great, but the new mechanics violate the ‘easy to learn, hard to master’ principle. To me that's just complexity bloat and annoying. I don't want those new mechanics and can't enjoy them. Same for cover mechanics as in Gears of War.

A similar thing is hyper cool John Woo karate movement skills and bullet time effects, first introduced with Max Payne. So far i have not seen a single game where this really worked well. Controls tend to become complicated for no real win. The recent El Paso maybe is an example for such games.

A good example of innovation is Dishonored, which i personally consider the last great FPS.
They added lots of special abilities.
Usually i truly hate this. Control for example. They have a levitation ability, but you have only one button to control both lift off and landing. It's terrible. I was constantly falling to death by accident. It was so bad, i did not even gave Alan Wake 2 a try after that, assuming the devs became just incompetent to make a playable game.

But Dishonored was somehow great. One special ability was to become a rat. And being the rat you could traverse narrow passages, e.g. the network of drain pipes in a building. Traversing the place this way gave me volumetric mental model of the scene, solving the primary ‘missing oversight’ problem of FP games to quite some degree.
If you look for innovations, i would recommend Dishonored.

But interestingly we saw the most progress by doing the opposite of adding new features: Removing the ability to shoot.
Penumbra did this. It's basically a narrative FPS but you can not shoot. You can only run and hide.
And the game has coined the Indie Horror genre, becoming only really popular with their follow up game, Amnesia.
But maybe it was also the first Walking Simulator in some sense. So a single game gave us actually two new genres.
Beside the reduction they also added a new and innovative feature: A realistic and robust physics interaction model, putting HL2 to shame.
It's really an underdog, but imo the most innovative and immersive game of all time.

JoeJ said:

But the genre is dead. Not because modern AAA studios fail on it, but because it's just done. It's over, like Rock'n'Roll.
You can't do it any better. No way.
If you want to do better, you have to do something different.

There is a lot of truth to that. The formula we have has been polished to seemingly near-perfection, and although there still is room for incremental improvements. diminishing returns have definitely set in.

But innovation is by definition surprising; if you could predict new ideas you'd have them already. It would be shocking if there aren't a few fundamental idea left that we haven't stumbled across yet in the FPS space. I doubt they will appear first in AAA projects though, since those have become so risk-averse.

JoeJ said:

Arena Shooter: gone.
Narrative FPS: gone.
Retro / Boomer shooter: gone.

The hero shooter has taken over from the arena shooter.

Aren't most AAA games more narrative than game these days? Which would you consider the last great narrative FPS?

There's been so many boomer shooters over the last few years, from Prodeus to Boltgun, and increasingly esoteric indie shooter show no sign of stopping?

What do think of new gameplay modes? Extraction shooters, battle royale, and whatever the hell Gray Zone Warfare is going to be classified as. That's probably something where we can continue to come up with ideas for a long time.

Brian Sandberg said:
But innovation is by definition surprising; if you could predict new ideas you'd have them already. It would be shocking if there aren't a few fundamental idea left that we haven't stumbled across yet in the FPS space. I doubt they will appear first in AAA projects though, since those have become so risk-averse.

Haha, yes. And actually, sitting here discussing FPS game design ideas, i do not feel like a dev. I feel like a gamer, who thinks he could do anything better, planning design ideas just in his imagination and assuming it would be just awesome. :D

That's not how it works, to me at lest. The process is coding the game as intended, then i see it does not work. But i might also see something else works or feels promising. So i abandon my initial design idea and follow the new path, to see where it takes me.

It's basically a random walk. But i guess you can't do this for a billion dollar AAA production either, so the AAA industry follows the same fate than their audience: They degrade to wannabe game designers, who think they could plan a video game on paper, then let others care about implementation details afterwards.

Brian Sandberg said:
The hero shooter has taken over from the arena shooter.

Ah yes. I forgot some other things too, just because they are not on my personal radar.

Brian Sandberg said:
Which would you consider the last great narrative FPS?

Maybe my definition is personal, but quite precise: HL is not a narrative FPS to me. The classical example is Singularity. In this game you are lonely. You arrive at some military research facility, where something went wrong. You pick up story fragments in from of notes. You read them if and when you want. There is a lot of environmental storytelling too, like in HL. But this can't tell an actual narrative. The collectable text / audio logs do this. They tell the story, and you put it together like a puzzle in your imagination. Only video games can implement such form of story telling.

That's what narrative FPS means to me. It does not interrupt gameplay, like cutscenes do. The player has full control of consuming story, and it's technically optional. (It's also cheap to produce.)
Modern AAA games are a lot about story telling, but they do so like Hollywood. And why not? People have just shitty gamepads, so an immersive experience is not possible anyway, and it's no big problem to interrupt gameplay.

And this gives us the hungry niche of former single player FPS players. They don't want to consume entertainment, they want an experience.

Brian Sandberg said:
There's been so many boomer shooters over the last few years, from Prodeus to Boltgun, and increasingly esoteric indie shooter show no sign of stopping?

Yes. It's not enough, but there are still new and good games to find. I still call it ‘dead’, but i just tend to exaggerate to make a point.

Brian Sandberg said:
What do think of new gameplay modes? Extraction shooters, battle royale, and whatever the hell Gray Zone Warfare is going to be classified as. That's probably something where we can continue to come up with ideas for a long time.

I'm already out of date. Just recently i asked myself: What the fuck is an ‘Extraction Shooter’? Idk.
Because i do not play MP anymore, i guess. I've played a lot of LAN MP Quake 3 and later some Quake Live over internet. That's awesome, and worth it's own discussion, but personally i just focus on single player design topics all the time.

However, i've noticed the ‘shrinking circle’ in PUBG, to shrink the map and enforce a winner in Battle Royale.
That's a brilliant idea ofc. If you can come up with such ‘surprising innovation’, you may be a made man. : )

Edit:

Brian Sandberg said:
Which would you consider the last great narrative FPS?

Scorn. So that's not too old. :D

Replying to the initial question: How to innovate FPS?

I think there has been a shift towards more realistic war games in FPS. However, even the “gritty” ones still let you play as a supersoldier with perfect aim and technique. Have you ever held a gun? Try aiming down the iron sights and walking. No matter how many hours you train with that, you will never get it perfectly steady, but in all FPSs that allow ADS besides CS that I came across, it works exactly like that. Another thing to try: Aim down the sights and then turn around. Even just moving your upper body will mess up the aim. But in all FPS out there, you can just jerk around 180° without losing precision. Even just keeping a steady aim while standing or being crouched is hard. Unless you have a proper rest and/or a bipod, even prone aiming is hard.

If you added a precision penalty for walking even while aimed down sights, and for turning, and maybe even limited the speed at which you can turn or accelerate your turning motion, that would completely change the game. You no longer get instantly killed after peeking out from behind cover, unless someone already aimed at exactly where you are. If anyone jerks his gun over to aim at you, he will incur a brief accuracy penalty. And aimbots become less impactful as a free bonus. These mechanics make for a less action-based and more strategy-based gameplay, as precision becomes harder to achieve, and you have to make tradeoffs between mobility and precision, but also allow you to execute more complex maneuvers that would get you instantly killed in conventional FPSs. Additionally, it will finally even the playing field between controllers and mouse a little.

Bullet drop mechanics are cool, too.

RmbRT said:
Replying to the initial question: How to innovate FPS? I think there has been a shift towards more realistic war games in FPS. However,

Basically you propose more realistic body and weapon physics for an innovation of FPS?

But there are some problems you did not address…

RmbRT said:
But in all FPS out there, you can just jerk around 180° without losing precision.

Yes, because they are FPS games. That's all they can do.

RmbRT said:
Have you ever held a gun?

If i do this, i can feel the weight of the gun and it's acceleration. Same for my arms and the entire body. I can feel my movement in the real world.

But i can't feel this for a virtual avatar while actually sitting static, clicking buttons and observing a screen. No haptics, no weight, no acceleration. And because i can not feel the physics of some accurate in game gun and arms simulation, my experience becomes something else: I feel lag and latency. I no longer feel able to interact with the game directly and instantly. I no longer feel like being in the game myself.

And thus we no longer talk about the first person experience which the FPS genre is known for.
I would even say such realism feels simply bad in any first person game, and so your ideas force you on using third person perspective instead.
In TP you do not play as yourself, you play a character. And because you observe your character while doing so, the realistic physics you propose become predictable to the player. That's not yet a solution of the lag problem, but a step into the right direction.

Or, ofc. you can make a VR FPS game, where most of the things you propose have been done already.

Sometimes the reason certain ideas don't exist is not because nobody had this idea yet, it's just that the idea does not work.

But your thoughts are not irrelevant. I think about this a lot too. But interestingly i do not see them as an opportunity to innovate, i rather see them as the biggest problem i have to solve. I have not played RDR2, but i remember people complained it feels laggy due to the more realistic animation. Maybe a good example. We really want better animation, but doing it causes a lot of problems regarding the loss of direct control. It seems you overlook this issue.

@JoeJ No, I was fully aware of that. It gives you hard constraints on the camera movement. It makes you unable to quickly turn around and face something behind you. You must be strategic in positioning and camera control, and you must make trade-offs between different targets to aim at. Aiming at one target makes more distant (in rotation) targets more costly to aim at.

You can give the player a 30° or so area of view that is quick to aim at, and then another area around that that is slower to aim at, and then finally make turning the whole body even slower. You can tune those numbers. It also completely changes the dynamics of snipers and rifles etc. And you can make pistols allow for a much quicker turn speed, for example, allowing you to react to close encounters much better than someone with an assault rifle.

And yes, you technically have similar mechanics already on controller or VR games, but those are enforced by the input device, not by the game. You can simply make a PC game have the same restrictions as if you were using a VR setup or a controller to play it.

By taking away absolute freedom of camera movement and aim, you disappoint people who love that feature, but you open the game up to completely new dynamics which require new ways of thought and strategising. Especially the 360° noscope crowd will hate this kind of game, but I don't care. I can't stand those people.

And I feel that it's not correct to term turn speed caps and/or acceleration caps as “lag” and “latency”. Your action does immediately happen, but you cannot perform arbitrary large actions. The same is also true for instant max-speed movement upon keypress. I don't feel it's correct to label fade-in or fade-out movespeeds as lag or latency. If the player wants to turn quickly, he has to perform a mouse motion over an amount of time. The mouse effectively becomes a joystick, and the act of turning is reflected by you moving your arm around over a short span of time, until thedesired rotation is completed. I would argue that turning actually probably feels more haptic with that mechanic, because you are investing more large-scale movement of your arm to control your character. You get to feel the weight of your loadout and weapon based on the increased or reduced turn rate. Compare that with CoD MW2 where you could take a shield and spin super fast and basically block all attacks. How does that feel real? You don't feel the heaviness of the shield at all, and you move in completely unrealistic ways. Just because you can move the mouse over a lot of pixels does not mean that ludicrous character turn speeds are somehow justifiable.

And I acknowledge that it's an annoying mechanic. However, permadeath in a roguelike is also annoying, yet it is a core pillar of the genre. Through overcoming friction and challenge with skill, the player feels meaningfully rewarded. And this is also not the kind of futile friction like early 3D platformer titles and their horrible camera & movement controls. You still get the same detail of input of a traditional FPS, but with some added friction that completely changes what the game is about.

RmbRT said:
And yes, you technically have similar mechanics already on controller or VR games, but those are enforced by the input device, not by the game. You can simply make a PC game have the same restrictions as if you were using a VR setup or a controller to play it.

The VR controllers do not ‘enforce’ limitations. They allow the player to move the gun as fast as he can, and the game maps this input 1:1 to its simulation. So there is no lag, beside technical limits such as latency,
But you intend to enforce limitations, and they will feel like that.
Did you make such prototype? And it feels good?

Personally i'm not ready yet to do this. My ragdolls can't hold a gun and aim it. I'm still busy with keeping them balanced and making them walk. But i could for example mount the camera to the ragdolls head body, and map mouse input to head rotation. I have never tried this, because i'm absolutely sure it's terrible.
I remember some dev talk about Star Citizen. They have tried this, but it was bad. The solution was to place the camera at a stable point ignoring animated head movements, and culling the head model from rendering in FP, so it does not clip the view. That's what everybody is doing, because it's the only thing that works. Replacing head animation with more predictable simulation won't change anything here.

But there is another, similar but minor problem where i have actual experience: Holding and carrying objects, e.g. the cubes in Portal. If those objects are physically simulated, it is difficult to make them follow a fast camera rotation without too much lag.
The a typical solution is to set their velocity directly, but then we loose the advantage of physics simulation. The physics engine can no longer solve for velocity itself, and thus the object jitters if it interacts with other objects. I did not want this, so i tried to control it only by using force. There is a standard solution for this as well. Physics engines usually have some kinematic joint, which tries to move objects to a given target position / orientation. But those joints are (for good reasons) not smart. Meaning, they drive towards the target as fast as possible by applying allowed maximum force. This causes to have the object at a high velocity when it hits the target. So it will overshoot and keep moving, oscillation forth and back around the target. The impression is again laggy (i lack a better term) and inprecise. The solution for me was a custom smart joint, introducing a given target velocity (usually zero), which starts to decelerate in time so it stops at the target as intended. The results feel very good, not any worse than setting position or velocity directly. I can also move my object with a gizmo, and it feels like positioning objects in a program like 3Ds Max. I still have a limitation on max acceleration to keep things stable, but under practical conditions it's not noticeable, and i'm happy.

However, if i set the max acceleration to a low value, e.g. to realistically model weight of a heavy gun, it feels just laggy and like a limitation again. Many games added such a penalty for heavy guns, e.g. slowing player movements and mouse sensitivity down, and yes this works.
But it works because the effect is predictable to the player. The effect is simple, like head bobbing an early FPS games. A simple sine wave. Not realistic, but predictable.

Head motion of a character you can not see is not predictable. Complex motion of a limb made of multiple bodies to hold a gun is not predictable. Because you can't feel this simulated limb, because it is not your limb. You never complained or thought about missing limbs in Doom, but now, after introducing realistic simulation, you will complain.

I do not think we can overcome those limitations with software, so we have to accept them, and design within and around those limits.
We can do an impressive simulation of physics, plus photorealistic visualization, etc., if we want. But we can not surpass the hard limits imposed by the fact that the player is not in our game for real. There is no way currently. Not even VR gets there, due to missing haptics, a sense of acceleration, and more. And there is no Holo Deck in sight. But that's what you would need to make your ideas work well, i'm afraid.

RmbRT said:
And I acknowledge that it's an annoying mechanic. However, permadeath in a roguelike is also annoying

That's comparing apples with oranges, really.

But maybe i'm wrong… : )

@JoeJ I have not yet made such a game and I'm not advocating for a ragdoll-based camera model. I'm saying standard FPS mechanics but without unlimited camera move speeds or acceleration. Just like games where you control positioning with a mouse also enforce limits to movement speed. Similar to what you said about head bobbing. I'm also not targeting VR. And yes, when speculating about nontrivial things, it's almost always subjective as to whether something is an appropriate opinion or comparison.
Certainly, such a setup will not work for VR if you can physically move faster than the character. But the same problem also stands for those VR platform things where you can physically run to move in game. If you're a world record runner, you still can't move beyond the max movement speed, I guess.

JoeJ said:
But i could for example mount the camera to the ragdolls head body, and map mouse input to head rotation. I have never tried this, because i'm absolutely sure it's terrible.

RDR2 actually does this AFAIK for the first-person view, at least for position if not rotation. I played it exclusively in first person and it's alright (the controls, game is great). There's a little bit of “lag” but it's not that bad. It seems like a more realistic control system than most FPS. The animation in that game is pretty incredible. It seems like a mix of motion capture used to drive physically-simulated ragdolls, kind of like it uses physics constraints to keep the skeletons aligned with animation, rather than driving the skeleton directly with animation. This way the animations never do anything physically incorrect, and allows physical interaction to affect animations.