I do love the way you describe the ffb. It makes me wonder if we might be getting to the point where a sim doesn't need a large helping of manufactured difficulty to satisfy the sim tag. I think as we get in to true sim territory, driving will become easier, because it will connect and extend from each person's real life experience. At the moment it is a exercise unto itself.
Makes me think of the Euphoria engine and how even the exact same input (eg. certain push on a shoulder) will yield minutely different results, because that's just kind of what happens in the real world. I suppose until utter perfection can be achieved in physics detail and calculations, somewhat deliberate imperfection might actually be closer to the real world. I can't even begin to fathom the number of variables that would act and interact in the real world.
Truth be told, I actually already use the experience I gain from real life driving in pCARS and visa versa. To be honest, I do drive very sporty on the open road with my Lancia Delta HPE (1999 model), which is a FWD car, and whatever I try in pCARS I also tend to try when safe in real life.
For example I try to compare the feedback I get from cornering in real life to that in the game. This led to Andrew Weber (Physics guru) throwing a few tests at me as well concerning for example tyre sidewall buckling and riding on the edge of the tread.
So actually you can already apply real life knowledge to the game and back. Besides, we community members have to do so because comparing Physics to last/current gen sims isn't the right thing to do. Now we can't all drive a Lotus 89T for example, but we can push our own cars to their limits to see how they react and see what we can apply from that experience to the game. While every car does drive differently, certain queues you get are in fact similar. Else we would always have to learn to drive each car all over again from scratch in real life as well, which we don't. We all can easily jump from a VW Beetle into a Porsche 911 and drive off.
How does applying such experience with our family trolleys apply to a monster like a Lotus 89T without making it feel like well... our own trolleys. lol
Simple... pure technical data about the cars we can get from A) The Manufacturer, B) Technical Books and (Haynes) Manuals, C) Magazines and Interviews, D) Racing Teams and Owners... etc.
Everything you can imagine has already been set in stone. Aerodynamics of the chassis, wings and undercarriage, geometry of the car and linkages, size of the rims and tyres, gear ratios, spring rates, fuel consumption, air fuel mixture, the bore and stroke, amount of cylinders... etc etc etc.
Those cold hard facts should in a Physics system based on causality (action->reaction, cause->effect) be enough to reproduce the character of each car, and even each track.
The biggest problem lies in the things that aren't a nice constant, mostly tyre related. Think about the contact patch dynamics, tread dynamics, carcass deformations, sidewall rigidity, flash heating, heating per layer, rim heat, brake hit heating the wheel arch and everything in it.
There are just so many highly dynamic things being calculated at once, especially with the flexi tyre model we have now, that it took us (devs and community) almost the entire project's development time to sort out what actually does what... and what action in one part of the tyre causes a reaction in the other.
In other words, SMS is trying to re-create and make sense of the black voodoo magic that is actually the rubber tyre.
The rest of the Physics changes have been more from going from a linear style approach to a causality style approach in handling things, where possible (we still haven't got a supercomputer at home) and most important.
The only other aspect, choice, is implemented by you the driver with linear or binary inputs. Steering, throttle, brake, clutch and shifter (SQ or H). After each adjustment to your choice on the inputs the Physics system of pCARS will go from that action into a long sequence of continues actions and reactions.
The fact that current Physics systems have been an exercise by itself is mostly likely because most systems do not interact and/or acted in a linear fashion. Something that is contrary to nature itself. After all, once you see something in nature that is straight cut you assume it is man-made for good reason.
Hence all other games, knowing pCARS, to me feel far too canned and directed by the game developer. The cars and environment often do not feel alive.
A reason why people raved about laser scanning tracks. Even if it is a snapshot of the track, the constant flux of nature acting upon the track's surface makes for that the shape, size, angle from every bump, curb and so on differs (even if it is by a tiny amount). This in return makes the track more organic and adds that undirected action into the mix upon which the Physics of the car have to react. However, if the Physics systems do not have a direction relation to one another, there will be little to no behavioral changes to the car. Even if the track surface itself looks and feels correct, the car will not react completely in the way you'd expect it too; making it an exercise by itself, especially with the minimal output from the game to your senses.
All in all, driving in pCARS is easy. Driving a car on the limit though... that is a different task. Personally I find humps the most treacherous of all, as you can't sense (by that elevator feeling) how far the car is lifting off of the ground at certain times. Such a thing just can't be reproduced in a simple home sim rig.
So even with perfect Physics in a game... if you can't sense it than you can't react to it either (causality link missing again).
Anyway, a perfect Physics engine is one that is based on causality in which you with all your senses will be connected. Maybe it sounds a bit like the Matrix in a way, but that's just the way it is... in my opinion at least.