I meant that as they tune all the components, regardless of what those components are, which, according to the origami metaphor, can only be done as you go along (because of the interactivity, unless you're like some distributed systems wizard), they have a strong focus on creating a cohesive feel in the actual interactive part of driving. You do that by including the player in your analysis of the feedback loop that constitutes the game in motion: (filtered!) input -> all outputs considered together -> input again, regardless of the supposed realism of the underpinnings, and the player in question. That feedback step is critical, because it's all the game is to us. There are other practical things you can do, such as ensuring your team has real-world experience to call upon, no matter how limited, but that's useless if the team doesn't get to use it in shaping the game. It's a design decision, really, and that kind of mechanical cohesion is a characteristic all of PD's games share. And no, it's not something only PD's games exhibit, obviously. Blade of Darkness, Super Meat Boy and Wipeout are probably three of many good examples of mechanically cohesive, feelsome games, at least for my tastes, that just sprung into my head. None of them are particularly realistic, by design; I think it's significantly harder to achieve great feel and absolute realism together, though, and bad feel overall can obscure any realism in the parts anyway. A semi-recent-by-PD's-standards, stand-out example is Tourist Trophy; it's effectively GT4's physics on two wheels, and inherits all the lovely foibles of that game, so the "accuracy" is often perceptually "wrong". Crucially, they started with GT4, and worked forwards to get TT; what does that say about the physical underpinnings of GT4? That it is suitably general, i.e. physically informed, such that it wasn't too hard to adapt the formulation of the specific collection of simplifications etc. to two wheels (which Akihiko Tan did). If it was all empirical, reverse-engineered with no physical foundation, it would be like doing all of the chassis dynamics work up to GT4 again in less than two years on a skeleton crew. Despite the inaccuracies, for most of the time, the feel of the game is spot-on, the way it communicates the natural flow and rhythm of each bike, and the limits of adhesion in all axes (which you should stay within for minimal GT4-tyre-model weirdness), all informing your control inputs, giving a natural "riding style" not too far removed from reality (much like the original GT did over "arcade" racers of the time), is like no other game I can remember. If PD can also add realism to that, in the way they have for GT since that time, well... If PD suddenly decide to use someone else's parts, that hard-earned cohesion is immediately compromised and will take effort to regain, probably best by (re)building everything around that new part. But you inherit the systemic design that created that part in the first place, and that system might not do what you want it to do, unless you change the part itself. This describes iRacing's troubles with third-party sound engines, because its sound, physics and graphics engines evolved on their own paths aside from the majority of mainstream games, such that tools made for mainstream games are not immediately compatible. I think it's safe to say that GT has evolved on its own path, too, for better or for worse.