1. Yes, purely because it would put Gran Turismo at the forefront of technological development and be a feature that PD could use as a USP at some point.
2. Also they would be giving us the technology to play around with.
3. Just because you think it's a bad idea doesn't necessarily mean it is.
4. I'm off to uni from 2013- 2016 so I won't be gaming much I think.
Unfortunately, it's not actually ready to be used for simulating real cars. There are obvious damping issues for a start, the lattice has a uniform rigidity (i.e. a single material) and it just hasn't been tested against real data.
The fact that the chassis is deforming means that there is a serious impact on the driving physics all the time. You'd have to match the dynamics of the lattice to the real car by using variable stiffness and damping (which I'm not even sure their model allows; at best the current homogeneity might just be a performance consideration) and make sure that any deformation is itself realistic.
PD will be working with manufacturer specs of chassis rigidity where available, and these could just plug into the suspension model, if they're actually a consideration in GT's physics engine. Most sims do it this way, because it's far easier to tame with just a handful of calculations, as opposed to a several thousand degree of freedom lattice, just for the chassis, and it's accurate because they're usually measurements. Given the FEM models the manufacturers use for pre-development are rarely "accurate enough", I doubt this crude, low-resolution, not-physically-based approximation will be anything like accurate without serious work.
With the lattice, the resultant rigidity matrix will be a function of both the individual spring and mass "rods" and their springiness and massiness (which needs to be spatially variable), but also their arrangement. A small change in the structure can cause a significant change in the result, so the designer will be chasing their tail a lot trying to get each number in the matrix correct. I'm not saying it can't be done, and I think these sorts of "complexity from structured simplicity" systems are fascinating, exactly the sorts of systems that dominate in our environment. It just needs to be done properly.
You could just model the "skin" this way, but that's not this method's real advantage, and PD are already working in that direction.
"Soft-body physics" will be
everywhere in due time, just not yet.