Here is my, admittedly speculative, theory on the FFB and floatiness. I think it stems from R3E's origins as more of an arcade/simcade title, which is still in the transition to a full fledged sim (thankfully, getting closer with each update). That said, I only jumped onboard a few months back, after that transition was well underway, so I don't have any firsthand experience of what it was like in its earlier incarnations.
I'll probably get flayed for even drawing this comparison, but what the hell... And I'll preface this by saying that, overall, I like R3E a lot. But the FFB and what I (and some others) perceive as floatiness reminds me a little bit of the exaggerated behavior of arcade titles, such as Grid Autosport. I bought Grid Autosport for cheap during the Steam Thanksgiving sales, and have played it a bit, and I perceive some similarities in those regards. Don't get me wrong, R3E is (IMHO) much better in every way, including the FFB and visual representations of movement. But it does feel to me like some of the things that I and others have reported are subtly reminiscent of the FFB and visual dynamics of arcade/simcade titles, although thankfully not to the same degree.
That kinda makes sense to me, given its original focus. When developing a simcade, there is little reason to invest the time to develop accurate dynamics at a fundamental level, b/c that isn't what those types of games are about. So, in making the shift to a full sim, it seems likely there would be a lot of foundational work that would need to be brought into alignment with those new goals. A ground-up overhaul would take a lot of time, and they probably can't realistically put everything else on hold until it is done. My guess is that they are taking a two-step approach to these kinds of things. First, apply "Band-Aid" fixes to temporarily alleviate some of these issues, while simultaneously and/or subsequently making more fundamental revisions in bits and pieces as time and resources permit.
As anyone whose done any amount of programming can attest, it is almost universally MUCH more difficult and time consuming to make changes to core functionality retroactively, after much has been built on top of it, compared to what is involved in doing so out of the gate. So, I have a lot of sympathy for the difficulty of their task, and how much it must take (and will continue to take) to make such a fundamental shift. On the whole, I think they are making great strides, even in the few months I've owned it, and am confident they will continue to do so. For now, I'm pleased that the recent update got things to a point where my overall enjoyment outweighs the niggles I still have with a few core aspects such as those discussed in this thread.
That ended up being longer than anticipated for a bunch of pure speculation! Just bored at the end of the day, I guess. But I'm curious if others think there is something to it, or if it is just a bunch of bunk.