Can go into oversteer in transient situations, not at steady 60km/h.
By looking at the data the problem is mainly due to the rear spring stiffness (any other change would make minimal difference with such an imbalance in roll stiffness).
P.S.
I've driven al least five S2000s at any speed and lateral acceleration allowed by the (weak) engine and by the tires. Still alive.
Yeah, it probably is the main reason coupled with the rebound ratios. The stock car's split is very high, and it kept getting smaller over the years. Also heard something about "excessive drop throttle oversteer", but even then it does *feel* a bit much, even if it's basically exactly the same transient response like on footage. Footage can still be deceiving.
The issue is, why is the the R's roll stiffness split well within the margin if the spring MR or force is off? If we calculate the rear MR differently and end up with 0.700~, the % gets skewed 10%~ potentially, hinting towards much stronger rear sta-bar or bumpstops, which are both out of the ballpark. 2.3's split is 65%~ front and the behavior's not even right.
I *can* believe that my methodology is off, or that Honda's data is backwards, with the stiffness being 45% rear, or 55% front. Gotta take another look at the source itself closer; but it sure does say "front". Some handling characteristics on-power also don't seem to match up completely, but that can be explained by tires or manufacturer trickery as well.
Anyway, treating it as a traditional DWB with LCA/UCA attached strut (Which it isn't), we get an MR of 0.720.
If you do the math:
94873.1141657 / (94873.1141657 + 80406.3849233) = 0.54126760208. With a tiny bumpstop change we'd have 55% front, so this does support the other hypothesis. Only question is; why? Mechanically, it makes little sense. I'll need to calculate the MR with a 3rd method and see if a bumpstop change will bring it in line with 45% rear; if that's the actual value. Or someone could just go and measure it.
On quite a few cars, the current method has produced results within 0.025 of experimental data, so I just rolled with it.
A hypothesis I'm thinking about is also that this roll stiffness % is the ratio of lateral flex, because they talk about stiffening subframe and tower parts; and it's a total coincidence it matches up either way you spin it. Wouldn't be a first for me. Which is why I try to go for a more mechanical approach.
Note how it says roll resistance, not stiffness.
I'll take a look later, but this week it's unlikely I'll find the time.