Actually the 2 MB proposal (BIP109) is flawed by design and that is one of the problems. The grace period is too short (even Garzik agrees with this and he 'supports' Classic), the consensus threshold is too low, it doesn't provide a solution for the quadratic validation problem (it adds a limit/workaround to prevent the problem). However, the problem with Segwit is that people do not seem to understand it (which is normal, they don't really understand how the underlying protocols work either) but they're being hyperbolic about it. There is also that group that would reject a perfect[1] solution to scaling (right now; with 1 Million TPS without harming any part of the network (e.g. decentralization)) just because it was presented by Core.
[1] Assuming that a 'perfect' thing could actually exist (the TPS is rather a random example).
I think that these are very important points that are explained in a fairly easy to understand way.
In essence BIP 109 - (referring to classic, right?), is attempting to undermine bitcoin with at least the couple of unacceptable terms that you highlighted above - 1) too short of a "grace period" and 2) too low of a threshold for consensus.
I don't understand the point about the "quadratic validation" problem, but the first two issues of undermining consensus seem sufficient in order to cause the proposal to be totally and completely unacceptable.