A hard fork is the only moral way to make serious changes to the Bitcoin protocol...
plus fucking one.
Here's mine:Minus two.
Segwit on it's own solves the transaction malleability problem, thus enabling payment channels and true scaling at Layer 2. 2mb, like segwit's ~1.75mb, is also a "slight boost in transaction capacity." I'm not sure how you've convinced yourself there's a huge "far more effective" difference between their tps increases.
Segwit has quite a few benefits, while a 2 MB block size limit has none (we can exclude the TPS increase as the both share it). I don't understand why this is so hard perceive. The only people that call out Segwit on its complexity are people who don't really understand code or are half-baked coders. Whoever I've asked previously (as I don't do C++ myself) said that the complexity is overblown by a 'certain group'.