* TX malleability fix
* payment channels
* fix to signature verification DOS attack vector
* block size dictated by the free market
SegWit as soft fork should be rejected at all cost, it's pure evil. However, all the features in this ugly hack SegWit should be implemented independently of each other and elegantly code-wise.
The only reason Core devs mixed all these fixes together in the abomination they call SegWit is to obfuscate the codebase of Bitcoin so that they would expand their control over Bitcoin even more. And the worst part about obscure code is that it could contain vulnerabilities (possibly deliberately injected). Heartbleed of OpenSSL anyone? Do you really want it to happen to Bitcoin too? It's much easier to review new features that are implemented independently of each other. So why didn't Core add them as such? Because soft fork is so much better? Yeah right, backwards compatibility is false advertising because SegWit isn't really backwards compatible since it requires super consensus to activate. So does a proper protocol upgrade.
The majority of the whole ecosystem doesn't give a shit about that...the most prominent names wants Segwitt because Segwitt is technical better and gives more security.