Governance processes can be designed, and different designs can have qualitatively different outcomes. Therefore, we can understand instituting (or evolving) governance processes as a design problem.
Clearly, there are risks associated with any sort of change to an existing system. And it is generally prudent to not try and fix what isn’t broken, especially in a complex system that is not fully understood.
However in the context of blockchain networks, where the software and state are both forkable. It is the blockchain’s network effect (including its community and culture) that drives its utility. Effective governance maximizes the voice of the stakeholders and reduces the risk of exit, thereby securing its network effect. The argument is that more important than specific features like Snarks or Sharding (which may attract more usage and grow the network effect), effective governance process will have a greater impact. With effective governance the network can make better collective decisions about which features to adopt, or allocate funding to support or expedite development of key functionality.
On-chain governance is not inherently problematic, current approaches may not adequately represent users (versus coin holders) and may arguably not be desirable at all. Still, It seems likely that eventually the blockchain with the most effective governance will be the dominant chain, and it also seems unlikely that the process that we have informally evolved thus far are perfect. Therefore it seems prudent to keep an open mind and continue to have a public dialogue about Ethereum governance structure, rather than suggest that dissenting opinions sabotage the legitimacy of the existing process.