>But solar isn't built like nuclear. Solar involves parallel exploration of device designs at very small scale, installed with massive redundancy and resilience. The real cost decline driver is manufacturing automation. Nuclear, even SMRs, have orders of magnitude coarser granularity.

Because the level of permitted development without being crushed by onerous regulatory burdens has been absurdly hamstrung on nuclear. All of the issues you add as "but" cases are things that many different innovations in a fluid market for research could have refined. The same has been done for many complex technologies over the decades, yet for nuclear there's always some excuse like the ones you mention. The comment you replied to is right. We're talking about something that since decades ago could have been improved enormously, and hasn't been thanks to a multitude of stupidities.

The United States Navy trusts extremely compact reactors (designed and working despite the DoD's notoriously lax financial and schedule stringency with defense contractors) to power its absolute most important, costly, defense-crucial war machines, and regularly docks them right inside the country's (and world's) largest urban areas, but somehow there's just no way to make nuclear power for civilian use more compact, cheaper and effective?

The regulatory burden argument doesn't explain why renewables are trouncing new nuclear in China. I view it as a universal excuse nuclear fans trot out to explain away inconvenient realities. They also never explain how the regulatory burden would be reduced in a way that doesn't compromise safety. And regulated safety is the price the nuclear industry pays for liability limits.