In his interview with Nick Chapsas (https://youtu.be/2uLGXe95kTo?si=uhErSaWoCbe87fl1&t=3552), Daniel Roth mentioned that he sees the auto render-monde as his personal "last resort" mode.
Hearing him saying this surprised me, because the Blazor team hyped up the auto render-mode so much when it was initially released! Roth mentions that he would avoid this mode because he "would have to think about both the concerns of Blazor Server and Blazor WebAssembly". But I'm not sure I'm seeing why this is such a problem. In auto mode, the Blazor Server concerns are only relevant the first time a user accesses the site and does not have the WebAssembly runtime downloaded. On subsequent visits/refreshes, only the WebAssembly concerns are relevant. It seems to me that for apps that have a lot of interactivity, and for which we don't want to delay the use of this interactivity by users (until the WASM runtime is downloaded), the auto render-mode is the best choice (if we don't want to use the pure Server mode to avoid the annoying SignalR disconnection issues).
But is there something I am missing? Should the auto-render mode truly be a last resort, like Roth says, even in a scenario like the one I described? If so, why?