This dance, shuffling code between backend and frontend, has been happening for a long time. It's not the first and not the last time someone claims to have found a solution (not even the best, just a solution for certain scenarios). That is fine until the client comes up with a new requirement that completely demolishes your assumptions, or some new awesome JS API appears, and then we're back to SPAs. Ups, who could have known, right? A couple years later, browsers adapt and a bunch of people, again, come up with a revolutionary idea: if we squint a little we can just about squeeze that code on the backend and save a bunch of work. Until new big products come out that change what users consider state of the art and...
It's a never ending cycle.
SPAs are a pain in the ass, like all distributed systems are, but they're also the most flexible and you don't need to reinvent half your framework or use dirty hacks if you have to do something slightly out of the beaten path. They're not the problem.
The real friction comes not from where you put the code but from the fact that you have to use different languages with different ecosystems. I'd claim most of the arguments against SPAs would suddenly disappear if one could write most of the code in whatever, compile to WASM, and slap a generic JS bridge on top.
It's a never ending cycle.
SPAs are a pain in the ass, like all distributed systems are, but they're also the most flexible and you don't need to reinvent half your framework or use dirty hacks if you have to do something slightly out of the beaten path. They're not the problem.
The real friction comes not from where you put the code but from the fact that you have to use different languages with different ecosystems. I'd claim most of the arguments against SPAs would suddenly disappear if one could write most of the code in whatever, compile to WASM, and slap a generic JS bridge on top.