[zelphirkalt]

I am surprised, that there is no mention of concurrency or running things in parallel in the article. That is one of the areas, where you profit the most from avoiding mutating state. Was there any parallelization effort for the example project "Archetype"? If any, how did that go? What were the problems with parallelizing to make things run faster?

I think there might also be another idea to avoid too heavy updating in case of the many many functional updates to the 30K data, but it depends on how Archetype works and what it does. It might be possible to not apply some updates immediately, but group them together with other updates or simplify multiple subsequent updates into one update in clever ways. This might be difficult to implement though and as I said, might not be appropriate for all situations, as you might want to give immediate feedback in the GUI for any change.