One block converted from bespoke script to directives, measured before deciding whether to convert the rest.
a carousel block, on a page with one instance:
bespoke script 22.4 KB (gzipped 7.1)
interactivity runtime 12.8 KB (gzipped 4.9)
the block's own store 1.2 KB
one block: worse by nothing much.
four blocks: 89 KB → 18 KB, because the runtime is
shared and the per-block cost is a store.
and the thing that does not appear in the numbers: the
directive version works before hydration, and the
bespoke one did not.The runtime is a fixed cost that only pays off across several blocks, so converting one is a loss and converting four is a large win — which means the decision is about the page rather than about the block. Progressive enhancement is the part that does not show up in a byte count and was the stronger argument.