back when HDDs were a thing, and sequential accesses were much, much faster than random accesses. So people developed algorithms that maximised locality of ref

back when HDDs were a thing, and sequential accesses were much, much faster than random accesses. So people developed algorithms that maximised locality of reference, like B-trees. An interesting line of work in this vein is that of cache-oblivious algorithms :)