The context was of speeding other computer operations by removing data.

With a sufficiently robust indexing system, file search time is independent of corpus size. E.g., Google isn't scanning the Internet each time you submit a search query. Time to build an index increases, but not searches against it.