Up to a point. But then, interesting work will need to be done with a large graph of small pieces of software, which starts to look a lot like a large piece of

Up to a point. But then, interesting work will need to be done with a large graph of small pieces of software, which starts to look a lot like a large piece of software. Delaying complexity is not the same thing as avoiding it; the former can be a reasonable approach, the latter is whistling past the graveyard.