While I quite agree with the idea, it rests on the assumption: "leaders also need to create a culture in which employees do not have every waking moment of every day committed to project work" Unfortunately, I am currently in situation where is lot more project work than people (and I wonder, is it different elsewhere?). In that case, I don't think the approach the author describes quite works unless you have a dedic…
There are some, who in ambition to lead (or work on highly visible projects), will step up and take on new things – meanwhile skimping on their core responsibilities and deliverables. Are these the people we want to promote into leadership roles?
I don't have a problem with people working on new things. The question of leadership - I think it's a little fuzzy. OS kernel decides what process gets to run and what not, but it's not like you can call it a "leader". He just does it on basis of processes' priorities, which may well be set by individual processes themselves or something else entirely. (And I am actually fan of direct democracy - where the leader designation is pointless, but that doesn't mean - in such system - you can't have project manager telling you what you should work on, as is obvious from the OS analogy.)