The problem I see with this is that humans are really bad at estimating how long tasks are to complete. So while perhaps procrastination isn't inherently bad, in the world of poor estimation it means things don't get done on time.
"On time" would be any externally imposed deadline, and I would say it's completely the fault of imperfect estimation practices.
If we could estimate perfectly, then deferring all work until it needed would be uniformly good. But many times the true size of the work is unknown and we procrastinate on estimating, and we only figure out how much work is involved until it's too late.
