There's a defined timeline right now and even then, it's long. To open that wide open, to me, it takes the teeth away from getting this implemented. The potential is there for it not to be implemented. It could be 20 years before it's finally studied enough that it's implemented. I see your reason why there should be three years, but even that is too long. Removing that, to me, is problematic.
On May 2nd, 2017. See this statement in context.