That was one issue that stood out about the claim that most NSF proposals are submitted shortly before deadline. Even if you have a proposal 95% complete month

That was one issue that stood out about the claim that most NSF proposals are submitted shortly before deadline. Even if you have a proposal 95% complete months early, there's no real benefit to submitting it, and some value to waiting - to polish, or perhaps to amend the proposal for a changing world. I've certainly seen professors go back and rework proposals at the last minute because someone else had published related work they needed to account for. The other issue that came to mind is that people finish work near deadlines with procrastinating, because work takes time. If there are two proposal deadlines per year, and the average person takes ~6 months to write a proposal, you'd see a normal distribution ramp-up of submissions, and never see the declining count because those people missed the deadline.