Wondering why the widget kept pushing the date further and further out as time passes lately, I started looking at how it was modeled. I also discovered Stephan Lagerholm's model (http://ipv4depletion.com/
), which seems to produce a more consistent and explainable result, and more interestingly, arrives at depletion dates significantly earlier than the Huston model that HE and NetCore appear to be using.
He also makes a compelling (to me, but note I am not a statistician) case here: http://ipv4depletion.com/?p=253
as to why Huston's model is problematic.
I wondered if you have looked at Lagerholm's work and have any insight into the relative merits of his model vs Huston's?