My bad.
Neven's link on ASIG doesn't go to the page with the explanation. The loss rate is just as alarming for 15%. Do they have a numerical archive? It looks like DMI lost 1,000k in a week. If true, and not a glitch of some sort, it ratifies my contention that there is no speed limit for extent loss when there is a low CAPIE and parallel melting. Where there was one when the melting was limited to the edges.
Vergent