Since 1979 there is a sea ice extent record for the Arctic with mostly daily measurements. The trend is clear and down, so what is the problem?
Imagine a single idealized ice floe: 1 kilometer long and 1 kilometer wide with a thickness of 2 meter drifting in the Arctic Ocean - a pretty flat cuboid. Roughly 20 centimeter are visible above sea surface as freeboard. NASA’s Arctic Mosaic at highest resolution would represent the floe with 16 white pixels arranged as square.
Let’s further assume bottom- and top melting are equal and and there is enough energy available to melt 1 meter of ice from each of the six sides within a month (30 days). How much ice is left after this period? Exactly, zero.
Buoys or satellites measuring freeboard and calculating thickness will tell you the melting rate was 6.667 cm/day. No surprise here. But have a look from an orbit at the development of the area of our ice floe. We started the thought experiment with an extent of 1 million square meters. On day 29 there are still 0.996 million m² and one day later - day 30 - all disappeared without a bang and extent = 0 m².
Here’s a simple chart:
You want to know when the Arctic gets ice free? Make sure to look at the right thing and in any case include thickness. Of course this experiment would look completely different if ice floes were perfect cubes, but those exist in your freezer - not in the Arctic.