You need at least two known values to be able to extrapolate unknown values.

This guenie slide uses just one value, 1 MB with 30 seconds as the latest time any node process a block today. Then just says lets extrapolate unknown values 3 MB, 8 MB with function O(n^2).

This is bad science, you need at least two known values like 0.5 MB and 1 MB with corresponding propagation time to all nodes in order to create estimation of the function for any unknown values like 3 MB or 8 MB. Obviously the more know values you have (over 2), the more precise function you can create and the more reliable extreme far unknown values like 8 MB becomes.

But the slide is big fail with just one 1 MB known value and presented function which comes not from fit with at least two known values but just from author willd guess, thus the obvious and eye catching absurd 3 MB and 8 MB propagation time predictions.