I think the classic definition of the singularity though (per Kurzweil's book) is precisely when the curve actually doesn't look flat on human comprehensible ti

I think the classic definition of the singularity though (per Kurzweil's book) is precisely when the curve actually doesn't look flat on human comprehensible timescales any more. I.e. One day there are significant overnight changes. Then the very next day hourly changes, soon thereafter every minute, second, millisecond, etc.