Importance-wise, delays in the singularity are trivial compared to a risk of it not happening at all.

So suppose we are worried that a die-off in industrial nations will actually make it impossible. Isn't this risk (that there is a major catastrophic disaster for industrialized nations and this disaster prevents a singularity from ever taking place) much smaller than the risk that the singularity goes badly when it happens? If so, then this supports the spirit of my original comment: worrying about the singularity going well is more pressing than worrying about major shocks to civilization which do not completely wipe out humanity.