> I find it strange that he says we have no way of imagining what post-singularity would be like.
It's by definition - the singularity is defined as the moment when technology advances so fast that we just can't keep up with it. If you can still reasonably predict how will it look like, it's not singularity yet.
It's the moment where we lack the capacity to work at its level and control it. That doesn't preclude imagining outcomes. His whole article was about how he tries to imagine the full scope of possibilities rather than narrow down to predictions (that are often false). It's not about reasonably predicting.
As an example, we don't yet have fine-grained, ubiquitous nanotech like that featured in Diamond Age by Neal Stephenson. But that book is all about imagining what it might lead to.
We extrapolate from assumptions all the time, why is this one any different?
It's by definition - the singularity is defined as the moment when technology advances so fast that we just can't keep up with it. If you can still reasonably predict how will it look like, it's not singularity yet.