I think good points can be made in favor of not ignoring the context when (a) the science cannot be evaluated independently (so you have to take it on trust or

I think good points can be made in favor of not ignoring the context when (a) the science cannot be evaluated independently (so you have to take it on trust or ditch it), or (b) there are reasons to suspect publication bias (i.e., results are reported/published more if they fit certain biases). So yes for many parts of biology and psychology, but no for maths or much of the hard sciences (at least not the older stuff that has been repro'd many times). Speaking of "isolation from the past", the linked opinion piece doesn't cite a single reference other than the author's own pop-sci book. An interesting decision for sure.