Randomness
"One of the advantages in thinking of randomness in terms of equipotential is that is allows us to bypass many of the tricky debates about causality in a rather clear and unambiguous manner."
In that sense it actually becomes an attempt to balance causality ( or confounding variable) across equally possible outcomes. As in taking a random sample; in order to make sure the groups are as similar as can be, randomization gives a high probability of fair distribution of the confounding variables.
" is concept far better suited to the laboratory than to nature: by design, it can only operate in the context of a stable, "
Right, I doubt there are situations, which are not orchestrated by humans, that have equal possible outcomes for all the variables being considered. However, I am reluctant to separate it along the lines of natural and unnatural, as I consider humans part of nature.
"It's a wonderful scientific tool, but a poor philosophical one."
That part I have to disagree on, as in a sense I feel science is an application of philosophy.