Just for context, this was discussed ad nauseam a long time ago, and
essentially the argument is that a) there are way too many places
where (-1)^(1/3) should be in fact the primitive complex root of
negative unity, even when it is not explicitly asked for via using the
complex number framework in Sage and b) it's not clear that there is
any way to make the plotting code "know" that one is plotting
something like this, because there are too many similar things to
catch. I also believe Maple does not plot the negative values.
I wish there were an easier way around this, but none has occurred to
any of us so far. Believe me, I'd be the first to implement something
which allowed a) to work but which still caught cases of b).
- kcrisman