But something else you wrote brought to mind this rather interesting statement from - of all places! - the introduction to Michael Barnsley's famous book, 'Fractals Everywhere':
'In deterministic geometry, structures are defined, communicated, and analysed, with the aid of elementary transformations such as affine transformations, scalings, rotations, and congruences. A fractal set generally contains infinitely many points whose organization is so complicated that it is not possible to describe the set by specifying directly where each point in it lies. Instead, the set may be defined by "the relations between the pieces." It is rather like describing the solar system by quoting the law of gravitation and stating the initial conditions. Everything follows from that. It appears always to be better to describe in terms of relationships.'
So, rather than condescending, I find this interesting: these are two dissimilar approaches to solving problems.
Your comment about baking an apple pie 'without being able to state any specific steps how to do it but instead describing the pie as some set of interrelationships of the ingredients' was also an excellent statement of the differences between these points of view.
Barnsley's perspective was that relationships may indeed be a better way to describe structures. His concern was fractals, but perhaps the analogy extends to the logical structures that comprise a program.
You can claim that it's interesting that there are these two approaches, but the quotation you gave was definitely condescending, because it flatly states that one approach is better than the other, despite the widespread success of the latter. The argument that it's good to abstract as much as possible from the machine is interesting, but to me it is by no means obvious.
I don't really see the relation with fractals. Although it is an interesting quotation, the relation of fractals to real world phenomena is often very dubious.
All functional languages can express a sequence of steps. More yet, on most of them you can later decide to transform the sequence, instead of just generating code from it.