"From an engineering perspective it would make the most sense for evolution to coopt components already available for molecular machines like the flagellum. However, that is not anywhere near enough. Structures like the flagellum bear such a striking similarity to human made machines that one would also expect them to also have a very tight specificity in both the components and the assembly instructions. The components that make up the motor, the bearings, the universal joint, the whip, etc are interdependent on one another not just from the conformal standpoint but also in their dynamic characteristics (i.e. tensile and torsional strength, toughness, torque, rpm, modulus of elasticity, temperature variances, fits and tolerances, and on and on. At least if one was to design such a device there would be hundreds of design decisions to make and exponentially many more ways to foul up than to make it work.
... Then even if all the components were just right, there still has to be tightly controlled assembly processes that are also interdependent. When I think of all the tightly specific mutation sets required for all this, I'm just bewildered how it could come about without an intentionality of some sort."
Steve Petermann commenting at Telic Thoughts