Here’s a nice video to help you visualize the process of self-assembly:
As Kathryn Applegate comments:
sub-assemblies form and break apart en route to the most stable structure, the full capsid. As the sub-assemblies begin to form, further associations with free subunits become more favorable and as a result occur rapidly, while the final steps may take considerably longer. While the subunits in the model are rigid, in reality the proteins take on multiple conformations, allowing the capsid to “breathe.”
Self-assembly is a brilliant strategy to facilitate nanotechnology. In their book on nanotechnology, Mark and Daniel Ratner outline several methods of constructing nanostructures and conclude that self-assembly holds the greatest potential. The alternatives are nanoscale versions of bulldozing, rubber-stamping, and writing with an old fashioned dip pen, all of which restrict your output. The only serious alternative to self-assembly is Drexler’s notion of an assembler, but that does not look feasible for many reasons as explained in The Design Matrix.
Yet, in essence, cells do have an “assembler” – it’s called the ribosome. But the assembly-aspect of the ribosome boils down to creating peptide bonds between amino acids. It is the physico-chemical properties of the linked amino acids which then enable the chain to fold into the 3D conformations that in turn tap into the realm of self-assembly. Outstanding. In this case, the assembler itself self-assembles (although the order of mixing becomes important).
Of course, self-assembly fits very comfortably within the hypothesis of front-loading.