Where do you get, from the network effect, that "necessarily the thing that gets chosen to be scaled up cannot be optimal"? The network effect just means that whatever is currently scaled up is hard to displace.
In fields where we gradually refine our definition of optimality, we don't tend to be aware of the retrospectively-optimal solution until after we've put previously-considered-optimal solutions through their paces, and found their pros and cons.
The network effect, however, tends to kick in when a solution is "good enough," which happens on some random iteration possibly long before the "truly optimal" solution is found.
The only time you'll see an optimal solution to a problem spreading by the network effect, is when the definition of optimality is clear-enough from the start that the optimal solution can be found before people have a chance to settle on anything lesser.
