[T]he starting point for becoming a good writer is to be a good reader. Writers acquire their technique by spotting, savoring, and reverse-engineering examples of good prose. (p.12)
Although spotting and savoring good prose is certainly easier said than done, the advice is at least clear. But what exactly is it to reverse-engineer good prose? "Reverse-engineering" is a metaphor here, but it is not altogether clear what it is a metaphor of. Pinker does go on to demonstrate how to reverse-engineer good prose. He picks "four passages of twenty-first-century prose, diverse in style and content" and tries to "understand what makes them work"(p.12). These passages are supposed to be examples of good prose, and what he does is analyze why they are good. It seems that by "reverse-engineering good prose" Pinker simply means reflecting on and analyzing what makes a piece of prose good. If this is all there is to reverse-engineering good prose, I don't think "reverse-engineering" is an apt metaphor for the practice.
Reverse-engineering is taking apart an object to see how it works, usually for the purpose of duplicating the object. The object is designed, with components organized and connected in specific ways so that the different functioning parts work severally or together to make the whole object work. It may seem that a piece of good prose is analogous to a designed object. Some may even think that a piece of good prose is a designed object, and hence that "reverse-engineering" is not a metaphor here. There is no need to quibble over the expression "designed object", for we can grant that a piece of good prose is a designed object and still have a good reason to insist that reflecting on and analyzing what makes it good prose is not like, or is not, reverse-engineering.
Here's the reason. Although there can be more than one approach to reverse-engineering a particular object, there can only be one correct result of "This is how it works". And the resulting "This is how it works", if correct, has to be complete --- nothing can be added. If two people attempt to reverse-engineer an object and both succeed, they will agree on how the object works. Reflecting on and analyzing good prose is not like this. Two people can agree that a piece of prose is good, but after reflecting on and analyzing the piece disagree on what makes it good prose. Their analyses may both be correct, and both be incomplete. This is because these two people pay attention to or value different aspects of the piece given their different sensibilities, and they may discover different new good things about the piece when they reflect on and analyze the piece further.
Spotting and savoring good prose already depends on our sensibilities; what Pinker calls "reverse-engineering good prose" is a practice that helps us strengthen the sensibilities we have, or even develop new ones. Calling it "reverse-engineering" is misleading because the practice is not merely a matter of making discoveries, but also a process of changing ourselves.