This was a very fruitfull essay to read, there is a lot to this essay, but it is still primarily remembered for one point. Consider the following two sets of prioritizations of important software features:
Do the Right Thing (MIT): correct, consistent, complete, simple interface, simple implementation
Worse is Better (New Jersey): simple implementation, simple interface, correct, consistent, complete
Lisp is prone to follow the MIT school of design, Unix and C follow the New Jersey school of design. The MIT approach is very hard to get right. The worse is better approach results in shipping products that become popular and then will improve. They’ll never be as good as the right design would have been, but they tend to win in the market, and eventually will become good enough.
My take is that he is right that the qualities that contribute to software being popular and high quality are different. The factors identified in the essay have not stood the test of time – for instance PHP has become very popular despite being a complex mess – but the underlying point remains. Things that become popular become successful then improve. Things that don’t become popular tend to die on the vine. If you look around, examples abound.