I think these kinds of articles are a bit glib. Developers implement architecture to solve problems confronting them. Sometimes a new problem then arises which must be dealt with, and so on, until the architecture in hindsight is quite complex. But you can only know this in hindsight. A few companies didn't run into the "piling on" of issues to be fixed, and so look back in hindsight, see their simple architecture, and think, "we know something that everybody else doesn't," when in fact they're simply experiencing some form of survivorship bias.