instead of arguing to stop custom solution and start building software and infrastructure as interchangeable parts, the article jumps to the very opposite conclusion
meanwhile, software as engineering actually exists and fly planes and bring spaceships to mun and beyond
the sad truth is that the customer often wants an engineered system but pretends it from college students paid in food stamps - well, you get what you pay for.
imho the first company that comes with an ikea model to the world of software - rugged, good enough, cheap - will dominate the next half century.
we're, at best, artisans and craftsman of software, but lets not pretend that's the trend. race to bottom will drive toward the win of mass engineering; we are actually already looking at this trend like now, where it is more and more reasonable to build an application out of existing services than rolling your own.
Just a decade ago would have been impossible in a year to bring about the same amount of feature today you can get for 9.99 off the shelf with the likes of intercom, keen.io, reveal.js, chargebee etc.
Software engineering is unfit for custom software development; that I concede; but it is just bound to happen as economy to scales drive do-or-buy decision toward microservices.