> When an engineer isn't asked to think about the ROI ahead of time, they are left free to come up with novel innovations.
In the cases I've seen, given free reign the engineers were happy to focus their effort on their personal interests, such as writing everything in a functional language or developing byzantine microservices + CI/CD.
Seems quite presumptuous to assume those choices were not adding value. Would you be able to tell if they were?
Most importantly, as mentioned in the article, engineering teams need (and want) clear business goals. There is a huge chasm between having an expected ROI vs having no targets at all.
The very complex and ambitious solutions they engineered, are currently a nightmare to maintain, as it's hard to hire people who can understand them (thanks to the technologies being very new and/or niche). Meanwhile, the requirements were bog standard and there's no demand to be scalable, to have high reliability etc. The system could probably be coded by a bunch of interns and grads in whatever language they fancy.
In the cases I've seen, given free reign the engineers were happy to focus their effort on their personal interests, such as writing everything in a functional language or developing byzantine microservices + CI/CD.