Most of mechanical engineering is still tinkering. Not all problems are solved. Many of the people here pining for "real engineering in software" need more interdisciplinary experience, ffs.
I can only speak from my undergrad experience, but almost every engineering discipline took the same background courses, as well as mechanical, thermal, fluid, and electrical engineering classes and from there, went on to take courses related directly to their field.
From my experience, tinkering is another way to say prototyping or building a proof of concept.
As we've gained experience, the need to prototype the same solutions decreases, but when learning new tools and methodologies, experimenting and prototyping is very much part of the engineering process.
Engineering involves risk tradeoffs, and if your developers are not making your risk decisions then they are not engineers -- and also you're going to spend a lot more money producing software (the authority to do risk assessment should always be as close to the "in the trenches" work as possible, every degree of removal invites the risk decisions to be made on lower-quality more-aggregated more-gutfeel information rather than clear metrics).
Most of computer programming is instead tinkering -- build a proof of concept, put it under stress, see what is being stressed, buttress those parts. The fundamental idea of architecture, "let's add some structure in advance so that we can build a self-sustaining system and take away the scaffolding later," is only really seen today maybe in test suites or so? Those are the scaffolds of our day I think. We're still in very early years for computing as a proper engineering discipline. But at least we do make some simple risk calculations.