Paid a Great Deal to be Terrible at Development Most of the Time
Let us define the business of software development broadly, as building working solutions and fixing problems relating to software as they arise in a development organization. That big tent covers quality engineering, devops, ops, the art of seeing when and how the building of new solutions can be evaded, random stable cleaning, and all sorts of other tasks that might not immediately spring to mind when thinking of the pristine software developer at his or her desk. It is my contention that if you are any good at the broader profession considered in these terms, and your talent and experience are recognized, then the further you advance in your career, the more time you will spend being an outright terrible, slow, poor developer - and while being paid a premium for it.
In any given organization there are always far more tasks to accomplish than there are people with the practiced experience to carry out those tasks. Who will make the critical updates to legacy Python devops scripts in a Java development group? Probably not the junior developers, but rather the experienced developer with a proven ability to pick up new frameworks and languages fairly quickly. He may not have touched Python in a decade, but he'll soldier though. For this and any similar task, he'll be terrible for a few weeks, then vaguely competent, and if the job lasts three months he'll be an informed practitioner, able to talk about many subtle issues with people who do nothing but Python development year in and year out. Skill and experience as a developer is precisely the ability to reliably become an informed, skilled practitioner in any branch of development given only a couple of months, and while building a solution that gets the job done at the end of the day.
If there is a lesson in any of this, it is that development as a profession isn't writing code and poking at servers. Those are just symptoms. Development is learning. We are all professional learners, and learning in this business is paramount over all else. The sprawl of knowledge and tooling, and the exuberant pace of change, has outpaced the capacity of any individual to be and remain a broad expert, secure in his or her throne of stasis. There are some futurists, such as Vernor Vinge, who believe it likely that as technological progress continues to accelerate, all professions from top to bottom of the tree of difficulty will end up looking like development in this respect. Schooling in its present very bureaucratic, lumbering form makes a poor introduction to a world in which everything changes on a yearly basis; it hasn't moved much past the original goal of turning out interchangeable citizen bricks for the factory wall, set up for a life in which the last job undertaken will be the same as the first. Something will eventually give, some profound shift in the way the world works and civilization orders itself. Schooling will change dramatically, or perhaps we will compensate by integrating more active exobrains more closely with our own processes: today Stack Overflow piped to the IDE via an expert system, tomorrow strong AI, visuals via contact lenses, and a neural tap to the world library.
Regardless of what is to come, there is the certainty that senior developers at the time will be paid well to be initially terrible users of many of its facets.