Early in my career I read "Peopleware" by Tom DeMarco and Timothy Lister. Between that book Brooks' "Mythical Man Month" and Humphrey's "Managing the Software Process" it seemed that there was hope for the software industry learning some necessary lessons and growing to become a true engineering discipline. Nope. Never happened. The industry standard, despite improvements in some areas, is still a farcical shitshow with little beyond lipservice to process, predictability or proper self-evaluation. I can only describe agile, as it is practised, more of a coping mechanism than an actual methodology. Indeed the methodology is embodied mostly in the infrastructure; issue tracking, version control, code review, continuous integration with as little methodology glue between them as required to produce output.

Modern first and second tier software management seems less professional, is contributing less and is generally worse than it was twenty years ago. The quality of the engineering and program managers, their training and commitment to their craft seems really low and is not generally valued. On average team level software management has gotten worse rather than better and, given what is expected and how it is valued, this shouldn't be much of a surprise. It is truly disappointing that what could have been a valuable and productivity enhancing role became so useless.

Things aren't going to change for the better though until the dust settles somewhat on the role of AI in software and systems development and we start again to consider how software should be developed in the 21st century. Maybe it is possible that with AI doing most of the low-level work that the focus will change to building and maintaining architecture and systems. Many programmers might become more like traditional engineers doing a lot more systems work than they do today and continuing to solve problems. Lord knows though it won't be today's software management doing this work; they have nothing in the way of skills to offer to the problem.