Just some random thoughts, but I've started during the era of J2EE monoliths still being on the rise, when Web 1.0 -> 2.0 was only starting its transition, Python still being niche etc. Amount of insanely boring and inefficient stuff like digging through all the XML APIs and CORBA was self-evident.

It was fun to build totally new (and much simpler) tools, shortcutting quite literal man-years of work with each solution. Like rewriting half of some insanely expensive and over bloated million-dollar Oracle enterprise product into web app that we crunched out in a weekend over pizzas, and then demoed and validated with our clients before next week ended.

Over the years all this exciting new way of doing things has somehow evolved into what feels very much like the older dig-through-XML-schemas-for-hours world.

What I miss the most is the shared mindset of focusing on the problem, using simple tools that were build for purpose. That mindset was commonplace back then, at least in my circle. Maybe that's just a phase (cycle?) the industry goes through.