The turnover is high and the individuals are significantly less capable than their predecessors. I witness basic programming flaws that were resolved twenty years go show up because the programmers have no depth of experience. Companies pretend their staff can do the work but they're learning as they go. The clients end up paying the learning and the mistakes. Adding to the problem is that client companies don't have staff with enough knowledge to properly test the deliverables prior to acceptance and implementation of the program.
I'm disappointed but no longer surprised to see new programs go live without proper testing in parallel to the existing programs. The humans then have to deal with all the fallout. I worked at a multi-billion dollar company that had several dozen analysts depart after the company rolled out a very broken planning tool. These planners weren't going to be guinea pigs and beta testers for a program that couldn't even match the program it replaced, much less offered any improvement. They weren't going to have their salaries and bonuses impacted by failures in the new tool.
We had a lot of going-away lunches in that division. The company closed a few years after I left. Their computer problems were part of larger issue. The company wasn't sold or merged; the board closed the doors and sold off the assets because it was never going to catch up.
“I’m disappointed but no longer surprised to see new programs go live without proper testing in parallel to the existing programs.”
Exactly! Parallel systems.
I set up an online employment application program for a city around 2005. We ran the old system and the new system parallel for a month before all the identified bugs could be identified and resolved.
Also was involved in setting up 411 information for a state. Testing before going live was at least a month. We tested every which way HOPING to find problems. Then we were focused on DUPLICATING the problem(s). Deep dive into the system to (hopefully) work everything out before going live. And when we went live, still ran parallel systems for a month.
Seems that everything is seat-of-the-pants now. Just throw out a new system online and hope for the best. At that point, the users are the testers.