Outstanding article. I encounter this all the time. My breakdown of the challenge:

* If the org isn't making changes fast enough or accurately enough when the logic is not configurable, what makes anyone think that putting in the massive extra effort to make it configurable will ever be done on time or right?

* Does the application logic change so frequently that this effort will ever be repaid?

* How do you test that the configurations made by the end user work right? If the actual IT org isn't strong enough to test logic well, what makes anyone think that an end user of some variety is going to go to a console, make a change, and it will as the end user thinks it will?

* The UIs and such needed to allow end users to make these changes tend to be so complex that no one actually ever uses them.

I remember a while ago I worked on a dashboard that allowed people to drag/drop widgets into place, we did some metrics a year after we released it and less than 1% of users ever actually customized it. It was a huge amount of code/effort to make it work (drag/drop in 2013 in jQuery was fairly miserable) for a feature that was barely used. Only reason we kept the feature was it demo'ed well.

Every time I've built a system like this, same stuff happened. It demoed well, was 10x (or more) the effort to develop compared to not making it configurable, and no one used it... and when they did, it never, EVER saved more work than it was to write it.

J.Ja