I was an applications programmer from the late 60's to the middle 70's when I moved on to a life of Systems Programming.
I have been there trying to debug mish mosh programs like the ones talked about in the article. They are NOT fun to debug in fact they were my worst nightmare. I was a typical programmer and I was called in in the middle of the night to try and fix these "bastards". Even with decent comments trying to figure out how you got there is often not easy. The dump you have is a point in time failure and just trying to figure out where in the program it failed is at times iffy. After complaining for years about "sphegetti" code management decided that structured programming was the way to go. Nice but it took too much time and deadlines went out the window. Then they came up with another name (long time forgotten) and it was semi workable and reasonably easy to code. We started to implement it and I will admit it was reasonably easy to debug and much more it was easy to figure out how you got there with a minimum of fuss. About that time I left programming for the wonderful world of systems and never looked back.
When I coded I used the philosophy of minimum branching and lots of comments. When somebody had to debug my code it was easily done and I got a few thanks along the way.
One time I coded up a fix for an IBM part of the OS and it took 1/50th the amount of time to run as the IBM version. I sent in the source for my fix and it was completely ignored and as of today 40 years later the program (or variation of it) is still running on the latest Z/os systems IBM is offering.