I started as a programmer with Chicago & Northwestern Railroad, and did the business requirements, programming, documentation and testing by myself. Today that would be at least 5 different jobs on multiple continents at least 5 times the cost, and they still wouldn't get it right.
After 2002 I couldn't find a programming job because they were mostly outsourced to India. Later I found a job in QA, a newly created position which checked for defects coming back from India.
I’ve been a manual QA analyst since then, because they think you forget how to program after a couple of years. Jason Christianson demanded that all regression testing be automated. His idea of 'automated' testing is a Javascript program that runs a job stream. ‘Automated Testing’ is actually a contradiction in terms. That program itself needs to be manually tested.
Also, ‘regression testing’ requires a comparison of ‘before’ and ‘after’ file results. That isn’t being done. The so-called ‘automation expert’ in my group couldn’t even write the Javascripts, she just ran them… manually.
A decade ago CVS was running 2.1 million claims through their programs, and regression tested by manually running the ‘baseline’ programs followed by the ‘changed’ programs, then ran comparators to determine that only the anticipated changes occurred. Since this was very time consuming, I suggested they automate the process.
I found a Headstrong contractor, and worked with him to automate the process in QTP. It saved a lot of time. A lot of the discrepancies were ‘date related’, meaning that they were caused by the different system run dates. I told them to reset the system date of the ‘changed’ run to match the system date of the ‘baseline’ run. This eliminated thousands of ‘false’ discrepancies.
I wrote a program at CNW using ‘Attachmate Now’ to automate their keypunch department, which I had mixed feeling about since a friend of mine worked there, but it worked and the keypunch department was eliminated.
I also wrote a program on the first handheld device called the Gridpad, developed by Jeff Hawkins. My program allowed the railroad workers to send in their worksheets without finding a fax machine in the small towns there were in.
They simply entered their hours into the Gridpad, signed it with the attached stylus, then plugged it into their motel room phone jack, and it was sent to the mainframe. Cutting edge at the time!