Entering the software world

My curiosity to apply machine learning methods to see what it looks like, when a computer starts doing some of the things I do professionally, keeps growing.  

The online impression is that applying deep learning methods appear to get easier everyday.  What that means is setting up (nicely explained here) the hardware configuration with the right gpu-s along with cpu-s and sufficient ram and then all the software libraries, is becoming easier.  It actually is getting easier and I can experience it.  Virtual machines are these wonderful magical things and once you have familiarized yourself with how they work and what they cost and how much time you spend getting them to work vs doing actual stuff, you go full circle and realize that the only real way to get work done is to build your own box (we grew up calling it a pc).  I haven't done that yet - waiting to reach cost parity (as estimated here).  

However, there is this basic universal activity in s/w engineering, called debugging.  It is almost like the entire s/w world implicitly works to self preserve the utility of this skillset.  To be very efficient at debugging code so that time spent figuring out why something didn't work is << figuring out how to make the code work for me, means becoming a programmer.  

I need to come up with the analogous activity in my experimental hardscience world to help correct my perspective.  I feel the analogy is to being able to see, hear, smell, touch so that an observation can be converted into an action using my fingers. In the physical world, I observe someone do something and I copy it, then modify it while copying and then imagine and execute something new. Like I just learnt how to make the perfect espresso for me, using a stovetop espresso maker, after several attempts since watching the youtube video.  

If I become a programmer will I still be able to make my experiments succeed in the first attempt?  Can I even become a programmer?