You don't become a master teacher overnight. It takes years, perhaps decades. Year one, you're just trying to survive. Year two is frequently a small step back. Year three on is slow improvement provide the teacher works to improve. To me and eight to ten year teacher is usually an advanced beginner, fifteen years? Seasoned. Master teacher? You're probably pushing close to 20 years or more. Of course, there are exceptions but this is the pattern I've most often observed (burnout notwithstanding).
Why such a long developmental timeline? Part of it is the feedback loop. At best your cycle is one semester but more often than not it's a year. You do something once and then you don't really know if it worked until way later if ever. You also can't try it again until next year and even then, it'll be with different students and possibly with all sorts of different variables.
That's why I was happy when this semester ended and I felt that I finally got one of my classes to where I wanted it. It only took five years. Actually five years but having taught it 8 times over those five years.
The class in question is the lab/recitation for Hunter's CSCI13500 - essentially CS1. For the non-Daedalus sections, they basically do a programming lab per week. For me, my students do those labs albeit usually outside of class time and we do enrichment. One of the challenges is that I have to follow the lead of the main course instructor who designs the labs and controls the class sequence. I have some leway but I can't for instance decide to teach classes way earlier.
Over the eight iterations I've tried different different things but I think I've finally settled on what and how things happen in the class.
While I still follow the thirteen labs given by the mane instructor in sequence, they use Gradescope and I use GitHub Classroom. I also use unit tests to grade the labs which gives me a platform to develop the idea of testing both in general and as an industry practice. We go from nothing to discussing testing and me providing test cases to them designing their tests for later labs. Since we're using GitHub Classroom that allows us to leverage that as well. We'll talk about SE practices like using version control, build tools (make), code review, etc.. This is all part of the enrichment which also includes students working on projects started by other students.
We also do some direct CS enrichment by extending some of the labs. Two labs that have been used recently have particularly lent themselves to extensions. One was a lab based on rotational ciphers which I extend by covering a decoder and there's a 2D image lab based on arrays that lends itself to all sorts of extensions - I like seam carving.
This is what I landed on but I also tried other things that didn't work and some of this will change in the future. The point is that it took years. I couldn't do something one week and then immediately try something else. Sure, you can do that with specific teaching techniques but not with a class overall.
One of the other classes I'm still tweaking is CSC23500 - essentially data structures. I taught it for decades as part of APCS-AB but Hunter's a new audience - different circumstances, different constraints, and differnet advantages. First time through I wasn't super happy with my performance. Second year I made some positive adjustments and was happier - Covid mid semester notwithstanding. This year, I think I got the structure down. It was remote and during Covid so there were other challenges so I won't really know until the fourth iteration but I think I'm getting there.
So, all this is to say that improvement in teaching and improvement in classes doesn't happen overnight. Education is the long game.