Entry 9

On the importance of feedback and our internal receptiveness to it.

Yesterday I received some negative feedback on the product stories I was writing and it was a treasure.

For weeks I had been asking for feedback and didn't receive any.

I had been writing super detailed (and rather low trust) stories. I would write up the core value propositions (most of the time anyway), po in a few screenshots and let my LLM, instructed to write in the pattern I chose, do the rest.

At first, this felt like a win. Stories were getting out fast, they were covering tons of edge cases and errors, some of which I wouldn't have thought of. As the weeks went on though, each story written this way felt worse than the last. Like, I really don't want to be specifying this level of detail for seasoned engineers. I suspected (but lacked the feedback) that if this was starting to feel bad for me it likely felt bad for them as well. I had even started doing a task breakdown of the stories, creating AI prompts since the engineers were using AI to generate and then scrutinize the code, such a crazy new world. Things weren't even moving all that much faster than I would have imagined without the AI. One story in particular stands out, it covered a complex screen and the AI created a comprehensive set of requirements which were incredibly redundant, made for one giant issue, and took me longer to pair down to something sensible than it would have writing the story from scratch. Still, I felt the engineers were using AI to generate the code so I kept in a lot of extra edge cases and error handling than I normally would have. Finally I got the feedback that all of these were making the stories super hard to read through (it was even hard for me) and making it take a lot longer due to the number of edge case covered.

I praise the feedback gods for this and I ask, why did it take me so long. I KNEW this style of story writing added complexity, added time, and reduced engineering autonomy. This was my experiment working through the new paradigm of AI and it felt really pretty darned bad.

While the feedback did sting for half a second (especially since I got it indirectly) I am super grateful to have received this gift of feedback. It allowed me to make real positive adjustments and move forward, so on to the next experiment, moving edge cases and error handling to sub-issues and actually prioritizing these.

Yay learning!!