I posit that in many software projects there is a small core of the “most interesting” work, surrounded by a larger core of “support engineering”. The “support engineering” is in service of the “most interesting” core in order to make it usable and a good product.
- Compilers: Core optimizations vs cli arg parsing
- Kernels: Core context switching vs some module that prints out the config the kernel was built with
- Audio software: Core engine, data model, or file format work vs UI work on the settings menu
- ChatGPT: Core machine learning vs front end web dev to implement the chat web UI
But the funny thing is that “interesting” is in the eye of the beholder. For every person that thinks X is the “most interesting”, perhaps most technical part of the project, there will be a different person that is totally uninterested in X and is delighted to let someone else handle this for them. This very well may be because X is too technical, too in the weeds.
The generalizes to work and society as a whole — people gravitate towards work that suits their interests. The areas they don’t find interesting are hopefully filled by others who are naturally wired differently. This does happen in real life but of course plays out less cleanly.