I come from the windows world with a strong background in PowerShell, and this article perfectly described my experiences with Python.
I spent 2 weekends trying to get JupyterHub up and running with the dotnet interactive kernels. And it all came down to ensuring that the right packages got installed at the right levels. Between the system, conda, and pyenv. And this is not the first time I've run into such problems.
I know it said anaconda is the worst offender, but honestly I wish there was a similar solution for PowerShell. I love the self-contained environments. It makes experimenting so much quicker and easier. But there is a learning curve.
And he right. I got so frustrated trying to figure out pip vs conda vs conda-forge vs pip3 vs pipx. For someone who only casually delves in python, it can be real off-putting.
However, nothing to me is more frustrating than running into package XYZ updated and now package ABC won't load. XYZ now requires python 3.10, but ABC can only run on 3.9 and below, etc. I have rage quit more than a few projects over stuff like that.
So, as someone who only dabbles in python, my number one suggestion is to use requirements files and put version number requirements in them. And if your project has some out of the ordinary combination and you use conda, provide a brief rundown of how to install and enable it. Those few lines in your readme could make all the difference for python noobs/hacks like me.