Focus group: What are users looking for?

[hamogu]

I did a “focus group” interview with an astropy user today about what type of “learn” resources they have been looking for or would use. I’m reporting this here hoping that others will do the same and we can build up a better understanding of what people are looking for so we can prioritize where we invest time into learn.

I'm broadly asking the following questions: How do you learn about astropy (and other astronomy packages) when you address a new task? Do you know about/use learn.astropy.org? What type of information do you find in the docs and what type is missing or can be improved? Do you or would you use API docs / narrative docs / examples / turorials / notebooks / videos / lectures with exercises / other material?
However, this is not a questionnaire, instead I'm having a conversation to follow-up on suggestions or problems so I can understand and interpret better how they work. I plan on asking these questions to several astropy users around me to and will post summaries where when I have done so and I encourage others to do that same, since we move in different social circles and thus cover different use cases.

[hamogu]

Summary interview 1:

• This person has been a Python/Astropy user for several years and is in a classing research scientist position (no teaching).
• They are not aware that learn.astropy.org exists, and when I described the intent, asked “What is that good for? I thought everyone uses google?”
• They are missing tutorials that walk though a full science analysis. The astropy API/per-package docs docs.astropy.org are clearly necessary, but have too small steps (e.g. load a fits file, but then not using it for anything).
• It is useful, but not absolutely required, if those examples are fully maintained (like the notebooks in the current learn infrastructure), to be able to follow every individual step.
• However, quantity is also important. It’s better to have many, many notebooks and find one that’s really close to the problem at hand, even if it’s not 100 % repeatable with current versions of astropy. This person has used astropy for years and can look at the API docs and modify code, if, e.g. a keyword has been renamed.

My take on the last two points: So, we could have both high-quality, Astropy maintained content and a list of links/webcrawl/… just indexing/linking to existing materials on the web that might be out of date or not curated, as long as it’s marked which material is quality A or B.