Running "brew/pip install datasette" isn't much help if you don't have a SQLite database already. So now you need to learn to use "sqlite-utils" or similar too to create one... by which point I could absolutely see most people having dropped off already.
I had a feature idea relating to this a while back which I should bump up the list: https://github.com/simonw/datasette/issues/1160
I opened a new issue based on this conversation here: https://github.com/simonw/datasette/issues/1845
Generally, the first-run experience is really important, because you may be getting lots of people in the funnel, but if you're losing them to complexity before they can get to the "this is nice" moment, it's for nothing.
Even something like finding plugins can be hard for first-timers, and you unfortunately don't hear about all the users who drop out, because they just stop using the software and never talk to you.
Maybe, though my issue hasn't been so much installing it as getting it to do useful things.
Your use cases sound great, thanks. Especially "Snowflake for small data" sounds great, I did want something like a simple PowerBI for my small SQLite database. If I could somehow create a dashboard, or some views, or something, and publish them along with my database, that would be amazing.
Plugins do sound good too, but when I download Datasette I already have something in mind so I already want it to come with the most common things I might want to use, along with examples on how to use them.
Vega was the plugin I found as well, but I remember it took me too long to figure out how to make a simple graph, so I quit at that point. If it were built-in, along with a bunch of other plugins, and I had an example of how I could create some simple graphs, I would have gotten much farther along, I think.