I think limiting yourself to research papers alone will deprieve you of ideas which are just developing, which may not have an official body of research or even any published papers about. Ideas we as a society or culture might only beginning to tackle.
I find the discussion around a thing to be sometimes more interesting than the thing itself - a research paper might spark way more interesting discussions than the paper itself.
Take the "Attention Is All You Need" paper. How many have actually read it vs. how much has been written about it, or about the things it led to?
There is a lot packed into those 10 pages, but I've found the thousands of discussions around transformers and GPT to be way more interesting.
I'm not sure where the absolute statements came from - only research papers, etc.
> find the discussion around a thing to be sometimes more interesting than the thing itself
Here we differ. The discussions are mostly BS, mostly misinformation, ignorance, jokes, etc. It takes a lot of reading to find a few gems. What an expert writes, in their own domain, about something they've specifically studied in detail, is far more valuable IMHO, filled with beautiful things.
I don't rule out all discussion (obviously!). But think of the papers as comments in forums - or as blog posts - but instead of a misinformed hot take, the commenter did a bunch of research in the existing literature, carefully constructed an experiment, and tried out their idea themself - and furthermore, the commenter is an expert themself, like some people who post to HN.
It would still be great to learn where you find good material, of any kind.
I find the discussion around a thing to be sometimes more interesting than the thing itself - a research paper might spark way more interesting discussions than the paper itself.
Take the "Attention Is All You Need" paper. How many have actually read it vs. how much has been written about it, or about the things it led to?
There is a lot packed into those 10 pages, but I've found the thousands of discussions around transformers and GPT to be way more interesting.