I still find libraries pretty useful while simultaneously being amazed at the quantity of papers online - I have found papers I would have nearly killed for decades ago now online at faculty pages of my favorite researchers. I'm guessing online is better for most, especially when coupled with browsing faculty pages for the author of your random linked papers. Also, look for citations listings that include authors of papers You like (it's printed and at your local research/eng lib for older obscure authors).
Before Eternal September and still today (though maybe less?), sending an intro email to someone that is studying a subject You are interested in is often useful (possibly better for niches). I'm guilty of not responding in a timely manner.
Probably Semi-OT: look at some indexes for CoEvolution Quarterly.
A million times this, and the value of libraries. There is a vast amount of information online, but the amount that isn't is still even greater.