>The last thing I want to do with my time is read data science and ML blogs and newsletters which are guaranteed to be 95%+ crap that's either irrelevant to me, wrong, useless or plagiarized from something I've already read.

Are you me? This is exactly how I feel. I've given up on trying to find a good source of Data Science news or articles.

If we're talking about ML models (e.g. Clustering Algorithms and Metrics), once a year I'll try to select a few books published in the last year read them. They'll always be some new ideas and algorithms, and I'll get to refresh my knowledge on existing topics as well.

Throughout the year I'll usually hear (HN/LinkedIn/colleagues) about related books (e.g. Designing Data Intensive Applications) and read them.

That's more than enough to keep me employed and ahead of the game at my job. I'm not a researcher nor a Twitter/LinkedIn "thought leader", so most of what I need to know needs to be tested and validated and not bleeding edge.