I learned this lesson the hard way. But now I resort back to books and obscure sources for my stats training. It's slow and painful, specially when you're alone and there's nobody to answer questions about this weird aggregation function you're learning.

Or perhaps less insidiously, a lot of answers are out of date. I remember that when I did AI game programming, I wanted to build a behavior tree, so in a recent project I Google for that. I get tons of great resources for behavior trees, libraries, examples, etc.

Except there's a lot more on the cutting edge I should be considering. Maybe GOAP is the right fit. Maybe HTN. Maybe something else. Sometimes these searches turn up results that are too specific to the old way of doing things and don't lead me to the latest and greatest.

The latest and greatest?

If that matters put in whatever subject and a year like 2019.

Be warned the latest is rarely the greatest.

True, but this isn't really case for optimization problems and results about theorems or lemmas.

I think one of YouTube’s strengths compared to SO is that you can find a few useful creators with good information on a topic and learn a lot from them.

On SO it’s the luck of the draw on each individual question.

SO also allows you to browse all answers (and questions) from a specific contributor.

I so don't like when DDG presents me with SO links and previews at the top, instead of Github issues and official library docs. 95% is garbage.