When you are learning on your own you will be biased towards things that interest you more. But the boring stuff is as important in the real world.
You can interview a candidate that can impress you with fairly advanced concepts described in depth.
But you may feel disappointed if probing for gaps... Sometimes having gaps in certain concepts come with risks.
A simple example: If someone does not know how floating point numbers work (e.g: default number type in JS) for instance, they may want to use them for finance with disastrous results.
If a project calls for knowledge in theory - hire one person who knows that aspect, and can either handle it, or teach it to others.
The rest should be people who know their craft. Finding those is hard enough.
Who would you rather hire, someone who knows what a logistic regression is but has to look up the API calls, or someone who doesn't and tries to write their own classifier from scratch, but doesn't need to look up any syntax?
I don't mind new hire to not know API either way - senior programmers should be used to working with APIs they don't know. Tech changes constantly.
I think all of it is wonderful. Knowledge about software development is much more accessible than, eg, architecture. And much more aknowledged without a degree.
But, with the myriad of paths, come more risks of one choosing the wrong path, both from honest mistakes or being mislead by dishonest people. And these mistakes can be very costly.
I have a hunch that sometime in the future almost all knowledge will be able to be learned like this. It is a good thing to study the advantages and disadvantages of this environment.