I don't think "reinventing solutions to common problems" is a bad thing. This is how we all learn how to do something new. And sometimes the new solution is better than any of the other solutions out there.
I agree that reinventing solutions to problems is definitely in the domain of the hacker.
When learning a new skill, however, you should understand what the common approach is before re-discovering the work of others. The article is about how to be more efficient in learning machine learning, not how to be a hacker.
It's akin to why (imho) better musicians learn to play other peoples styles before developing their own.
