I would add that this isn't a new trend at all, at least for the industry as a whole. 30 years ago (geez, has it been that long...) I was developing embedded graphics code, and we always had situations where various developers wanted to use libraries of code for easy functionality adds and faster time to market, whereas other developers wanted to code everything in assembly, and not use any pre-developed solutions, for speed and code size. This was in an era where we needed everything to fit into 128kb, typically, so usually the assembly/hand code vote won out.
Most of my career was in the programming field, and I've seen this same trend all along. Frankly in virtually every case, additional functionality wins over dissecting, optimizing, and re-coding on one's own. Since that one instance of embedded code though, the functionality/get it out the door vote has always won the argument. It's a business, and using up an extra 26k here or there is far less important than getting product out the door.
For those not in it for a business, it becomes (IMO) largely a hacker's hobby if one really wants to know what's going on within a whole system. A small percentage of people have that interest, but most will be more focused on additional functionality rather than performance, code size, or other hacker's pursuits.