I don't make my own clothes, don't slaughter my own beef and don't distill my own gas. And I probably would fail miserably at trying to do so. Then again, I also don't bemoan that people lose the skill to do so.
The point was that it didn't matter in the "old days" whether you're curious enough so you'd want to do it. Back then, you simply HAD TO learn this to get things done. There was no way around it. If you wanted to program a game in the days of the C64, you HAD TO learn assembler since there was no other language remotely fast enough to accomplish what you wanted to do. And for the longest time, games and programming them meant riding the bleeding edge of technology development because only then you could actually squeeze those few last cycles out of the CPU or GPU and abuse some quirks of this or that chip to your advantage.
Advances in compilers meant that eventually we could dump asm in favor of C and C++ because those compilers were actually advanced enough to optimize sensibly, and soon their level of optimization beats the average asm programmer. But when I look at the development and how games are today written in C# or, like the Minecraft you used as an example, in Java, it mostly means that game programming now, too, has left the field of "bleeding edge tech" in favor or making it easier. Computers are fast enough that cycles wasted don't matter no longer.
But this also always means that certain skills will no longer be learned simply due to a lack of need. Which may be good, since it frees up time for learning something else, but it also means that some foundations are lost. Convenience always entails not knowing, or rather, not having to know something. Whether someone will learn it depends only on him wanting, not him needing, to learn it.
That difference is crucial. Some times are simply no fun to learn, but they may well be very important for understanding something and doing it the right way. Learning Big O Notation is no fun. None at all. But it is crucial if you want to understand (and not just accept as given) the advantages and drawbacks of various algorithms. And yes, you can simply learn that this algo is "better" than that one, but you will always be dependent on someone who learned it to calculate it for you.