I think people are being overly pessimistic. It's not that ambitious of a project (particularly a hobby project at that). Why? Well, he doesn't have to support thousands of different drivers and hardware configurations -- he only needs code that supports his own. That eliminates a good chunk of the OS code.

Second, a lot of grunt-code can be found in open source projects, so most of the tedious/time-consuming programming can be eliminated if he chooses to follow this option.

Drop preemptive multitasking, cache-optimization, modes, virtual memory, and networking and there's not too much left.

And finally, it doesn't take that long to understand the GDT. Mine became corrupted once so I took a day to learn how it worked -- fixed that crap in a hex editor.

So no, he's probably not going to invent the next highly polished OS that handles every edge case and has been rigorously tested against bugs, but then again I don't think it's unreasonable to see a simple little functional OS.