When you're writing a bootloader and an operating system kernel, it's kind of necessarily true, because if you're a midwit programmer, your machine won't even boot. It comes down to what you consider "good quality code"; something that's robust, not too complicated and you can iteratively build a an operating system on? Terry's is. He spent ages pondering whether an architectural choice makes things unnecessarily compex vs. what you get in return.
In operating systems programming, you can have a file allocation table or a syscall index be a big ol' array of pointers and it can make complete sense from a simplicity pespective. In applications programming, if all your data is in one global 2D array, it's probably because you're a dumb fuck. You should compare Terry and Jason neither in their actual programming ability nor in the sensibility of their choices, because their development efforts are from different worlds. I would love to hear Jason pretend like he understands anything Terry is saying, though.