Kay waxes Hegelian here, but I wonder if there's a knowledge component as well: when one builds tools properly at level N, one learns which parts of a problem are best addressed at that level, and which should be handled elsewhere, but that learning involves an opportunity cost, and if that time could have been spent learning something more directly useful to one's project, it's better to use existant tooling, no matter how suboptimal.
... in programming there is a wide-spread 1st order theory that one shouldn't build one's own tools, languages, and especially operating systems. This is true—an incredible amount of time and energy has gone down these ratholes. On the 2nd hand, if you can build your own tools, languages and operating systems, then you absolutely should because the leverage that can be obtained (and often the time not wasted in trying to fix other people's not quite right tools) can be incredible.
— Alan Kay, VPRI Memo M-2004-001 6