It just follows from the rules.
The rules were concieved, back in 3.0, as, like "Rogues are the skillz class!" and they didn't think too hard about, what if a rogue managed to finagle Spellcraft or Knowledge: Arcane as class skills and max them out... or, for that matter, an NPC Expert could maximize those skills (in spite of no spellcasting ability) quite easily, since it picked its own class skills. 3e even coined, at some point, the term "Theoretical Thaumaturge" for a character with high Spellcraft and knowledge who didn't cast spells, at all. 4e & 5e, while they have simpler skill systems, did not back off from allowing non-casters to have caster-associated skills, so the idiosyncrasy remains. It's a system artifact.
Whether it's a legit way of modeling a world of magic, I suppose, depends on how you imagine worlds of magic being, and how you concieve of magic. Is magic just like, an alt.physics and scientists studying magic in a D&D world would be able to divine it's mysteries via experimentation (given cooperative casters to experiment with), or is magic something else entirely that defies such analysis?