Serum-esque? Sure, selecting waveforms from a lookup table isn't difficult. However implementing clean sounding alias-free oscillators that can themselves be modulated at audio rate and so on requires years of accumulated domain knowledge. On top of that serum has all kinds of fancy features, such as the ability to import wave files and generate good-sounding wavetables from them, high quality filters, (look up zero-delay-feedback filters) complex modulation routing and most importantly, a GUI that lets you navigate all this complexity and actually compose music with it.
>you would have a lot more freedom using raw math over visual gui.
Exactly. Freedom is the enemy of creativity. Hence why the majority of max/msp users don't actually make music with it. The more layers of abstraction and choices there are standing between your musical idea and making actual sounds, the harder it is to ever realise those ideas. By the time you've finished messing around with all the technical stuff and making endless choices, you've lost the idea. Hence the endless appeal of the acoustic guitar and human voice combo.
It's certainly possible to make music the other way around, where you let the machines have the ideas and you simply guide them, but it's a very different thing. Using pure math and code really shines here, as well as modular synthesis equipment.
EDIT: talking of modular, Waldorf, the german company who pioneered wavetable synths in the 90s (you'll take my uWave XT from my cold dead hands!) have just released a wavetable-based eurorack module, and it's pretty extra-crunchy, no anti aliasing here ;)