> According to Eugenio Martínez Celdrán, however, systematic classification of Spanish allophones is impossible due to the fact that their occurrence varies from speaker to speaker and from region to region. According to him, the exact degree of openness of Spanish vowels depends not so much on the phonetic environment, but rather on various external factors accompanying speech.
So parent point stands. That's my opinion as well as a Spaniard.
The very fact that there are individuals and whole group of people who actually talk differently can, has been, and still often is dismissed as either ignorance, class, personal idiosincrasies, ...
There is a different way at looking at the thing: observing it objectively as other things in nature. This descriptive linguistics is endlessly fascinating to me, as it's less infected by human desires that have nothing to do with linguistics and all to do with politics.
An example of where those two viewpoints begin to present reality in a quite different way and yet don't require to be a trained linguist to grasp it is the definition of where a dialect ends and a language begins.