Gregor asks:
Who thinks global state lists for something like shape (already separated into 2D-shape, 3D-shape), texture, smell, color, are possible and useful?
Short answer: yes they would be useful for some purposes, but horribly limiting for others.
Long answer: this is a complicated question, and depends to some extent on your philosophical position and vision of the future!
Again, I think it's important to keep in mind that there are two broad uses/classes of described characteristics in good descriptions/keys. "Comparative" characters are the sort that we all aim to record in standard fashion for all taxa in a group/treatment - with respect to shape, this would be "leaves ovate", "leaves elliptic" etc. "Discriminative" characters are much more specific, and are used to discriminate two or more closely similar taxa, or to describe a distinctive and unique (hence usefully discriminative) feature of a taxon. Examples of these:
Leaves subspirally twisted and often undulate with a spreading incurved or recurved acuminate apex .............Andersonia latiflora vs Leaves not twisted or undulate, apex triquetrous or more or less cylindric, mucronate............................Andersonia aristata
and the following description of Acacia horridula "leaves":
Phyllodes crowded, narrowly semitrullate, broadest below the middle at the slight gland-tipped angle
(both examples are from the Flora of the Perth Region. If I'd looked harder I could have found wackier examples - Nepenthes would be a good place to look).
These shapes can't be adequately described in words, let alone using the IAPT shapes, and I'm damn sure they wouldn't be adequately covered in any agreed lexicon of shape terms.
And smell! There are two species of Caladenia (Orchidaceae) in Victoria that are best discriminated by smell - one smells like a wet cat, the other like a dry one!
The philosophical thing is that modern systematics is moving towards a position of decrying any description that's not fully comparative - the whole idea behind DELTA, for instance, is to produce fully comparative, complete descriptions and we tell our undergraduates to avoid at all costs describing some feature for one taxon in a monograph and then not mentioning it for others. The outcome of this is a trend towards the bland, generalised description that, while fully comparable with all other descriptions, fails to get across the essence of the thing (and of course, computer-generated descriptions are the blandest of all).
There will always be a tension between these two aspects of a description. I'd suggest that the reason that the fully comparative description, and with it the idea of lexical lists, is becoming so seductive is because we increasingly represent the world the way it can be represented in computers and databases - we're adapting to the machine rather than vice versa.
Sorry about the diatribe.
Cheers - k