I don't like the idea of using arXiv tags. While arXiv itself is great, I don't find its classification particularly useful. I would prefer to use standard tag system, given people do it precisely.
- arXiv tags are not equally distributed (by means of scientists, fields closeness, ...)
- arXiv tags are heavily influenced by the history of arXiv (i.e. evolution from hep-ph, through quant-ph to other branches)
- arXiv tags are rigid - they do not evolve in the same way as the interest of scientist evolve
- one or two arXiv tags means one or two less better suited tags
By 'standard tag system' I understand naturally emerging tags, instead for ones imposed from above.
Moreover, I consider mathematical tags one arXiv better organized, so the MathOverflow example on that matter is not that relevant (at least, IMHO).
This post has been migrated from (A51.SE)