Thread · 4 tweets · 12 Nov 2020

Except LaTeX isn’t universal, it’s about 15% of papers — most of the rest is Word. Both Word and LaTeX suffer from being rendering formats rather than semantic, which is what you’d want for this to work. Not to say this is impossible — I know it can work because I built it once.

But getting authors to switch to better formats is nigh impossible (especially in eg. physics where people think that LaTeX is actually a good format and make fun of Word, even though they’re both pretty terrible). So that means processing a standardised form of some sort.

If people are willing to give up their local LaTeX tweaks, a standard LaTeX-alike should be possible. The same might work with Word. Both would work better if you’re willing to spend some time tweaking the submission in the tool (because neither is good at semantics).

From there it should be possible to output JATS (maybe JATS4R) and be more or less in business — for rather generous values of “more or less” given how pigheaded publishers and journal editors can be about their personal little preferences.