But precomposing all the potential combinations is less sane than the current mess (and you can outlaw Zalgo in the standard if you think it's a serious issue)

But precomposing all the potential combinations is less sane than the current mess (and you can outlaw Zalgo in the standard if you think it's a serious issue) Also, the % should measure people, not languages, that would greatly decrease the imaginary 99%