I post here to clarify the choice and record it in the post-finalization archive,
since I didn't find this issue raised in the mailing list.

srfi-1:           (fold cons '() '(a b c d))  => (d c b a)
srfi-13, 130:  (string-fold cons '() "abcd") => "dcba"
srfi-133:        (vector-fold cons '() '#(a b c d)) => ((((() . a) . b) . c) . d)

That is, srfi-133 passes accumulated value first to KONS procedure,
while other *-fold/*-fold-right passes it last.

As far as I can recollect, this is historical because srfi-43 used (<index> <seed> <element> ...)
and we simply dropped <index>.   Correct?