It's probably better if I just type out a larger excerpt:
"Unlike Beethoven's sonatas, but like his own song cycles, Schubert's
piano sonatas were not of a nature to inspire the need for public
performance for a long time. Sviatoslav Richter's comprehension of this
special intimate nature can explain his interpretation of some of the
late sonatas. His very slow tempo in the first movement of the last
sonata in B-flat Major (marked only Molto moderato) excited the derision
of Alfred Brendel. As I remember, Richter takes almost half an hour for
this movement alone, with three more still to go. Brendel was right in
thinking the tempo incorrect or inauthentic, but he also appeared not to
feel that the intimacy of the work was also essential to its
authenticity, and contented himself with a large-scale rendition. The
movement is indeed of grand dimensions, but the paradox of Schubert's
style here is the astonishing quantity of dynamic indications of
pianissimo and even ppp, broken most memorably just before the repeat of
the exposition by a single fierce and unexpectedly brutal playing as
loudly as possible of the trill of the principal motif, heard so far
only very softly (a repeat that Brendel refused to perform, perhaps
because the unprepared violence is awkward in a large hall, although
paradoxically more convincing in an intimate setting). Richter was an
extraordinarily intelligent musician: whenever there was a significant
detail in the score, it was always signaled by a reaction in his
interpretation, not always, perhaps, the reaction that one would have
liked, but no matter. For a recital in a large hall, he played the last
Schubert sonata on a darkened stage, with a single small lamp near the
keyboard, and the piano completely closed. The lowered light, the
muffled sonority, and the exceedingly slow tempo increased the sense of
intimacy, forcing listeners to concentrate to hear details. Even on
Richter's recording of this piece, I find that he was impressively able
to sustain the long line of the piece at the unnaturally reduced pace.
However, when he employed the same tactic with the G Major Sonata, where
Schubert himself does not control the line so intensely, the effect was
intolerable, as we waited impatiently for the next note.
"In short, there are innovative ways of interpreting the past, and they
can be magnificent even when wrong, but only if there is an awareness
and at least a mitigated respect for the original historical conditions.
There are also ways, of course, that are simply dead wrong. We have
reasonably objective criteria of judgment, socially and professionally
established, but no definitive ones. Some interpretations are
persuasive and may seem almost ideal, but every one always contains the
roots of a misconception or casts the shadow of a significant aspect
overlooked."
from "Old Wisdom and Newfangled Theory: Two One-Way Streets to Disaster"
in Charles Rosen's book of essays "Freedom and the Arts"