> I don't know any other piece of music that has such a thick layer of false tradition on it like this one.
As well as taking issue with the name - deriving from the poet Ludwig Rellstab which has "stuck like glue" - he notes that the manuscript for the slow first movement, adagio sostenuto (slowly and sustained), is marked alla breve[2] meaning "count two in a bar" so that each beat counts for a pair of the famous triplets[3] - effectively doubling the tempo:
> If you would play it in this tempo that is usually played [plays a bar at conventional speed] you can have breakfast and lunch and dinner and the poor pianist is still playing this first ...
Not content with this disruption of "Moonlight" convention, he relates that the manuscript writes senza sordino, instructing the player to play the whole first movement "without the dampers".
Despite the protestations of his colleagues that "you cannot do it on the modern piano":
> I beg to differ. Why can't you do it on the modern piano? Have you tried it? They said "No I haven't tried it. But you cannot do it!"
> It's an argument just not good enough. Beethoven is a great great enough composer that you have to take him very seriously. [...] He wanted a very special sound, that the harmonies swim together like in a wash and the overtones are strengthening each other.
His solution to the "modern piano" problem is simple: just partially raise the dampers by pressing the pedal down "about a third".
All this might sound like a recipe for disaster, but his interpretation sounds wonderful!
A lot more follows, and I can also recommend his "Moonlight" masterclass[4] as a demonstration of the cogency of his position on the piece.
1: https://www.youtube.com/watch?v=fW_Dv_GNQAo
2: https://en.wikipedia.org/wiki/Alla_breve
Also bear in mind that Beethoven was obsessed with the idea of variation: he took a theme and knew how to run with it. So in his later years, I expect he was also open to much more experimentation. One of his great themes is in the opening of the 2nd movement of the symphony #7, which one of my favourites of all my many favourites of this exquisite human being.
Similarly, the syncopated harmonic rhythm in the final movement of the last Sonata #32 (opus 111) doesn't throw you off as much when you are reading the score and have the bar lines to reinforce where the measure boundaries fall.
I speculate that if Beethoven had been able to hear those pieces played aloud, he would have made adjustments to them. I hear similar glitches in much of late Beethoven, but those two sections from those two sonatas are the ones that stay with me the most.