No. 3 is, for all intents and purposes, impossible at the marked tempi. But the other four don't really rank among the super-difficult modern/contemporary works. 4 and 5 are actually pretty playable; No. 1 is stupidly hard in places, but nothing like No. 3. No. 2 is also not bad, because all of the leaping around is very repetitive. De la Nuit is harder than any of the Sonatas, other than No. 3.
I'd agree with this. What you've posted is Damerini, n'est-ce pas? How do you think that Hodges' performance compares?
As far as pieces that actually have the word "Sonata" in the title, Opus Archimagicum is the hardest that I can think of, too. But besides being a silly thread for the obvious reasons, it seems even more silly to pigeonhole ourselves to "things called a Sonata." That term has basically had no meaning since the turn of the 20th century, which is of course when all of the most difficult music was written.
I wouldn't say that it has "no meaning", since there are plenty of works with that tile written since around 1900 that are recognisable as "sonatas" on the basis of what we understand that term to have meant from, say, Haydn to Brahms, but I do of course agree that its use has inevitably become much looser than it was before then - and Sciarrino's examples illustrate this fact as well as any. Sorabji's last sonata (known as his fifth but actually his sixth, as there's an unnumbered one from 1917 which is his first known extant piano work) breaks the bounds of convention in terms of its sheer scale and it's interesting that he abandoned the term thereafter and all his subsequent large-scale piano works ae either called symponies or something else.
Specifically regarding Sorabji, the two works of his which I personally consider the most difficult have been performed by Powell, now (Solo Concerto and the Super "Dies Irae" w/e variations, whose title is too silly for me to ever remember).
It's not silly; it's just long (like the piece itself). For the record, it's
Sequentia Cyclica super Dies Iræ ex Missa pro Defunctis in clavicembali usum - i.e. cyclic sequence on the Dies Iræ from the mass for the dead, written for piano - but of course it's usually known simply as
Sequentia Cyclica.
But of course the most difficult pieces overall have come out of the New Complexity school. Wieland Hoban's "when the panting STARTS" is definitely the most difficult piece I have ever come across by a long way. Downie, Barrett and Finnissy wrote some quite ridiculous works, and the piano part in Emsley's "The Juniper Tree" is ultra-difficult. Not from the New Complexity school, but Xenakis and Bussotti have also written some totally-unplayable-but-still-plausibly-attemptable pieces. There are a slew of others maybe a half-step down in difficult from those, but I won't unleash a laundry list.
This is true up to a point, but only, I think, in that there are quite different difficulties in these works; I can still imagine pianists (albeit not that many!) who can develop their facilities sufficiently to present decent performances of
Tract, Downie's Piano Pieces, Finnissy's
all.fall.down / Piano Concerto No. 4 /
EC-T, Evryali or whatever who'd still struggle to cope with Sorabji's Piano Sonata No. 5,
Sequentia Cyclica, Piano Symphony No. "0" or Symphoic Variations, partly (though not entirely) because of the sheer resources of stamina required for these monumental works. All of those others are very short by comparison, but I suspect that even Finnissy's
A History of Photography in Sound which, exceptionally, is of Sorabjian dimensions would not present quite such challenges to the pianist as do any of the Sorabji works that I mention here.
By the way, the works of Sylvano B remain an utterly closed book to me; no doubt that's down to my own density, but so be it, I fear!...
Best,
Alistair