MissScarlett: <what's up with 30 Bg8?>
It blocks the eighth rank, of course, allowing White's rook to enter.
Harding in his Blackburne biography (2015), p.19:
<30.Bg8! Good enough; in a blindfold game Paulsen could hardly be expected to find 30.Re7!! Qxe7 31.Be6+>
Quite, and yet it's always instructive to look at the difference between the great and the good.
After 30.Bg8, Black's only realistic try is 30...Bc6, covering the d7 square.
Then we have the line <31.Re7 Qxe7 32.Be6+ Qxe6 33.Qxe6+> and now 33..Bd7 means that the Bg3 will fall.
Without the interpolation of Bg8-Bc6, 32.Qe6+ can't be met by ...Bd7 and the Bg3 will live on.
Interesting but academic, you say. But more importantly, what does Harding say?
After 30.Bg8, he has: <If 30...Bc6 31.Re7 Kc7 32.Bxf4+ etc.> But ...Kc7 isn't even legal, so Harding's cocked it up. This error seems to have evaded detection, so far: http://www.chessmail.com/research/B...
Even so, there's another consideration - after <30.Bg8 Bc6>, White's only winning line is with 31.Re7. If Paulsen could hardly be expected to find 30.Re7!!, why should he find 31.Re7?