I bought The Shadow of the Torturer when it came out in paperback, I guess in response to a positive review. I found it kinda difficult to read, but I wanted to know what would happen next, so I bought volumes 2, 3, and 4 when they came out too. By the time I was done with the whole series, I was actually enjoying it, and I went back to the beginning and read the whole thing. I reread the whole series a couple more times. Also read Forlesen (quoted here) and a bunch of other great stories, but I couldn’t quite bring myself to read any other novels. Then a bunch of years ago I read an interesting book about The Book of the New Sun which answered various mysteries to me, and I was motivated to read The Urth of the New Sun, which tied up some loose ends. I was absorbed by the Urth of the New Sun while I was reading it, but once it was over I was unsatisfied. And I think I know the problem: the book was enforcing a level of coherence that is just not possible with fiction.
Lots of writers create memorable fiction within self-contained worlds: Anne Tyler, Scott Turow, John Le Carre, J. R. R. Tolkien, Jane Austen, etc etc etc. These worlds seem real—-but you can’t stare at them too closely, or you’ll reach the resolution limit and everything starts to pixellate. The world of The Book of the New Sun was amazing, hinting at mysterious depths, but it fell apart upon under pressure. Best to let some things just be implied. It was wise of Turow, Le Carre, etc., to not push it so far.
And this reminded me of my argument that in the best alternative histories, the real world is what’s ultimately real.
And this in turn connects to statistical model building, and the idea that if we get enough data, we will be able to refute any model or any series of models (recall Cantor’s corner).