I have altogether about 4000 physical books, though I pretty much stopped buying new ones a decade ago. (There’s actually a list of many of my books, as “NKS references” at https://www.wolframscience.com/reference/books/.)

On my desk I have to say I have only one book: A New Kind of Science. And I refer to the paper version with some regularity. (I used to have a physical Mathematica book too, but that’s now been completely superseded by what’s online.)

But within reach… I have a bunch of reference books that I’ve been using as benchmarks for Wolfram|Alpha… as well as a few “classics” that I just feel I should have nearby.

Let’s see… there’s Newton, Darwin, Euclid, Galileo, Boole, D’Arcy Thompson, Linnaeus. (I used to have Turing’s collected works, but they seem to have gone missing.) (In addition to the famous books by these folk, I see I also have e.g. Darwin’s autobiography, D’Arcy Thompson’s book on Greek fishes, Darwin on barnacles, and Newton’s chronology of ancient kingdoms…)