Sometime in the late 1980s, I was talking about writing with my friend, the mathematician Jim Propp, and he said, as a sort of throwaway challenge, to try imagining what it would take to write a novel that a 20th-century writer could not write. I remember being intrigued by that, and almost immediately realizing that hypermedia was the answer. I couldn’t write such a novel, of course, but what I could do was create a space within which such a novel might emerge. I had the great good fortune to stand on the shoulders of giants. I was a student in Neil Postman’s Media Ecology program at NYU, and the invited guest at their 1987 conference was Jay David Bolter, who introduced me to Michael Joyce and Mark Bernstein. I was working in Expository Writing at NYU with [Jane] Yellowlees Douglas, and travelled to Yale to meet Stuart Moulthrop and to Brown, where Eli Mylonas, George Landow, and Robert Coover were both doing pioneering work. There was a fantastic energy in the air around hypertext, like a Tesla coil that we were all tuning in.

Uncle Buddy’s Phantom Funhouse was an attempt to use hypermedia to embed the elements of story entirely within situationally appropriate artifacts. The conceit of the fiction was that an attorney has passed into the reader’s hands the literary estate of science fiction writer Arthur “Buddy” Newkirk. At one point, I had thought about literally distributing it on a hard drive — something that would only become possible much later — but settled for a hybrid approach, with a marked-up print proof of an unpublished story and two cassettes of music accompanying the digital traces of the vanished author.

As much as possible, I tried to adhere to one of the axioms our little cabal of theorists had cooked up: “all possible endings.” While there does exist some set of actions, pegged to times, that can be encountered in the fiction, how the reader makes sense of those will vary depending on where they have come from and where they go next. There are some characters, and two of them — Art Newkirk and Emily Keane — appear to have a good deal to say. But again, how the reader makes sense of their embedded narratives (and the degree to which they are trusted) will depend.

For me, it was important to have a balance of word and image, of discursive and presentational, and using HyperCard made that possible. To keep things loose, I added a strictly projective element, a set of “Oracle” fortune-telling cards that emerged for me as I was writing the text (and, in some cases, from which the text emerged.) And in sections of the fiction, I used the programming language, HyperTalk, as both a generator of randomness and as an embedded text.

The result, hopefully, is a text with emergent properties that could be read (and in the process, created) by a non-20th-century audience.