Leiber and AI

A Grandmaster and the first two-time winner of Hugo for novels, Fritz Leiber is probably best known for his Fafhrd and the Grey Mouser series of stories.  However, some of his SF and horror stories remain well-regarded and accessible in recent anthologies, including the Library of America’s American Fantastic Tales (“Smoke Ghost”), and American Science Fiction (The Big Time).

LoA selections are almost always a solid bet: I recently read The Big Time (1958) and was surprised how different it is to the standard SF fare. It describes an ongoing “Change War” involving two sides fighting over the sweep of history (naturally, with the end of times arriving with an atomic bomb). We are not exposed to the battlefields or any famous historical figures; instead, the story is told through the actions and dialogue of convalescing soldiers and their caretakers. Several personalities from all over space and time populate a sort of vessel that is partially isolated from the rest of the Universe in a sort of vessel.

Jack Woolhiser cover for Ace (isdb.org)

The Big Time was written after a long break from productivity, when Leiber drew inspiration from a novel he read that featured an “intensified and embellished first person viewpoint”, creating a beatnik-era young woman who was scooped away from certain death into the refuge. She splits time between a principled but abusive German officer and a strange alien creature from the earliest, abundant years of the moon. It’s her role to attend to these other characters, but her own backstory and motivations are the most compelling. We indirectly experience the Change War through her eyes and ears, as soldiers from different eras party, brawl and incompletely recover from shell-shock. The dialogue (internal and external) is certainly unusual for SF and will likely be distracting for a lot of readers, but I thought it encapsulated the frozen-in-time nature of the character.

The other distinguishing feature of The Big Time is the single setting of the novel, strongly resembling a one-act play. The plot resembles that of a “locked room” mystery, where one of the characters manages cut off the vessel completely from the rest of the universe, and another accidentally arms a smuggled nuclear bomb. Leiber had a background in theater, and we can see the careful timing of plot events among the character interactions and expository descriptions. It made me think back on how several Fafhrd and Grey Mouser stories seemed to be so carefully staged and populated.

I would not claim The Big Time to be a favorite novel, but it’s definitely worth picking up for anyone interested in early Hugo winners, 1950’s postwar fiction or Leiber’s work in general.

The Leiber short story collection A Pail of Air (1964, stories originally published between 1950 and 1962) is rather scarce, but I was able to find its components online (thank you, SFF audio) and in other collections. It is one of the best single-author collections of short fiction that I’ve read over the past few years.

Richard Powers cover for Ballantine (isfdb.org)

“The Coming Attraction” (1950) is a strikingly pessimistic imagining of a future New York where human relationships are corrupted by violent thrill-seeking.

“A Pail of Air” (1951) is a rich survival story of a future Earth buried under layers of frozen atmosphere, after an astrophysical catastrophe named The Big Jerk pulled the planet away from the warmth of the sun.

Leiber had an enthusiasm for Greenwich Village bohemians. “Rump-tity-tity-tum-tah-tee” (1958) and “Pipe Dream” (1959) feature this lifestyle in entertaining, but dated, stories. A stronger statement is made in “The Beat Cluster” (1960), which places beatniks squarely (heh) in the center of the future of space exploration (after they hang out in the periphery, of course). All are recommended for anyone who enjoyed Kerouac’s onomatopoetic On the Road.

Most interesting to me, however, is the 1962 novelette “The 64-Square Madhouse,” about a chess tournament where world-class players encounter a new contestant – a machine featuring artificial intelligence. We experience this event through the eyes of a newspaper reporter and chess novice. She works for the Chicago Space Mirror, but that’s to put your mind at ease the humans will remain the best at chess for a long time after the moon is settled, and so on. It has its pedestrian moments, but the story is on the whole very readable.

The real gem of “The 64-Square Madhouse” occurs early, when our reporter is interviewing the oldest contestant before his first match. He explains how the computer must play in order to overtake the humans:

“If you had,” he said, “a billion computers all as fast as the Machine, it would take them all the time there ever will be in the universe just to play through all the possible games of chess, not to mention the time needed to classify those games into branching families of wins for White, wins for Black and draws …. So the Machine can’t play chess like God. What the Machine can do is examine all the likely lines of play for about eight moves ahead—that is, four moves each for White and Black—and then decide which is the best move on the basis of capturing enemy pieces, working toward checkmate, establishing a powerful central position and so on.”

This does not exactly anticipate the means by which IBM’s Deep Blue finally defeated Kasparov in 1996: that computer really did massively compute entire games between moves. However, Leiber does anticipate the changes needed to produce a system capable of conquering the game of go last year.

Unlike chess, which has maybe 20 to 30 choices for a typical move (creating about 15,000 board positions after 3 moves), go consists of placing a stone on a large (19 by 19) grid, creating a state space of possibilities far greater than that of chess (200 to 300 choices, and about 15 million board positions after 3 moves). Slogging through these massive options between turns like a typical chess AI system is simply impractical. The successful AlphaGo system was designed from the beginning to build intuition and play go like a human. It built a large information bank of every championship-level go match that could be found (also preceded by the “book-learning” in Leiber’s story) and kept multiple versions of itself to self-train, developing an ability to read the board and react correctly, just as an experienced go player would. So we reached this point (long) before settling the moon, but it’s interesting to see a bit of insight featured in Leiber’s short fiction.

The Big Time gets 7/10 and A Pail of Air gets 8/10. Leiber continues to be an author of considerable interest.

Note 1: The novel that inspired Leiber’s creation of Greta Forzane in The Big Time was Joyce Cary’s The Horse’s Mouth – an example of the useful content found in the Notes section of the LoA edition.

Note 2: A pretty good description of the challenges posed by chess and go to AI developers (prior to Deep Blue and AlphaGo) can be found in Introduction to Artificial Intelligence by Philip Jackson Jr., 1985 (2nd ed.)