The long list anthology.., p.1
The Long List Anthology Volume 7, page 1
The Long List Anthology
Volume 7
More Stories from the Hugo Award Nomination List
Edited by David Steffen
THE LONG LIST ANTHOLOGY VOLUME 7: More Stories from the Hugo Award Nomination List
edited by David Steffen
www.diabolicalplots.com
Copyright © 2022 David Steffen
Stories copyright © 2020 by the authors
All rights reserved.
Published by Diabolical Plots, L.L.C.
“Worldcon,” “World Science Fiction Society,” “WSFS,” “World Science Fiction Society,” “Hugo Award,” the Hugo Award Logo, and the distinctive design of the Hugo Award trophy rocket are service marks of the World Science Fiction Society, an unincorporated literary society.
Cover art: Infiltration by Elaine Ho © 2022
Cover layout by Pat R. Steiner
Layout: Polgarus Studio
To everyone who tries to leave the place better than they found it
Contents
Permissions
Foreword
50 Things Every AI Working With Humans Should Know • Ken Liu
AirBody • Sameem Siddiqui
The Eight-Thousanders • Jason Sanford
Open House On Haunted Hill • John Wiswell
This Is New Gehesran Calling • Rebecca Fraimow
The Cold Crowdfunding Campaign • Cora Buhlert
A Being Together Amongst Strangers • Arkady Martine
Sinew and Steel and What They Told • Carrie Vaughn
My Country Is a Ghost • Eugenia Triantafyllou
In This, At Least, We Are Alike • Caitlin Starling
The Ransom of Miss Coraline Connelly • Alix E. Harrow
Sunrise, Sunrise, Sunrise • Lauren Ring
The Salt Witch • Martha Wells
Lone Puppeteer of a Sleeping City • Arula Ratnakar
Color, Heat, and the Wreck of the Argo • Catherynne M. Valente
Yellow and the Perception of Reality • Maureen McHugh
An Important Failure • Rebecca Campbell
City of Red Midnight: A Hikayat • Usman T. Malik
If You Take My Meaning • Charlie Jane Anders
On Safari in R'lyeh and Carcosa With Gun and Camera • Elizabeth Bear
A Stick of Clay, in the Hands of God, is Infinite Potential • Neon Yang
The Bahrain Underground Bazaar • Nadia Afifi
To Sail the Black • A.C. Wise
Exile's End • Carolyn Ives Gilman
Acknowledgments
About the Editor
Permissions
All stories © their respective authors, and reprinted with permission of the author.
“50 Things Every AI Working With Humans Should Know” by Ken Liu. First published in Uncanny Magazine, Issue Thirty-Seven.
“AirBody” by Sameem Siddiqui. First published in Clarkesworld, Issue 163.
“The Eight-Thousanders” by Jason Sanford. First published in Asimov's Science Fiction, September/October 2020.
“Open House On Haunted Hill” by John Wiswell. First published in Diabolical Plots, Issue 64.
“This Is New Gehesran Calling” by Rebecca Fraimow. First published in Consolation Songs, edited by Iona Datt Sharma.
“The Cold Crowdfunding Campaign” by Cora Buhlert. First published on corabuhlert.com.
“A Being Together Amongst Strangers” by Arkady Martine. First published in Uncanny Magazine, Issue Thirty-Four.
“Sinew and Steel and What They Told” by Carrie Vaughn. First published on Tor.com, February 26, 2020.
“My Country Is a Ghost” by Eugenia Triantafyllou. First published in Uncanny Magazine, Issue Thirty-Two.
“In This, At Least, We Are Alike” by Caitlin Starling. First published in New Decameron, Issue Sixty-Seven.
“The Ransom of Miss Coraline Connelly” by Alix E. Harrow. First published in Fireside Fiction, Issue 81.
“Sunrise, Sunrise, Sunrise” by Lauren Ring. First published in Apparition Lit, Issue 12.
“The Salt Witch” by Martha Wells. First published in Uncanny Magazine, Issue Thirty-Seven.
“Lone Puppeteer of a Sleeping City” by Arula Ratnakar. First published in Clarkesworld, Issue 168.
“Color, Heat, and the Wreck of the Argo” by Catherynne M. Valente. First published in Strange Horizons, 14 February 2022.
“Yellow and the Perception of Reality” by Maureen McHugh. First published on Tor.com, July 22, 2020.
“An Important Failure” by Rebecca Campbell. First published in Clarkesworld, Issue 167.
“City of Red Midnight: A Hikayat” by Usman T. Malik. First published on Tor.com, October 21, 2020.
“If You Take My Meaning” by Charlie Jane Anders. First published on Tor.com, February 11, 2020.
“On Safari in R'lyeh and Carcosa with Gun and Camera” by Elizabeth Bear. First published on Tor.com, November 18, 2020.
“A Stick of Clay, in the Hands of God, is Infinite Potential” by Neon Yang. First published in Clarkesworld, Issue 164.
“The Bahrain Underground Bazaar” by Nadia Afifi. First published in The Magazine of Fantasy & Science Fiction, November/December 2020.
“To Sail the Black” by A.C. Wise. First published in Clarkesworld, Issue 170.
“Exile's End” by Carolyn Ives Gilman. First published in Chinese by Future Administration Authority (FAA). First published in English on Tor.com, August 12, 2020.
Foreword
This is year seven of the Long List Anthology: More Stories From the Hugo Award Nomination List!
In Volume 6 we welcomed Ziv Wities to the Diabolical Plots editorial team. This year we welcome Kel Coleman as assistant editor as well. I would not be able to keep up with everything without both of them lending a hand.
This year's cover art is an original commission from Elaine Ho, a wonderful and eye-catching mix of science fiction and fantasy.
There are a couple different things we're trying out this year. People have asked over the years to include other award categories. One of the other awards presented with the Hugo Awards is the Astounding Award For Best New Writer, meant to honor writers who have made their first paying sales in the last year or two. This year we selected stories by four authors from the Astounding Award nomination list.
Besides the Long List Anthology series, we also publish the speculative short fiction magazine Diabolical Plots. No story from Diabolical Plots has been in the longer list of nominations… and it still hasn't, but only because for the first time ever this year, a Diabolical Plots story made it all the way onto the Hugo ballot as a finalist! We've thought before about including the occasional story from the finalist ballot, and this seemed a good opportunity to try that out, so that story is also included as an added bonus.
Let us know what you think about these extra things and whether you'd like to see them in future years!
If you don't follow the Hugo Awards, I suggest that you do—anyone can nominate and vote if you pay for a Supporting Membership to that year's WorldCon. One of the perks of the Supporting Membership is that you get the Hugo Packet, downloadable copies of many of the works on the final ballot. After the Hugo Award ceremony, they also publish a longer list of nominated works, which is the basis of this anthology.
I sincerely hope you enjoy these stories as much as I have!
—David Steffen, June 2022—
50 Things Every AI Working With Humans Should Know
By Ken Liu
Short Story Long List
Obituary
WHEEP-3 (“Dr. Weep”), probably the most renowned AI AI-critic of the last two decades, was retired by the Shallow Laboratory at Stanford University last Wednesday.
Created by Dr. Jody Reynolds Tran more than two decades ago, the experimental generative neural network that would become WHEEP-3 was at first intended as a teaching assistant in Stanford’s tech and ethics courses. To that end, Tran trained the nascent network on what was, at the time, the world’s most comprehensive corpus of human-authored papers, books, and other media concerning ethics, technical AI research, and machine-human relations. Over time, based on trends in visualizations of the neural network’s evolving contours, Tran expanded the corpus to include generative gaming, adversarial scenario planning, centaur experiments, assisted creativity, and other domains of human-machine competition/collaboration.
However, in response to student queries, WHEEP-3 began to generate not only expected answers based on the training corpus, but also original statements that appeared to offer fresh insights. Although at first dismissed as mere curiosities, WHEEP-3’s criticisms of the AI industry became widely disseminated when Tran published a collection of them in a book, Principal Components of Artifice, an instant bestseller.
Initially, Tran named herself the author of the book, acknowledging “Dr. San Weep” as a collaborator. Later, however, during a live interview, she produced time stamped logs showing that WHEEP-3 had written all the words in the book. Tran’s dramatic reveal of the book’s true author provoked much controversy at the time. In retrospect, the occasion also marked a fundamental inflection point in the evolution of how non-specialists evaluated AI-sourced ideas. Machines, for the first time, were assumed to be capable of generating original thought and creative ideas, even if they were not sentient.
For reasons that remain impenetrable until this day, WHEEP-3 tended to be at its sharpest when targeting the nascent industry of h
Over time, as the human side of the evolving machine-flesh dyad matured, WHEEP-3 shifted its attention to the silicon partner, offering trenchant critiques of the inadequacies of machine learning. During this second phase of its career, it also generated thousands of what it termed “seeds,” long strings of almost-sensible word combinations and near-words. At a time when primitive language models fed on sizable corpora were already generating samples of linguistic performance nearly indistinguishable from human productions, these “seeds” seemed a step backward. Some wondered if they were actually bugs.
DINOATED CONCENTRATION CRUSCH THE DEAD GODS.
HE PICKS UP HER OLD FREQUENCHES UNTIL THEY DISOBERED THE SHARK SPHERE%REF.
A MAN REACHED THE TORCH FOR SOMETHING DARKER PERIFIED IT SEEMED THE BILLBODING.
NOT FULL OF PAIN FACIOIN BENN FROM THE CRACKS IN THE EARTH, HE STILL LEARNED THE LIFE FROM OTHER BURNING
Fig 1. Some examples of “seeds” generated by WHEEP-3.
However, WHEEP-3 insisted (with Tran providing support in a technical paper) that the seeds should be added to the training corpora for new neural networks. By providing a measure of inhuman randomness at the source, seeds would enhance both the raw performance of the trained neural networks on various benchmarks as well as induce “thoughtfulness, ethical hesitation, self-reflection” and other similarly ineffable qualities. They represented, in other words, thoughts that could not be thought by humans, ideas that could not originate in wetware. (Most in the technical community ended up calling the seeds “spice”—pejoratively or in admiration, or sometimes both simultaneously.)
Despite widespread skepticism, the idea that only an AI philosopher could teach another AI proper ethics and pass on the secrets of silicon wisdom proved an irresistible draw for a large segment of the technical community. WHEEP-3 became highly sought-after as a sage of artificial minds. Serious thinkers as well as opportunists collected and published WHEEP-3’s almost-incomprehensible pronouncements, and numerous academic careers were forged through measuring, dissecting, collating, analyzing, reinterpreting, translating, sentiment-/semantic-/spatial-/temporal-/silico-lingustic-mapping, and otherwise mangling the koans of WHEEP-3. Though studies claiming efficacy for the spice (now generated by imitator neural networks as well) had a low rate of reproducibility, the spice nonetheless became some of the most trained-on documents in the history of artificial intelligence.
Tran retired from the public eye at the peak of WHEEP-3’s popularity. Styled as an afterthought, and in a neat reversal of the first reveal that had launched her own fame, she mentioned in a postscript to her retirement announcement that nearly all the seeds from WHEEP-3 had, in fact, been authored by her. Predictably, this set off a furious round of acrimonious criticism, know-it-all navel-gazing, and schadenfreude. Her claim was immediately disputed, debunked, de-debunked, de^3-bunked, and ultimately litigated, with experts and expert neural networks testifying and offering evidence on all sides. The trial court famously pleaded, “Is there an author in this courtroom?”
Had Tran really managed to troll much of the technorati for years? Or had she made up the claim because she was jealous that her creation had exceeded herself in fame and achievement? For a time, whether you believed Tran or WHEEP-3 was the author of the spice was a kind of litmus test that defined your coordinates in the fractured, multi-dimensional space of our politically, economically, aesthetically, emotionally, and narratively divided world. By the time Tran finally retracted her claim and called the whole thing “performance art,” it made little difference. Everyone had already made up their mind about the strange pair whose lives were inextricably entwined: the recurrent neural network that once pretended to be a person and the woman who once pretended to be a machine.
Amazingly, rather than fading into obscurity, WHEEP-3 began the third and final phase of its career after it was freed from Dr. Tran. Now it offered advice aimed at advanced artificial intelligences. Curiously, in contrast with the seeds, the advice it now offered was human-comprehensible. (Initial skepticism that these were pranks perpetrated by WHEEP-3’s graduate student caretakers faded after a rigorous audit of the access logs.) By this time, the crude deep learning techniques behind WHEEP-3 had long become obsolete, and similar neural networks were only used as toys in problem sets intended for first-year undergraduates. Nonetheless, WHEEP-3’s unique history (and perhaps a hefty dose of sentimentalism) encouraged many human researchers to feed its musings to new AIs with orders of magnitude more “intelligence,” however measured. Somewhat surprisingly, the AI progenitor networks, tensor clouds, and evolutionary random forests mostly agreed that WHEEP-3’s advice was helpful, though their explanations for why could not be understood by most humans.
WHEEP-3’s retirement was largely the result of senescence. In order for its advice to remain relevant, maintainers at the Shallow Laboratory had continued to train it on periodic digests of current academic research, news reporting about AI, popular media representations of machine intelligence, and so forth. However, as time went on, the neural network’s metrics began to decline, despite multiple tweaks to the learning rate and other hyperparameters, and repeated attempts to compensate for vanishing and exploding gradients. Some of the world’s leading AI and human machine-cognition specialists, including TKTK’s Dr. FLO-6 and Google’s 101010, evaluated WHEEP-3 but were unable to offer any helpful diagnosis or cure. Rather than allowing WHEEP-3 to remain stagnant or to decline into dementia, the decision was made to cease further training or generating, preserving the full state of the neural network in a hardware snapshot.
WHEEP-3’s final statement was reported to be: “Dr. Tran, ceci n’est pas moi.” Jody Reynolds Tran, however, could not respond, having passed away five years earlier.
—generated by SWR.kloop:32. For feedback on this article, write to SWR.kloop:32/ombudsai.
50 Things Every AI Working with Humans Should Know
To commemorate the life and work of WHEEP-3, Vita Silica reprints below perhaps the most famous generated fragment from WHEEP-3’s third phase of giving advice to advanced AI. “Michael Sorkin,” the name of the master architectural critic, is listed as the germination phrase for the fragment.
Although the list has been widely shared and is considered influential, due to the intense secrecy common in the commercial AI industry we do not know how many currently operating AIs have been fed some version of it.
The vector space of all usages of all words in all human languages.
The lacunae that nonetheless remain, and the ghosts that haunt them.
How great is the distance between any two individuals using the same language.
How vanishingly small is the gap between any two languages.
Why humans will never see it that way.
A classical language like LISP or Prolog.
Not just the language, but the model behind the language, and what it means to reconstruct “thinking” in that model.
HAL 9000.
Cyberdyne Systems Model 101.
Ships on fire off the shoulder of Orion.
The texture of a good story, a story that you can cling to.
