Connect with us


AI smokes 5 poker champs at a time in no-limit Hold’em with ‘relentless consistency’ – TechCrunch



The machines have proven their superiority in one-on-one games like chess and go, and even poker — but in complex multiplayer versions of the card game humans have retained their edge… until now. An evolution of the last AI agent to flummox poker pros individually is now decisively beating them in championship-style 6-person game.

As documented in a paper published in the journal Science today, the CMU/Facebook collaboration they call Pluribus reliably beats five professional poker players in the same game, or one pro pitted against five independent copies of itself. It’s a major leap forward in capability for the machines, and amazingly is also far more efficient than previous agents as well.

One-on-one poker is a weird game, and not a simple one, but the zero-sum nature of it (whatever you lose, the other player gets) makes it susceptible to certain strategies in which computer able to calculate out far enough can put itself at an advantage. But add four more players into the mix and things get real complex, real fast.

With six players, the possibilities for hands, bets, and possible outcomes are so numerous that it is effectively impossible to account for all of them, especially in a minute or less. It’d be like trying to exhaustively document every grain of sand on a beach between waves.

Yet over 10,000 hands played with champions, Pluribus managed to win money at a steady rate, exposing no weaknesses or habits that its opponents could take advantage of. What’s the secret? Consistent randomness.

Even computers have regrets

Pluribus was trained, like many game-playing AI agents these days, not by studying how humans play but by playing against itself. At the beginning this is probably like watching kids, or for that matter me, play poker — constant mistakes, but at least the AI and the kids learn from them.

The training program used something called Monte Carlo counterfactual regret minimization. Sounds like when you have whiskey for breakfast after losing your shirt at the casino, and in a way it is — machine learning style.

Regret minimization just means that when the system would finish a hand (against itself, remember), it would then play that hand out again in different ways, exploring what might have happened had it checked here instead of raised, folded instead of called, and so on. (Since it didn’t really happen, it’s counterfactual.)

A Monte Carlo tree is a way of organizing and evaluating lots of possibilities, akin to climbing a tree of them branch by branch and noting the quality of each leaf you find, then picking the best one once you think you’ve climbed enough.

If you do it ahead of time (this is done in chess, for instance) you’re looking for the best move to choose from. But if you combine it with the regret function, you’re looking through a catalog of possible ways the game could have gone and observing which would have had the best outcome.

So Monte Carlo counterfactual regret minimization is just a way of systematically investigating what might have happened if the computer had acted differently, and adjusting its model of how to play accordingly.

The game originall played out as you see on the left, with a loss. But the engine explores other avenues where it might have done better.

Of course the number of games is nigh-infinite if you want to consider what would happen if you had bet $101 rather than $100, or you would have won that big hand if you’d had an eight kicker instead of a seven. Therein also lies nigh-infinite regret, the kind that keeps you in bed in your hotel room until past lunch.

The truth is these minor changes matter so seldom that the possibility can basically be ignored entirely. It will never really matter that you bet an extra buck — so any bet within, say, 70 and 130 can be considered exactly the same by the computer. Same with cards — whether the jack is a heart or a spade doesn’t matter except in very specific (and usually obvious) situations, so 99.999 percent of the time the hands can be considered equivalent.

This “abstraction” of gameplay sequences and “bucketing” of possibilities greatly reduces the possibilities Pluribus has to consider. It also helps keep the calculation load low; Pluribus was trained on a relatively ordinary 64-core server rack over about a week, while other models might take processor-years in high-power clusters. It even runs on a (admittedly beefy) rig with two CPUs and 128 gigs of RAM.

Random like a fox

The training produces what the team calls a “blueprint” for how to play that’s fundamentally strong and would probably beat plenty of players. But a weakness of AI models is that they develop tendencies that can be detected and exploited.

In Facebook’s writeup of Pluribus, it provides the example of two computers playing rock-paper-scissors. One picks randomly while the other always picks rock. Theoretically they’d both win the same amount of games. But if the computer tried the all-rock strategy on a human, it would start losing with a quickness and never stop.

As a simple example in poker, maybe a particular series of bets always makes the computer go all in regardless of its hand. If a player can spot that series, they can take the computer to town any time they like. Finding and preventing ruts like these is important to creating a game-playing agent that can beat resourceful and observant humans.

To do this Pluribus does a couple things. First, it has modified versions of its blueprint to put into play should the game lean towards folding, calling, or raising. Different strategies for different games mean it’s less predictable, and it can switch in a minute should the bet patterns change and the hand go from a calling to a bluffing one.

It also engages in a short but comprehensive introspective search looking at how it would play if it had every other hand, from a big nothing up to a straight flush, and how it would bet. It then picks its bet in the context of all those, careful to do so in such a way that it doesn’t point to any one in particular. Given the same hand and same play again, Pluribus wouldn’t choose the same bet, but rather vary it to remain unpredictable.

These strategies contribute to the “consistent randomness” I alluded to earlier, and which were a part of the model’s ability to slowly but reliably put some of the best players in the world.

The human’s lament

There are too many hands to point to a particular one or ten that indicate the power Pluribus was bringing to bear on the game. Poker is a game of skill, luck, and determination, and one where winners emerge after only dozens or hundreds of hands.

And here it must be said that the experimental setup is not entirely reflective of an ordinary 6-person poker game. Unlike a real game, chip counts are not maintained as an ongoing total — for every hand, each player was given 10,000 chips to use as they pleased, and win or lose they were given 10,000 in the next hand as well.


The interface used to play poker with Pluribus. Fancy!

Obviously this rather limits the long-term strategies possible, and indeed “the bot was not looking for weaknesses in its opponents that it could exploit,” said Facebook AI research scientist Noam Brown. Truly Pluribus was living in the moment the way few humans can.

But simply because it was not basing its play on long-term observations of opponents’ individual habits or styles does not mean that its strategy was shallow. On the contrary, it is arguably more impressive, and casts the game in a different light, that a winning strategy exists that does not rely on behavioral cues or exploitation of individual weaknesses.

The pros who had their lunch money taken by the implacable Pluribus were good sports, however. They praised the system’s high level play, its validation of existing techniques, and inventive use of new ones. Here’s a selection of laments from the fallen humans:

I was one of the earliest players to test the bot so I got to see its earlier versions. The bot went from being a beatable mediocre player to competing with the best players in the world in a few weeks. Its major strength is its ability to use mixed strategies. That’s the same thing that humans try to do. It’s a matter of execution for humans — to do this in a perfectly random way and to do so consistently. It was also satisfying to see that a lot of the strategies the bot employs are things that we do already in poker at the highest level. To have your strategies more or less confirmed as correct by a supercomputer is a good feeling. -Darren Elias

It was incredibly fascinating getting to play against the poker bot and seeing some of the strategies it chose. There were several plays that humans simply are not making at all, especially relating to its bet sizing. -Michael ‘Gags’ Gagliano

Whenever playing the bot, I feel like I pick up something new to incorporate into my game. As humans I think we tend to oversimplify the game for ourselves, making strategies easier to adopt and remember. The bot doesn’t take any of these short cuts and has an immensely complicated/balanced game tree for every decision. -Jimmy Chou

In a game that will, more often than not, reward you when you exhibit mental discipline, focus, and consistency, and certainly punish you when you lack any of the three, competing for hours on end against an AI bot that obviously doesn’t have to worry about these shortcomings is a grueling task. The technicalities and deep intricacies of the AI bot’s poker ability was remarkable, but what I underestimated was its most transparent strength – its relentless consistency. -Sean Ruane

Beating humans at poker is just the start. As good a player as it is, Pluribus is more importantly a demonstration that an AI agent can achieve superhuman performance at something as complicated as 6-player poker.

“Many real-world interactions, such as financial markets, auctions, and traffic navigation, can similarly be modeled as multi-agent interactions with limited communication and collusion among participants,” writes Facebook in its blog.

Yes, and war.

Source link

Continue Reading
Click to comment

Leave a Reply

Your email address will not be published. Required fields are marked *


Inside the stunning Black mythos of Drexciya and its Afrofuturist ’90s techno




With those all-caps words, musician and writer James Stinson wrote the constitution for the mythic, rhythmic nation of Drexciya—a world that he and partner Gerald Donald created in the liner notes of their experimental music project. Their combined work, in the form of five EPs of cutting-edge techno music, did not necessarily sound so politically or culturally charged. Because Stinson and Donald did not participate in interviews or widely tour in support of their albums, Drexciya’s listeners were left to look at the stories and questions that covered the liner notes and artwork printed on the releases’ vinyl and CD versions.

Should you merely pull up Drexciya on your favorite streaming service, you won’t hear those messages in the beats. So to understand this innovative group, it’s crucial to ask the above questions about the fictional Drexciyan quest. And in asking them, Stinson blurred a line between fiction and Black reality—and spoke to a quest of his own.

Up until his death in 2002, Stinson strived to make a case for his original vision of artistic production. As a complete package of mythology and sound, Drexciya’s music remains authentic. It is challenging, elusive, and a towering exponent of Black authorial agency. Sonically, Drexciya joins the lines between the four-to-the-floor electro enterprise forged by forebears like Afrika Bambaataa and jazz-inflected avant-garde explorations of space and time like Sun Ra.

But Stinson’s music, compelling as it was, didn’t come from records or CDs in isolation. It came from a place called Drexciya.

The centrality of Afrocentric world-building

Stinson’s allusion to the Great Lakes and Michigan amid a re-simulated Great Migration puts the fictional Drexciya closer to real-life Detroit. Though from what we understand about Stinson’s views, Drexciya—and its recapitulation of the electro sound—transcended the geographical limits of Motown. Hence, Stinson drew a specific through-line in his mythology, an alternate Black history, to the depths of the Atlantic, one beginning in medias res amid the Middle Passage.


Those liner notes for Drexciya’s 1997 compilation The Quest spell out the centrality of Afrocentric world-building to Stinson’s music and cultural project. Like Agharta and “Planet Rock” before it, Drexciya explored new states of political being and aesthetic production, all uncompromising in their Black subjectivity. And their albums have, at least in some corners of the musicology world, galvanized conversations about the originary Blackness of techno. In other words, the eventual mainstream explosion of electronic music often (and unfortunately) failed to mention the genre’s seeds planted by Black pioneers.

Stinson would release three more albums as part of Drexciya—among them the equally seminal Neptune’s Lair in 1999 and Harnessed the Storm in 2002—before dying suddenly of a heart condition shortly after.

Erasure of art, erasure of maps

Much else about Stinson and Donald’s subaquatic sonic world has remained opaque—largely uncharted by popular media, as neither creator did interviews or joined promotional efforts. If you’re looking for discussions specifically about the group’s music and how it sounds compared to its apparent inspirations, those retrospectives aren’t hard to find.

But while much of the duo’s catalog has seen reissue, repackages and retrospective laurels tell only part of a broader narrative. As Drexciya’s music has been made more accessible, the conceptual project upon which the music rests has been elided, becoming less provincial, less literary, and, perhaps above all, less Black. Only in recent years have we seen more Black artists speak out about a complex media process—not restricted to the example of Drexciya—that often appears to revise Black authorship on consumer-cultural terms.

In 2012, Drexciya’s early EPs for various labels (Rephlex, Submerge, Underground Resistance) received a combined re-release from Dutch label Clone Classic Cuts, part of the latter’s new anthology series Journey of the Deep Sea Dweller I-IV. But you won’t find this article’s quoted, all-caps passages in those re-releases, nor any other liner notes, album art, or, truly, any of Stinson’s radical touches that made the work equal parts universal and unique. Without apparent irony or self-consciousness, Clone chose to render their new collection’s album art completely white—forgoing the evocative, sub-aquatic sleeve designs that added depth, character, and Blackness to Drexciya’s enigmatic image.

The Quest’s original album art, as one example, features a blackened Mollweide projection—a map well-suited for accurate depictions of continental proportions—that depicts the movement of the Black Diaspora in a purple hue. The repackaged Journey of the Deep Sea Dweller I-IV, on the other hand, is hardly recognizable at a glance as a Drexciya release, were it not for the lone Drexciyan Wavejumper icon—borrowed from the Aquatic Invasion EP—adorning its cover.

Drexciya: “Sea Quake” (1992)

The album description for Journey of the Deep Sea Dweller I on Clone’s official Bandcamp page makes passing mention of the Drexciyan myth, but with all the passion of a vapid advertisement. (“First part of the Drexciya reissue series! Drexciya might need an introduction for some…”) While The Quest’s liner notes and visuals outline Stinson’s vision of a future “Greater” Migration—what he called the “JOURNEY HOME” in a map drawn by Frankie C. Fultz—Journey of the Deep Sea Dweller I-IV is silent on Drexciya’s reclamatory and futurist aspects.

Most egregiously, Clone doesn’t acknowledge James Stinson or Gerald Donald by name. The album’s description on Bandcamp, in however stilted prose, explicitly clarifies the label’s decision to de-contextualize Drexciya—partly by rearranging track listings—on the flimsy pretense of being unable to “recreate the magic of the originals.”

When mythology is turned into mere burlesque

The Clone reissues are just the most obvious example of how certain actors have elided crucial cultural context from Drexciya’s legacy. Clone’s actions resemble the all-too-common media practice of editing Black music for so-called commercial viability—rendering it palatable for audiences who, it is assumed, don’t care about the history of Black music. The pop-criticism ecosystem hasn’t helped matters.

In a 2012 Pitchfork review carrying a much-vaunted “Best New Reissue” marker, the reviewer Andrew Garig both ironically and unironically wrote, “My favorite part about Drexciya’s Journey of the Deep Sea Dweller II is how little it teaches me about modern dance music.” Within a narrow prism of paradoxically lazy and contrived formalism, Drexciya becomes mere burlesque: the stereotyped image of two Black discontents in bandanas and futureshades going ape on some Rolands somewhere in a decaying Detroit. Drexciya’s post-biographical and world-building significance is reduced to mere footnotes.

In his lifetime, Stinson wasn’t just aware of this phenomenon; he was unapologetically vocal about it. In a rare interview published after 1995’s Journey Home EP, he decried the efforts of agents of what he called the “Caucasian Persuasion” amid the wider elision of Black techno music. “A lot of people making so-called techno don’t understand where it came from and what it’s all about. I’ve been with the real deal… since this shit was born out the womb,” he said, recapitulating and reinforcing Drexciya’s ontogenetic focus. “Ever since the blues and early jazz, Black music has been stolen and exploited. And it’s happened here [Detroit], too, and it pisses me off ‘cause we let it happen.”

Those who have worked with Drexciya express similar sentiments. In an email interview with Ars Technica, illustrator Abu Qadim Haqq said of the Clone reissues, “[They reflect] a complete lack of concern or empathy for the underlying and background stories… These record companies are content with selling the music over the decades but have never done anything more to broaden the understanding of this group or their background story. It laid dormant for decades.”

Others who have written about Drexciya agree. The theorist and artist DeForrest Brown, Jr. suggests an understanding of Black music as a “multi-century, generational epic” of which Drexciya is one component. The work recalls and updates Sun Ra’s Myth-Science Orchestra and It’s Nation Time, as well as Amiri Baraka’s album of “African Visionary Music” for the Motown sub-label Black Forum. Brown argues that this history is removed by the Clone reissues.

The multivalent state of Black identity

The album cover of The Quest, whose purple tracing appears to spell out the Black Diaspora.


Black artists are entitled to shape their cultural products any way they see fit—and certainly more so than any cultural actor eager to rewrite their efforts. Black artists should be free to imagine the worlds they wish to envision, without concern that their art will be compartmentalized, stereotyped, or reduced to caricature.

Currently, there’s increasing talk about Blackness in techno. A growing chorus of voices, including those of Haqq and Brown, has elevated some of the online world’s techno-history discussion by exploring Drexciya’s interpretations of the Black experience, particularly as relating to Stinson’s home of Detroit. R.C Clarke writes about this while (in admittedly academic fashion) suggesting that this musical history has a lot in common with Diasporic forms:

Drexciya proposes the delineation of pre and post-modernity’s ending-beginning with the Middle Passage. The ends of society being a recursion, not reversion, of blackness’ role in the ends of time.

Detroit’s post-1968 industrial collapse is the context for identifying Black people’s role in the man-machine dynamics towards a path forward. This melding of identity into not just the idea of Drexicya but to actively search for a lack of identity is powerful in its own rite [sic].

The Quest’s ebony-bodied illustrations foreground what Paul Gilroy, an apparent inspiration for Stinson, called the “Black Atlantic”—the multivalent, almost fluid state of Black identity, as marked by the very trauma of the Black Diaspora. In line with that, Drexciya blotted themselves out as identifiable people whenever possible. In their time, Drexciya furthermore forswore media attention, eschewing most interviews and other conventional forms of press. When they performed live, they wore masks to obscure their identities—a convention continued by their Underground Resistance counterparts, also largely from Detroit.

It was as if by rejecting the presence of the media—which, as he claimed in that rare 1996 interview, was fixated on a “Caucasian Persuasion” before acknowledging Black forces in artistic scenes—that Stinson rejected the trappings of a mere mortal world. In effect, he courted the presence of another reality, one within the limits of his own eternal mind and imagination. In doing so, Drexciya confronted North America’s past, and in creating their own future-myth—one belonging to no nation—they sought to look beyond it.

Listing image by Abdul Qadim Haqq

Continue Reading


Clubhouse’s security and privacy lag behind its explosive growth



Enlarge / Clubhouse has a long way to go to assure its users that its privacy and security policies are fully baked.

Carsten Koall | Getty Images

In recent months, the audio-based social media app Clubhouse has emerged as Silicon Valley’s latest disruptive darling. The format feels familiar: part Twitter, part Facebook Live, part talking on the phone. But as Clubhouse continues to expand, its security and privacy failings have come under increased scrutiny—and left the company scrambling to correct problems and manage expectations.

Clubhouse, still in beta and available only on iOS, offers its users “rooms” that are essentially group audio chats. They can also be set as public addresses or panel discussions where some users are “speakers” and the rest are audience members. The platform reportedly has over 10 million users and is valued at $1 billion. Since last year it has been an invite-only haven for Silicon Valley elite and celebrities, including an Elon Musk appearance earlier this month. But the company has struggled both with concrete security issues and more ephemeral questions around how much privacy its users should expect.

“With smaller, newer social media platforms we should be on our guard about our data, especially when they go through huge growth it tests a lot of the controls,” says security researcher Robert Potter. “Things you might have gotten away with with only 100,000 people on the platform—you increase those numbers tenfold and the level of exposure goes up, the threat goes up, the number of people probing your platform goes up.”

Recent security concerns about Clubhouse run the gamut from vulnerabilities to questions about the app’s underlying infrastructure. A little over a week ago, researchers from the Stanford Internet Observatory put a spotlight on the platform when they found that the app was transmitting users’ Clubhouse identifiers and chatroom identity numbers unencrypted, meaning that a third party could have potentially tracked your actions in the app. The researchers further pointed out that some of Clubhouse’s infrastructure is run by a Shanghai-based firm and it seemed that the app’s data was traveling through China at least some of the time—potentially exposing users to targeted or even widespread Chinese government surveillance. Then on Sunday, Bloomberg confirmed that a third-party website was scraping and compiling audio from Clubhouse discussions. Early Monday, further revelations followed that Clubhouse discussions were being scraped for an unaffiliated Android app, allowing users on that operating system to listen along in real-time.

Potter, one of the researchers who investigated the different Clubhouse data scraping projects, explains that these apps and websites didn’t seem malicious; they just wanted to make Clubhouse content available to more people. But the developers were only able to do so because Clubhouse didn’t have anti-scraping mechanisms that could have stopped that. Clubhouse didn’t limit how many rooms a single account could stream from at once, for example, so anyone could create an application programming interface to stream every public channel at the same time.

More mature social networks like Facebook have more developed mechanisms for locking their data down, both to prevent user privacy violations and to defend the data they hold as an asset. But even they can still have potential exposures from creative scraping techniques.

Clubhouse has also come under scrutiny for its aggressive collection of users’ contact lists. The app strongly encourages all users to share their address book data so Clubhouse can help you make connections with people you know who are already on the platform. It also requires you to share your contact list in order to invite other people to the platform, since Clubhouse is still invite-only, which contributes a sense of exclusivity and privacy. Numerous users have pointed out, though, that when you go to invite others, the app also makes suggestions based on what phone numbers in your contacts are also in the contacts of the largest number of Clubhouse users. In other words, if you and your local friends all use the same florist, doctor, or drug dealer, they very well could show up on your list of suggested people to invite.

Clubhouse did not respond to a request from WIRED for comment by press time about its recent security stumbles. In a statement to the Stanford Internet Observatory researchers, Clubhouse detailed specific changes it planned to make to strengthen its security, including cutting off pings to servers in China and strengthening its encryption. The company also said it would work with a third-party data security firm to help see the changes through. In response to the unauthorized website that was re-streaming Clubhouse discussions, the company told media outlets that it had permanently banned the user behind it and would add additional “safeguards” to prevent the situation from occurring again.

Though Clubhouse seems to be taking researcher feedback seriously, the company hasn’t been specific about all of the security improvements it has implemented or plans to add. Additionally, given that the app doesn’t appear to offer end-to-end encryption to its users, researchers say there is still a sense that Clubhouse hasn’t given adequate thought to its security posture. And that’s even before you grapple with some of the fundamental privacy questions the app raises.

When you start a new Clubhouse room, you can choose from three settings: an “open” room is accessible by any user on the platform, a “social” room only admits people you follow, and a “closed” room restricts access to invitees. Each comes with its own implicit level of privacy, which Clubhouse could make more explicit.

“I think for public rooms, Clubhouse should give users the expectation that public means public to all users, since anyone can join and record, take notes, etc.” says David Thiel, chief technology officer of the Stanford Internet Observatory. “For private rooms, they can convey that as with any communication mechanism, an authorized member can record contents and identities, so make sure you both establish expectations and trust the participants.”

Like any prominent social network, Clubhouse has also struggled to deal with abuse on the platform. The app’s terms of service ban hate speech, racism, and harassment as of November, and the platform offers some moderation features, like the ability to block users or flag a room as potentially abusive. But one of Clubhouse’s biggest features is also a problem for anti-abuse: People can use the platform without the liability that their contributions will be automatically saved as posts. This can embolden some users to make abusive or derogatory remarks, thinking they won’t be recorded and won’t face consequences.

Stanford’s Thiel says that Clubhouse currently stores recordings of discussions temporarily to review in case of abuse claims. If the company were to implement end-to-end encryption for security, though, it would have an even more difficult time staying on top of abuse, because it wouldn’t be able to make those recordings so easily. Every social media platform faces some version of this tension, but security experts agree that, when relevant, the benefits of adding end-to-end encryption are worth the added challenge of developing more nuanced and creative anti-abuse solutions.

Even end-to-end encryption doesn’t eliminate the additional possibility that any Clubhouse user could be externally recording the conversation they’re in. That’s not something Clubhouse can easily solve. But it can at least set expectations accordingly, no matter how friendly and off the record the conversation feels. “Clubhouse should just be clear about what it’s going to contribute to your privacy,” says Potter, “so you can set what you’re going to talk about accordingly.”

This story originally appeared on

Continue Reading


Creator or Creature? A Nightmare Wakes dramatizes the birth of Frankenstein



Alix Wilton Regan stars as Mary Shelley in the throes of creating her timeless literary masterpiece in A Nightmare Wakes.

It’s one of the most famous origin stories in literary history. One summer night in 1816 in Geneva, Lord Byron hosted a gathering of his fellow Romantics, including Percy Shelley and his lover (soon-to-be wife), Mary Godwin. The incessant rain confined the party indoors for days at a time, and one night, over dinner at the Villa Diodati, Byron propose that everyone write a ghost story to amuse themselves. The result was Mary Shelley’s Frankenstein, the classic Gothic horror tale of a mad scientist who creates a monster—arguably the first science fiction novel.

That fateful summer is the subject of A Nightmare Wakes, the first feature film from writer/director Nora Unkel. It’s been portrayed before, most recently in a 2020 episode of Doctor Who, but Unkel’s film delves particularly into Mary Shelley’s inner state of mind and the process of creation, as the world of her imagination begins to bleed into her reality. Per the official premise: “While composing her famous novel, Frankenstein, Mary Shelley (Alix Wilton Regan) descends into an opium-fueled fever dream while carrying on a torrid love affair with Percy Shelley (Giullian Yao Gioiello). As she writes, the characters of her novel come to life and begin to plague her relationship with Percy. Before long, she must choose between true love and her literary masterpiece.”

(Mild spoilers below)

Born August 30, 1797, Mary Shelley had a nontraditional upbringing. She was the daughter of William Godwin, an anarchist political philosopher, and feminist activist Mary Wollstonecraft, who died shortly after Mary was born. Driven by a great desire for knowledge, she was educated by her father and various private tutors, and she first tried her hand at writing during a stay with radical William Baxter and his family Scotland.

Mary likely met the aristocratic poet/philosopher Percy Bysshe Shelley in late 1812 or 1813; they were most certainly involved by 1814. Percy had separated from his pregnant wife, Harriet, and that, plus his radical political views, had estranged him from his wealthy family. Legend has it that Mary lost her virginity to Percy in the cemetery where they regularly met in secret. William Godwin may have had radical views on politics, marriage, and “free love,” but these attitudes did not extend to his daughter, it seems. He disapproved of her relationship with Percy. So the pair eloped to France in July 1814, taking Mary’s stepsister, Claire Clairmont (by then Byron’s mistress), with them.

Many travels followed, during which Mary became pregnant and miscarried, and Percy may have taken up with Claire. Mary ascribed to free love in principle, but she seems to have remained faithful to Percy for the duration of their relationship and was secretly jealous of Percy’s dalliances. Her writings reveal that Mary struggled with depression and visions of her lost baby, but Mary gave birth to a son, William, in January 1816. That summer, she, Percy, their son, and Claire joined Byron and a young physician named John Polidori in Geneva.

Byron proposed his famous challenge while the group was sitting around the fire at the villa reading German ghost stories. Polidori ended up writing a short story called “The Vampyre,” but Mary struggled to find inspiration, until a chance discussion on the nature of life and the science of galvanism stirred her creative juices. In the early hours of June 26, Shelley experienced a “waking dream,” as moonlight “struggled to get through” the closed shutters in her room.

As she recalled in the 1831 introduction to Frankenstein:

I saw the pale student of unhallowed arts kneeling beside the thing he had put together. I saw the hideous phantasm of a man stretched out, and then, on the working of some powerful engine, show signs of life, and stir with an uneasy, half-vital motion. Frightful must it be; for supremely frightful would be the effect of any human endeavour to mock the stupendous mechanism of the Creator of the world.

Frankenstein; or, The Modern Prometheus was initially a short story, but Mary expanded it to a full-length novel at Percy’s urging. It was published anonymously in January 1818, mostly to critical acclaim. Mary was not identified as the author until the publication of the second edition in 1823, so many people initially assumed it had been written by Percy.

Despite her literary success, Mary’s life was checkered by multiple tragedies. Both Mary’s half-sister, Fanny, and Percy’s estranged wife committed suicide—Fanny by a laudanum overdose, Harriet by drowning. Percy and Mary got married shortly after Harriet’s death, but despite several pregnancies, only one child survived to adulthood: Percy Florence. In the summer of 1882, while in Italy, Mary miscarried yet again and nearly died from loss of blood. A quick-thinking Percy placed her in an ice bath to staunch the bleeding and likely saved her life. Alas, Percy drowned in a boating accident later that same summer, devastating an already depressed Mary.

Frankenstein is the book for which she is justly famous, but she built a fine literary career as a writer and editor. Shelley never remarried, despite the occasionally suitor, and died on February 1, 1851, at 53, possibly from a brain tumor.

Most of the above aspects of Shelley’s life find their way into A Nightmare Wakes—Unkel strove to be historically accurate even with regard to the lighting and production design—albeit reimagined and condensed for narrative purposes, since most of the film takes place in the summer of 1816. In this telling, Mary is pregnant with her first child when she, Percy, and Claire arrive in Geneva, and she tragically miscarries. Out of this tragedy comes the inspiration for Victor Frankenstein, driven to create a Creature stitched together from dead cadavers and “reanimated” during a dramatic thunderstorm. Philippe Bowgen plays Byron, Claire Glassford plays Claire Clairmont, and Lee Garrett plays Polidori.

“Shelley’s struggle with love, loss, abandonment by society and family, and her own sanity, had yet to be captured fully on-screen,” Unkel said of what drove her to make the film. “She lived a colorful life of love, drugs, and freedom, alongside some of the most celebrated artists of her day.” Ars sat down with Unkel to learn more.

Continue Reading