Quantum Thief (The Quantum Thief, 2010; The Fractal Prince, 2012; The Causal Angel, 2014; TvTropes) by Hannu Rajaniemi ( Twitter) is a hard SF trilogy set in a post-Singularity Solar System (cf. Eclipse Phase/Orion’s Arm), starring the distant descendant of Maurice Leblanc’s gentleman jewel-thief Arsène Lupin, a certain amnesiac Jean le Flambeur.

It follows him as he escapes from prison and attempts to retrieve his memories hidden in various places throughout the Solar System; each novel focuses on one world-building setpiece: Quantum Thief is cryptopunk Venice; The Fractal Prince is demoscene Baghdad; and The Causal Angel is MMORPG space cities. Le Flambeur succeeds in stealing back his old self, only to discover his old self was entangled in a web of plots by all the major players fighting over a ‘jewel’ of universe-destroying power, destabilizing the fragile balance of power, and sparking a Solar System-wide war which causes the end of the universe—but don’t worry, it all works out OK (mostly).

I read Quantum Thief in May 2014, and Fractal Prince/Causal Angel in January 2015, and enjoyed it, but was left too confused to write a proper review. I could only capsule-summarize it as “Accelerando meets Lupin; uncompromising SF, stuffed full of interesting tidbits—the game-theoretic prison at the beginning is only the beginning; self-recommending.” In August 2022, after visiting DEFCON, I caught an unpleasant case of COVID, and while discussing the best remedies¹ with a ML researcher, I recommended Quantum Thief as a good read but admitted I didn’t get it, in part because it was so in media res.

I was not up to my usual work, so I thought I’d take a second whack. I was right, and the trilogy made sense to me the second time around. So, this is my attempt to explain The Quantum Thief Trilogy, its “emotional core” (to quote Hannu).

Stylistically, QT is relentlessly in media res: neither we nor le Flambeur know why he is in prison, and little is explained thereafter. Hannu makes no concessions to the casual reader, as he mainlines straight into his veins the pre-deep-learning 2010-era transhumanist zeitgeist via Silicon Valley—if it was ever discussed in a late-night bull session after a Singularity University conference, it might pop up here. Hannu stuffs the novels with blink-and-you’ll-miss-it ideas on the level of Olaf Stapeldon. A conventional Verne gun is too easy a way of getting to space—how about beating Project Orion by instead using a nuclear space gun (since emulated brains don’t care about high g acceleration)? Or for example, the All-Defector reveals that, since other universes could be rewriting their rules to expand at maximum speed, erasing other universes before they know it, he plans to rewrite our universe’s rule to do so first (ie. he will defect at the multiversal level against all other universes); whereas beginner-level SF like The Three Body Problem would dilate on this for half a book, Hannu’s grand reveal gets all of 2 paragraphs before crashing into the eucatastrophic ending.

For world-building, he drops neologisms left and right, and hard ones at that—few enough American readers will be familiar with the starting premise of “Arsène Lupin in spaaaace!” (probably more are familiar with the anime Lupin The Third these days), but his expectations go far beyond that: the ideal reader of the trilogy is not merely one familiar with the Prisoner’s Dilemma but also with the bizarre zero-determinant PD strategies discovered ~2008, and not just with such basic physics as quantum entanglement or applications like quantum dots, but exotic applications to quantum auctions & game theory (including Prisoner’s Dilemma) & pseudo-telepathy (yes, those are things), and it would definitely be helpful if that reader happened to also be familiar with Eliezer Yudkowsky’s c. 2000s writings on “Coherent Extrapolated Volition”, with a dash of Nikolai Fyodorovich Fyodorov’s Russian Cosmism for seasoning (although only a dash²).

This leads to an irony: I noted while reading Masamune Shirow’s Ghost in the Shell cyberpunk manga that almost everything technical in the GitS manga turned out to be nonsense despite Shirow’s pretensions to in-depth research & meticulous attention to detail; while in QT, most technical things sound like cyberpunk nonsense and Hannu doesn’t insert any editorial notes like Shirow does to defend them, but are actually real and just so arcane you haven’t heard of them.

For example, some readers accuse Hannu of relying on FTL communication via quantum entanglement, which is bad physics; but Hannu does not! If they had read more closely (similar to the standard reader failure to understand the physics of “Story of Your Life”), they would have noticed that at no point is there communication faster-than-light, only coordination faster-than-light—‘spooky action at a distance’. He is instead employing advanced forms of quantum entanglement which enable things like secret auctions or for coordinated strategies of game-playing (quantum coordination, like treating the particle measurements as flipping a coin and one person does the ‘Heads’ strategy and the other person does the ‘Tails’ strategy does not require communication, obviously, but surprisingly, quantum coordination can be superior to all apparently-equivalent communication-free classical strategies). He explains briefly that the zoku use quantum entanglement in these ways, but a reader could easily miss that, given all the other things they are trying to understand and how common ‘quantum woo’ is.³

So, in such hard SF, if we do not handwave some sort of magical DRM for AIs, we face serious world-building problems with AIs and mind uploading: assuming we can understand the entities at all⁴, how do we tell meaningful stories about the entities in that world? In a world where minds can be copied & forked indefinitely over time, how do you maintain personal identity—not necessarily in any philosophical sense, but simply pragmatically, as a security matter? Cryptographic keys can be copied, quantum keys can be stolen, memorized passwords extracted, trusted hardware cracked, and so on.

The zoku are anarchic organizations descended from MMORPG ‘clans’.⁵ Their use of quantum entanglement represent one solution: ‘you’ are defined by your unique physically-uncopyable quantum states, which can be used for entanglement & secure communication, and your jewels slowly brainwash you into an appendage of the zoku (immortality, of a sort). The zoku promiscuously use their ‘jewels’, accepting as a fact of life that there will be individual losses such as theft (hence the eponymous “quantum thief”), but the zoku itself lives on.

But jewels (and the sacrifice of an autonomous self) are too restrictive for a thief like le Flambeur, while the Sobornost are committed to using only technologies that allow copies which can be preserved forever (ruling out all quantum technology with religious fervor), enslaving minds forever to whoever runs them.

So Hannu’s answer for them harks back to that classic of American SF TV, Babylon V (1993–1998). In B5, characters are interrogated by the benevolent status-quo overlords with the “Vorlon Question”: Who are you? And their opposites, the changing Shadows, ask instead the “Shadow Question”: What do you want?

Hannu’s answer to the personal identity question is the Shadow Question—that Jean & the Sobornost Founders & the zoku elders are all defined by what, at their core, they want. Anyone who wants the same thing is, for all (their) intents and purposes, the same person as them; because they want the same unchanging things, they can be trusted as the original. The ‘Founder codes’, and Jean’s final password to unlock his sealed memories, are all memories of what defines their wants: the Founder Sumanguru wants blood & fire & electricity & screaming children, and enemies to destroy; the Founder Chen recall the trauma of livestreaming their father’s assassination, remaining eternally resolved that the last enemy that shall be defeated is death; while seared into the minds of the Founder Joséphine Pellegrinis is the final thought of their founder, her desperate dying wish that her lover Jean le Flambeur someday return to her… (And the zoku elders want to empower their zoku clans.)

But even personal identity frays under the power of time: given freedom to change, sooner or later, like the Ship of Theseus, the mind which sets out is not the mind which arrives. So the price of immortality must be that one cannot change: one is condemned to want the same things, forever.⁶ (“There is no prison, except in your mind.”) Joséphine Pellegrini cannot stop seeking after her lost Jean—nor can Jean stop his thieving nor trying to escape her, because le Flambeur, what does Jean le Flambeur remember?

He remembers that every wall has a door; that every system has a flaw; every proof its assumptions, however poor; and whether man or physics, there is no absolute law. He remembers long ago an impoverished boy locked in a desert jail cell, wanting freedom, and certain of only one truth (repeated throughout): “There is always a way out.”

The Quantum Thief trilogy is not a jewel theft cat-burglar story, but its opposite: a prison escape-artist story—from the universe!

It is the opposite, because prison is, of course, the final destination of the thief eternally pursued by the detective. And the escape from prison is the inverse of being sent to prison: in a heist, the thief starts on the outside of a guarded space, wealthy, free, with access to all the resources of the wide world and a lifetime of allies, using his brain to single-mindedly plot himself into the goal of the inaccessible space with the precious object; but in a prison escape, the thief is the precious object and already in the guarded space, and must, naked, solely with his brain, unsteal himself from the prison and vanish back into the goal of the wide world. The Kaminari Jewel, it turns out, is not the goal. It is only a door to step through to the goal—freedom.

The Kaminari Jewel, the wish-granting Grail sought by all parties, is omniscient because it traces causality, cannot be fooled, and rejects those whose ideals are unworthy to rewrite the universe. How then does one become worthy, if one’s wants, one’s identity, are impure, and one will not, cannot, change without ‘dying’? Then someone else must use the Jewel, someone who is pure. Thus the overlapping plots: Founder Chen seeks the lost copy of his childhood, still pure and untraumatized; the zoku elder seeks a way out of the zoku, to be allowed to want something other than the prosperity of the zoku; Founder Pellegrini places her last hope in a pure young knight(ess) worthy of the grail, of her making but gifted free will; and the weary le Flambeur discards his past yet again and arranges for his imprisonment in the impregnable Dilemma Prison (but—there is always a way out…), the only place where he may be able to change into a kinder le Flambeur, able to discard his past, and become worthy of the Jewel.

And at the end? The Jewel is invoked. The world ends. Everyone gets what they wanted, good and hard, in their own pocket universe: the Sobornost are left behind in a classical universe to pursue the Great Common Task and ensure nothing is lost or forgotten again, the zoku get a new universe even more suitable for game-playing, and as for Jean le Flambeur?

A door opens for all the le Flambeurs, even the ones still trapped in the Dilemma Prison. Where does that door lead? It doesn’t matter, and the readers will never know.⁷ The gentleman thief walks toward it, as he knew he would, “whistling as he goes”, because, as the last line of The Quantum Thief Trilogy reminds us:

---

1. Summary: phenylephedrine is a gross imposture practiced on the American public as an ineffective placebo offered in place of pseudoephedrine. You want pseudoephedrine, paracetamol, guaifenesin, pholcodine/dextromethorphan, and honey—I wound up going through a whole bottle of honey in my second COVID case because my coughing & throat irritation became so bad.↩︎

2. Hannu tells me he learned of Cosmism from blog posts in the mid-2000s like Nader Elhefnawy’s citing of a (2002?) translation of a Nikolai Alexandrovich Berdyaev essay (and later, John Gray’s 2011 The Immortalization Commission), and recalls talking about Cosmism & QT with Charles Stross. (They met through Writers’ Bloc while Hannu was a grad student in Edinburgh.) Stross helped further popularize Cosmism by pushing the odd claim that Western transhumanism does not merely share a few superficial similarities with it, but is historically descended from and is a secret Russian Cosmism.

   The ignorance of Hannu’s readers makes the case against this claim, and in any case, hardly any of Federov’s material was translated or discussed in the West in the relevant time period (his major work not until 1990) so it could not have affected Western transhumanism much (nor does Stross present any evidence that anyone in the West knew or cared, making only a post hoc ergo propter hoc argument)—you will look long & hard through Robert Ettinger’s books (inventor of cryonics) for any trace (or J. B. S. Haldane or J. D. Bernal or Aldous Huxley or…), or any such claim by Fedorov scholars reviewing his influence, like Koutaissoff & Minto1990, which is silent about any influence on the West—because there was none. (As one enthusiast puts it, “practically nobody outside of the country had heard of him until well into the 1970s.”)

   Hannu does not endorse this thesis, presenting Cosmism as a much later influence, and it serves its defamilarizing purpose of not being Western transhumanism but a different hypothetical ideology/morality, (which helps the worldbuilding by avoiding degeneration into a simplistic ‘red vs blue’ morality play).↩︎

3. This makes QT an example of “things you can’t countersignal”: because it looks like such a common abuse of quantum mechanics, and the narrative doesn’t infodump on it, uncharitable or ignorant readers will naturally assume that Hannu is doing the bad thing, instead of wondering if he was doing a good thing which is simply beyond their understanding.

   For comparison, many readers would assume that much of Peter Watts’s novels make up stuff like the ‘screaming faces’ or the superintelligent insect-mind, but Watts includes appendixes with references “showing his work”; while a reader of Greg Egan’s Orthogonality will never be doubtful because the novels spend so much time explaining the novel physics & why it would do what Egan says it does (and referencing his website with exhaustive exploration of the mathematics). As much as one hates to explain the joke, it is probably best to explain it lest one be traduced by reviewers online–more so, that is.

   A third way I have experimented with is my AI hard takeoff short story, “It Looks Like You’re Trying to Take Over the World”. The purpose of the story was mostly to list references I thought were interesting, and cobble them together into the thin veneer of a short story. This was a dilemma: if I left the links in, they were quite distracting and defeated the point of fictionalization; but if I left them out, readers would usually assume that almost everything was made-up fever dreams. We considered providing two copies of the story, first without and then with, but that would make editing it difficult and was inelegant.

   We recalled that readers of Gwern.net had complained that the heavy hyperlinking was often a distraction and they were using (not always successfully) various ‘readability’ or ‘reader-mode’ gadgets, so we took the opportunity to create a “reader mode” which hide all links—and then reader-mode was simply automatically enabled on the story at the beginning & disabled at the end, as a “punchline”. We don’t know how well it worked because readers never tell you about that sort of thing, but we think it’s neat. (As an amusing & instructive side-effect, this auto-reader-mode led to a number of amusing instances of readers—often machine learning experts—scoffing at parts of the story as absurd, but which had multiple references. They were simply both ignorant & had failed to notice. Machine learning progress & publication volume has been so rapid over the past decade that very few people have even an overview of all relevant material.)↩︎

4. SF giant Vernor Vinge died shortly after I finished this review; one of the things that makes both Vinge & Hannu’s SF hard is that they take seriously the idea that human brains are not some sort of ultimate limit to intelligence, and that the progress of technology may result in minds which are vastly more intelligent than humans in the same way that there vehicles which are vastly faster than humans can run, or further beyond human understanding than human activities like the global economy are beyond the understanding of a chimpanzee. If you accept that brains can be emulated or AIs created, why is there any humanly-comprehensible story? As SF editor John Campbell told Vinge when rejecting a proposal for the human-level sequel his first monkey-level ‘singularity’ story: “You can’t write this story. Neither can anyone else.”

   Vinge famously invents new ‘physics’ ruling out superhuman intelligence, providing refuges where the story can play out. Hannu takes a different tack: in QT, superhuman intelligence is possible, but is too possible—unleashing the full potential of optimization results in inhuman “dragon” superintelligences, which sacrifice all human values in the pursuit of power & growth. The results are of no moral value, and are too powerful to use except as a last resort—the in-universe equivalent of nuclear bombs. (Indeed, it is implied that one reason the Sobornost have not conquered the solar system is that they are distracted by containing the existing dragons, like the one ultimately deployed to destroy Earth.) All factions mutually agree to a fragile equilibrium of banning anything which could become an abominable dragon, and to build everything out of human-like brains or ordinary non-AI software.↩︎

5. Stated in passing, and Hannu throws in a few obscure MMORPG references. For those wondering what that ‘sleeper legend’ allusion was, it was to an EverQuest dragon designed to be unkillable—but players did anyway in 2003.↩︎

6. Hannu in 2010 on the themes:

   The plot was inspired by one of Rajaniemi’s favourite characters as a teenager…what intrigued Rajaniemi were the cycles of redemption and relapse Lupin goes through as he tries to go straight, always falling gloriously short. At various stages in his career Lupin gets married, joins the French Foreign Legion and even spends 5 years as a chief of police, investigating himself—but always finds himself drawn to life on the wrong side of the law.

   “It’s somehow clear that he’s never going to succeed, that the pull of the illicit is too strong, that somehow fundamentally, he is Arsène Lupin—though of course Arsène Lupin is not even his real name. Arsène Lupin is the identity he has created for himself, but he can no longer escape it.” This underlying thread offers an element of tragedy, Rajaniemi says. “He can break the rules, but in the end he’s also imprisoned by some higher order of rules, rules of identity.” What would become of Lupin in a future where people really can switch identities or bodies? Could he change at heart? Could he actually redeem himself? “That became the central theme not only of the first book, but probably the sequels that will be forthcoming.”

   …After examining questions of identity through memory in The Quantum Thief, the second installment will revolve around how we construct stories of ourselves…with a third in prospect that will focus on how game-playing allows us to adopt radically different selves.

   ↩︎

7. One is reminded of one of the other famous exits in SF, the escape of Paul Atreides from the trap of prescience at the end of Dune Messiah: “Now I am free.”↩︎

Similar Links

[Similar links by topic]

Link Bibliography

[Bibliography of links/references used in page]

[ Send Anonymous Feedback ]

[adblock public service announcement]