/newwave/ fiction review- 'Quarter Lives' by Cairo Smith
Deepseek, agentic competition, and the simulation of agency as a generator of meaning
This week, I’ve been busy messing around with Deepseek, the new Chinese open source LLM that’s causing a massive stir in Silicon Valley and tech circles because of how cheap and good it is.
The significance of Deepseek is multi-fold:
(a) It’s comparable to frontier American models from OpenAI and Anthropic, disproving the notion that the PRC was greatly behind the US on frontier AI,
(b) Many prominent techbros are claiming that the Deepseek team lied about their extremely low training costs, but the SV-consensus on Chinese innovation capacity appears to be shifting. The new narrative as Chinese corporations being ‘live players’ undermines widely-held notions that Chinese people in China are (racially/culturally/institutionally) incapable of intelligent creativity and are only capable of g-loaded tasks so long as they’re strictly mental computations related to IP-theft,
(c) Because Deepseek is so good (and cheap), it may threaten the valuations of American tech companies who’ve invested enormous capex into huge data-center buildouts, potentially causing the NASDAQ to crash due to a form of market-driven narrative collapse,
(d) It suggests American hegemony, which relies on American technological superiority (particularly in AI), is now increasingly likely to erode in the face of Chinese technocapital acceleration (“Neo-China arrives from the future,” etc.).
(e) Given (d), PLA supersoldiers in flying mecha-suits are now only months away from occupying Washington, D.C.
(f) Alternatively, the AGI arms race is now in full swing and both American and Chinese labs are going to birth a machine god that’s going to annihilate the human race.
I suspect most of the hypotheses above going to come true (I’ll let you figure out which ones).
My base case is now as follows: barring a scenario where the US achieves a decisive strategic advantage in frontier AI (e.g. a ‘superintelligence) and immediately deploys it as a weapon to destroy the Chinese state, I expect the PRC to rapidly stack a series of increasingly-obvious Sputnik moments that become progressively harder to deny until they reach technological parity in every tech-tree of geopolitical and economic importance.
I don’t have a clear intuition about what the future is going to look like, but I think the arrival of Deepseek is something akin to a psychological “future shock.”
I am of the mind that we are about to experience a sequence of staggeringly disruptive technological, social, and economic changes that will be maximally disorienting in just a short number of years. (Maybe they’ll add “Future Shock Syndrome” as a novel anxiety disorder to the ever-expanding Pokédex of the DSM.1)
Let’s set aside the question of who will “win” the AGI/ASI arms-race (and it is now emphatically a race).2
At this point, if you have even a notional familiarity with the capabilities of frontier models, it feels very likely that mass technological unemployment is probably only several years away for huge swathes of white-collar workers (manual laborers and skilled trades might take longer, but advances in humanoid robotics will eventually hit them too).
If you’re reading this post, it’s likely that you will have zero marginal economic utility to the economy within single-digit years.
This is a transition of unfathomable importance.
My critique of liberalism is that liberalism is largely just utilitarianism and utilitarianism inexorably devolves into wire-heading, i.e., into cumming-as-much-as-possible.
The trajectory of Western cultural development is such that we’ve sequenced through two epochal meaning-collapses: first, with the death of God from Darwinian evolution, Biblical errancy, and secularism, second, with the death of romantic love due to social atomization and commoditization effects from modern-dating dynamics.
What meaning still remains is a third, largely-untouched pillar: work, our ability to hustle and grind, to fulfill our purpose (a deeply held committment to making-the-number-go-up).
Yes: what remains are our careers.
And now, these, too, will be taken from us.
Yikes!
I’m now several months out from when I promised Cairo Smith a review of his short-story collection, Quarter-Lives. Apologies to Smith for the delay—as you can see, I’ve been preoccupied with a lot of doomscrolling.
Quarter Lives is a short-story collection that presents itself in the right mode for indie literature: a short, light read that’s meant to be fun—genre-fare written with a decidedly cinematic tint to it. I read it on short commutes around town.
Smith is a filmmaker, and his visually-loaded imagination is both a strength and a limitation: the stories, while generally enjoyable and good, often read like screenplay adaptations rather than more obsessively-crafted works of literary fiction where writers spend years with sentence-optimization.3 Since I’m a sentence-optimization-autist, I typically prefer works with a lot of the latter.
Note that this isn’t a critique of the book so much as it is an observation about genre-work in general: I liked the book, but I often couldn’t help but thinking that what I really wanted to see was a televisual adaptation of these stories (‘Spirit of ‘87, a youthfully charming ghost-story with an 80’s vibe, is something that would take well to an adaptation).
Overall, I’d say that Smith’s general strength as a stortyeller lies in narrative metastructure. If literary structure comes down to scene composition from balancing narration with description and dialogue, literary metastructure comes down to the desire you have to keep reading forward. He has an appropriate sense of pacing, and never lingers for too long.
(This is something literary fiction writers often lack because they overoptimize for the sentence-level rather than the macro-level).
And then there was one
The truth is that there is one standout story in the collection: Network States.
You can read it for free here, but I’d encourage you to support the author by grabbing a copy of the book on Amazon.
Network States, was, to my first impression, likely to be a narrative account of Balaji Srinivasan’s model for future permutations of statehood that start out as spatially-distributed, online communities and only later cohere into IRL-land-states.
What it’s actually about is something that is much more interesting: a model of the future where human beings derive most of their value from a simulated world that increases their agency in that plane.
To simplify, Network States is an inventive faux-journalistic piece about one man’s account of a complex WW2 simulation conducted online and the low-status extremely-online IRL autists who are taking it way too seriously.
The story itself is well-executed, funny, and captures a counterintuitive layer of joy around internet culture in a way that’s feels youthful and charming.
It’s salience lies in its premonition about the future: if most human beings in base reality will have zero economic value to contribute, a unitary descent into wireheading is going to similarly drive them insane.
What is to be done about this problem?
Smith suggests an answer.
If human beings cannot generally engage in agentic competition in the base reality of the economy, then the simulation of agency will be necessary to provide their lives with meaning. For this reason, the faux-MMORPG world of Network States and its overly-enthusiastic community of WW2-simulation competitors isn’t a mere joke.
Rather, it’s a model for meaning creation, and a possible avenue into humanity’s primary coping mechanism for the future.
What Network States anticipates is a world in which we cope by ascending/descending into the realm of persistent simulation where the individual person still matters because of their effect on that layer of reality.
For this reason alone, I think the book is well worth your time—and your pondering.
This joke was plagiarized from my novel.
"Winning” in this scenario might mean “we’re all dead,” lol.
The front half is much stronger than the back half.
Interesting post on the AI side,if interested: https://youtubetranscriptoptimizer.com/blog/05_the_short_case_for_nvda
Ordered Quarter Lives. Sounds good. Hope I get to it ...