The Future Comes by Break-In

The Bulgarian physicist who mapped the AI century from a website that no longer exists — and did not live to see the world catch up.

A forecast with a print run of 200

One of the most uncanny forecasts of the AI century was not written at MIT, or at a think tank, or by anyone you have heard of. It was written over two days in August 2003, in Bulgarian, as conference theses for a local science-fiction convention, by a physicist named Ivan Popov — and published on a hobbyist website that today exists only in the Internet Archive.

Popov was born in Sofia in 1970. He studied physics and spent three years at the Space Research Institute of the Bulgarian Academy of Sciences, in a department that was experimenting with digitization and early machine learning — neural networks, in the years when that phrase drew blank stares even in the West. He left science to become a literary translator: it is through him that Bulgarian readers got Stanisław Lem, Victor Pelevin and Nick Bostrom. In 2004 he published what is often described as the first Bulgarian cyberpunk novel, “Hackers of Human Souls.” It won the Eurocon debut award. Its print run was two hundred copies.

That novel, and one essay — “The Future Comes by Break-In” — are the core of one of the strangest predictive records I have ever audited.

Two futurologies

Popov’s essay opens with an epigraph he borrowed from the Russian futurologists Sergey Pereslegin and Andrey Stolyarov: “For 90% of people, the arrival of the future looks like the end of the world.”

There are, Popov argues, two kinds of futurology. Classical futurology extrapolates: it takes the present’s curves and extends them. It is always wrong, he says — the extended present never coincides with any version of the real future. (His example: Mendeleev’s careful 19th-century projection that the cities of Europe would drown in horse manure.)

Non-classical futurology does something else. It hunts for what Popov calls locuses of the future: elements already present today — technologies, ideas, practices — that are incompatible with the present’s line of development. Borrowing an image from the Strugatsky brothers, he calls them “tentacles the future has pushed into the present.” The present, as a rule, fights these tentacles and tries to cut them off.

From this he derives the sentence that organizes everything else he wrote:

“The real future is not what we strive for, but what we fight against.” (translation mine, as are all quotes below)

The future, in Popov’s model, does not arrive by consensus. It breaks in — through civilizational crisis — and it first appears, to the people of the present, as madness or as deep, irreversible degradation. Every major transition in history looked that way from inside: the accumulating burgher was a degenerate miser to medieval eyes. And it breaks in at the periphery, not the center — because the center always has the regulatory strength to cut the tentacles, and the periphery does not.

The receipts

What makes this more than elegant philosophy is what Popov did with it. Writing in 2003–2004, he worked out, in cold specifics, what the “information phase” transition would look like. Judge the record:

The collapse of the price of intelligence. If the industrial phase meant machine production of goods, Popov reasoned, the information phase means machine production of knowledge. And exactly as machine production proletarianized the weaver, AI would proletarianize the knowledge worker: “The prestige and pay of most intellectual professions will fall sharply… they will work with computer systems and merely press the buttons of the AIs.” He predicted “proletarians of mental labor” and falling costs of innovation itself — in 2003, twenty years before the phrase “prompt engineer” existed, he described both the job and its wage curve.

The shape of AI. “The history of intelligent machines began with a heap of idiotic delusions,” a character in his novel says. “They thought it had to reach human-like intelligence. But what the devil for?!” In the essay he is precise: expect “arbitrary types of intelligence, including maximally non-human ones,” and he dismisses the humanoid-AI program as the equivalent of the 18th century’s clockwork androids — exquisite toys built while the actually transformative machine, the filthy, hissing steam engine, was reorganizing society from below. It is difficult to read this in 2026, watching alien, non-anthropomorphic language models restructure white-collar work, without a shiver.

The fragmentation of shared reality. As communication density grows, he wrote, the cost of reaching society rises — “the more people shout, the lower the audibility of any single message” — until only a handful of large agents can run mass campaigns, and society decomposes into “information cocoons” with “weakly connected reality tunnels.” The term “filter bubble” would become mainstream with Eli Pariser’s 2011 book, years after Popov’s phrasing.

Bots all the way down. In the novel (2004), a character discovers his personal-communication bot has been switched off for a month: his mother in Chicago is used to “his” calls every few days — though she may have delegated her end to a bot as well. “Nowadays it’s fashionable for bots to talk to each other, since people have no time.” This is a punchline about 2026 written when Mark Zuckerberg was a sophomore.

Language as the weapon. The novel’s thesis — “Man is a device for processing information. And an object of control by means of that same information” — with language as the primary instrument of social engineering, is the entire modern discourse on memetic warfare, LLM persuasion and platform manipulation, compressed. Yuval Harari would later popularize the theme globally as “hacking humans” (Davos, 2018). Popov had titled his novel “Hackers of Human Souls” fourteen years earlier.

Industrial feudalism. Among his crisis-exit scenarios: a partial “naturalization” of the industrial economy — demarketization, dependence, lords and serfs of infrastructure — which he provisionally called “industrial feudalism.” The word “technofeudalism” entered the mainstream lexicon around 2020–2023. His periphery principle, in the primary essay, points to the weak but technologically receptive peripheries of the West — Eastern Europe and the Balkans, Latin America; a later-cited version of the text names Ukraine, Moldova and Colombia. The war in Ukraine — today’s laboratory of battlefield autonomy — has made the periphery thesis newly suggestive, though as prediction it is loose rather than precise.

The compression of knowledge. Perhaps his deepest and least noticed idea: aging civilizations die of information overload — “explosive complexification with falling efficiency” — and the transition requires what systems theory calls secondary simplification: radical compactification of the infosphere without loss of function, including a “metatechnology” for re-deriving lost knowledge on demand instead of maintaining it all. If you want a 2004 description of what large language models actually are — a lossy but functional compression of civilization’s textual accumulations, with regeneration on demand — you will not find a better one.

What he got wrong

An honest audit lists the misses. Psychedelic technologies as a “threshold technology” of the transition: no. Biofeedback games as the dominant art form of the future: no. The deterministic crisis-machinery — phases, inevitable leaps, “3 to 6 great transitions” — is inherited scaffolding from the Russian systemological school (Pereslegin above all), which habitually presents bold analogies as laws. And the strong form of his periphery thesis has a problem: the LLM future consolidated first in the very center of the old world, three blocks from Stanford, not in Moldova.

Read him, as a colleague of mine put it, as a conceptual radar, not a calendar.

The lens, pointed at 2026–2035

But the radar still works, and pointing it forward is the best way to honor it. Popov’s test: ignore the roadmaps; list what the present is fighting hardest; assume the future is on that list, disguised as degradation.

What sits on that list today? Autonomous agents with money and legal agency. Open-weight frontier models. Cognitive and pharmacological enhancement. Battlefield autonomy. AI-mediated relationships and outsourced memory. Every one of them is described, by serious people, in exactly the vocabulary Popov predicted the present would use: madness, degradation, threat.

His model’s near-term implications, stated as predictions: the proletarianization of cognitive work will complete, and status will migrate away from credentialed knowledge toward whatever machines cannot yet certify. Institutions will quietly stop maintaining full knowledge stacks — curricula, documentation, mid-tier expertise — and keep machine-mediated regeneration instead; from inside, this will look like the collapse of education, and it will be secondary simplification. Bot-mediated social life will cross from shameful to normal, complete with etiquette. And the new social forms will consolidate first where regulation is weakest — watch the peripheries.

The quiet departure

Ivan Popov withdrew from literary life around 2011. One story appeared in English — Sci Phi Journal, 2015 — and was reprinted in the 2016 “Up and Coming” anthology of Campbell-eligible authors. He died in August 2019, at forty-nine.

I will not pretend to know the private shape of his final years, and nothing about his death changes anything about his ideas. What can be said safely is this: the work survived him, and people who see too far ahead are often too easy to miss while they are alive.

Coda: he was the locus

In Popov’s own model, the future is carried by locuses — elements incompatible with their present, which the present starves, represses and fails to recognize, until a crisis proves them right. By that definition, Ivan Popov was a locus of the future: a mind the present was not ready to recognize, published at the edge of visibility.

The world his essay described became legible three years after he left. He was not late. We were.

(In 2020 his father compiled his essays into a collection — “Cyberpunk, Heuristics, Futurology” — which has no price: it is only gifted or exchanged. A samizdat from the digital age. If you can translate, publish, or simply read him — the door is open. Start with Georgi Karamanev’s profile below.)

— -

Sources and further reading

• The essay, part I (Bulgacon-2003 theses, Aug 2003), Wayback Machine: https://web.archive.org/web/20050206070122/http://www.bukvite.com/k-e-f/matr/ip_vzlom.htm
• The essay, part II (Eurocon-2004 notes), Wayback Machine: https://web.archive.org/web/20050206064304/http://www.bukvite.com/k-e-f/matr/ip_eucon.htm
• The essay, live mirror: https://zemlemoria.org/aqualonia_files/2004_files/future.htm
• His K.E.F. site (cyberpunk, heuristics, futurology), archived: https://web.archive.org/web/20070202195246/http://www.bukvite.com/k-e-f/
• Georgi Karamanev’s profile “Ivan Popov, Who Caught the Future” (Bulgarian, 2025): https://karamanev.me/istorii/ivan-popov
• Bibliography (Bulgarian SF wiki): http://bgf.zavinagi.org/index.php/Иван_Попов
• His story in English: Sci Phi Journal #5 (May 2015); reprinted in “Up and Coming: Stories by the 2016 Campbell-Eligible Authors”

All quotations are my working translations from the Bulgarian originals.