The Internet and the "Empty Cave"
This article was written with the assistance of a chatbot. Japanese
The Story That AI Made the Internet Boring
Lately, I keep hearing that "AI has made the internet boring."
Morley Robertson, who recently passed away, wrote much the same thing in what turned out to be his final column for Weekly Playboy[1]. You search for something and all you get are content farms and SEO articles—you can never reach what you actually want to know. AI-generated summaries have become ubiquitous, and context is vanishing from all information.
As if the old internet was ever that great.
A Prophet in 1997
There was an astronomer named Clifford Stoll. In 1995, he wrote a book called Silicon Snake Oil (published in Japanese as The Internet Is an Empty Cave)[2].
Stoll was no technophobic outsider. He'd been a heavy user of Unix and networking since the 1970s. He even tracked down and helped arrest a KGB hacker—an episode he recounted in The Cuckoo's Egg[3]. In other words, this was someone on the inside of the internet, criticizing it from within.
The original title, Silicon Snake Oil, is a reference to the 19th-century American con men who peddled "snake oil" as a miracle cure-all. Stoll's implication: the tech industry is hawking the internet as its own brand of snake oil.
For a long time, this book was dismissed as a "failed prophecy." His technical complaints—slow modems, scarce content—had become outdated. He apparently even argued that word-processed résumés lacked the human touch of handwritten ones, which, admittedly, is pretty funny[4].
But reading it again in 2026, what the book was really trying to say lies elsewhere entirely.
It Wasn't the Cave That Was Empty—It Was Us
If you had to sum up Stoll's central concern in one sentence, it would be this:
"A system that provides easy answers robs us of the ability to think."
He recounts something he learned from an astronomer during a stay in China[5]: The hard part isn't performing a Fourier transform or operating a program. The hard part is understanding your data and figuring out what to do with it.
A computer lets you run through trial and error. But the computer itself has no insight. What cultivates that kind of thinking is reading, writing, analysis, and teachers.
In other words, Stoll wasn't saying "there's no content on the internet." Regardless of how much content exists, the internet as a medium is structurally geared toward mass-producing easily accessible answers, and that structure makes human thinking shallower.
"The empty cave" was never about a lack of content. It was Plato's cave allegory in its purest form: we who mistake the shadows on the wall for reality are the ones who are "empty."
The Shadows Have Only Grown More Convincing
What changed in thirty years?
Online content exploded. We got arXiv, GitHub, access to every kind of primary source imaginable. Compared to Stoll's era, the number of shadows on the cave wall has multiplied ten-thousandfold.
But the more shadows there are, the harder it's become to find the real fire.
Search engines were corrupted by SEO. Social media fell under the rule of algorithms. And then AI arrived, offering pre-digested summaries as "overview" answers. The "uniform answers with no room for reflection" that Morley Robertson lamented are the logical endpoint of these thirty years.
But this is not a problem AI created. AI is merely an accelerator. The desire to "efficiently consume nothing but answers" was in us from the start, and technology has simply been faithfully serving that desire all along.
Content farms, SEO articles, AI summaries—they're all products of the same impulse. Skip the tedious context, just hand me the conclusion. Good or evil, dangerous or safe, buy or sell. Binary answers, delivered at maximum speed.
Stoll saw through it in 1995.
Was That Encounter Really by Chance?
People say, "The old internet was full of serendipitous discoveries."
But were they really serendipitous?
Kobayashi Hideo wrote about his encounter with Rimbaud as though he'd been struck by lightning[6]. But Kobayashi was studying French literature at Tokyo Imperial University. He was sitting at the very center of the French literary world in Japan, embedded in its reading circles and personal networks. His finding Rimbaud wasn't so much an "encounter" as an inevitability within that context.
The internet of the '90s was the same. People who "stumbled upon" interesting sites were already situated in a context that made such discoveries possible. They had the literacy, knew how to search, and could judge the value of what they found. What looked like chance was really the gravitational pull of a prepared mind. (Okay, maybe I'm romanticizing this a bit.)
If that's the case, then the internet wasn't interesting because of the internet. It was interesting because there were interesting people using it. And if the internet has become boring now, that's not the internet's fault either.
This hasn't changed. Accepting whatever answer AI hands back to you is no different from reading only the first search result and thinking "got it." The tool changed; the structure didn't.
Stay in the Mud
Morley Robertson closed his final column like this: We must doubt the answers AI offers, seek out context on our own, and keep getting our hands dirty.
Stoll said the same thing thirty years ago, in different words: Understanding your data and figuring out what to do with it—that is the one thing a computer will never do for you.
The internet was an "empty cave" from the very beginning. That hasn't changed.
And now, for the first time in a long while, we find ourselves swept up in the old excitement again. The internet became an advertising wasteland, but AI has brought the fervor back. We're about to buy the snake oil all over again.
Footnotes
[1] Morley Robertson, "AI's 'Invasion' Has Stripped the Internet of Context and Made It Boring," *Weekly Playboy*, issue released January 26, 2026 (column series *A Provocative Plan for Japan's Revolution*, No. 490). Morley Robertson died of esophageal cancer on January 29, 2026, at the age of 63. This column was effectively his final published work.
[2] Clifford Stoll, Silicon Snake Oil: Second Thoughts on the Information Highway, Doubleday, 1995. Japanese edition: translated by Akira Kurahone, published by Soshisha, 1997.
[3] Clifford Stoll, The Cuckoo's Egg: Tracking a Spy Through the Maze of Computer Espionage, Doubleday, 1989. Japanese edition: translated by Hiroki Ike, published by Soshisha, 1991.
[4] Ibid., p. 135. I haven't actually read the book, so this is cited secondhand from a review. Both the quip about it being "pretty funny" and this very footnote were written by Claude. You could fairly argue that if I'm going to write about a book I should read it first, but a new deluxe edition of Deleuze and Guattari just came out, so…
[5] Ibid., p. 52. An anecdote involving Dr. Li Fang. On p. 244 of the same book, Stoll writes: "The computer itself has no insight. What cultivates that kind of thinking is reading, writing, analysis, and teachers."
[6] Kobayashi Hideo, "Various Designs" (*Samазama naru Ishō*), *Kaizō*, September 1929. Kobayashi's experience with Rimbaud is also recounted repeatedly in works such as "Letter to X" (1932).