opinion tech single source: New Scientist

The Echoes of the Void

Picture this: eight hundred years from now. A grad student sits in a lecture hall dedicated to what they call the Void Ages. Their reading list is painfully thin—a few novels, some blurry clips labeled "movies," and maybe a clumsy assembly of gear meant to run those archaic contraptions called "video games."

The professor, perhaps equipped with something unsettlingly metallic, tries to explain why those old controllers looked like molded fists. The core question hanging there is heavy: why is everything so broken? Why is our history reduced to fragments cobbled together by probability engines?

“No war or fire caused this,” the instructor replies gravely. “We fed our whole culture into these massive language models and threw away the originals. What remains is just algorithmic output. We are left trying to reverse-engineer human existence from code.”

That's the chill of it. We might end up studying a beautiful simulation of ourselves, mistaking statistical coherence for genuine feeling. It feels like burning libraries just to train a predictor function.

But sometimes, the true measure of humanity isn't in perfect records; it’s in specificity. When I was panicking during the 2020 uncertainty, turning to Samuel Pepys—a guy dealing with plague back in 1666—was grounding. He wasn't giving political treatises; he was noting how one hides a wheel of cheese while fleeing London.

That tiny detail anchors the vastness of fear in something tangible and real between two people separated by centuries. Even fiction achieves this weight; George Eliot and H.G. Wells stick around because they caught the nervous ticks of their age—the specific ways people grappled with industry changing everything around them.

Those quirks are seismic readings of a living mind, not polished summaries generated by chance patterns. My own attempt at keeping track—my note about what one particular person experienced in San Francisco—is an insistence on that singular record over the smoothed-out average provided by any machine learning process.