On Wednesday, the Vesuvius Challenge announced that researchers had virtually unwrapped an entire carbonized Herculaneum scroll for the first time. The scroll—designated PHerc.172, recovered from the Villa dei Papiri in the 1750s and held in Naples ever since—yielded a complete, end-to-end text through a pipeline of X-ray tomography, machine-learning segmentation, and ink-detection models trained on sub-millimeter variations in papyrus density. The winning team, a consortium led by the University of Kentucky’s Brent Seales, will split a prize purse that now exceeds $1.8 million. The papyrologists who reviewed the results have identified the author as Philodemus of Gadara, an Epicurean philosopher who taught in the Naples area in the first century BCE. The work’s title: On Vices.

This is, by any measure, a staggering technical achievement. Two thousand years of volcanic carbonization, and a text that no human eye has seen since the reign of Nero is now, in a very literal sense, in print. The AI community has been celebrating for 24 hours straight. The computer-vision papers will write themselves. The TED-talk invitations are probably already in inboxes.

Which is why someone should say the thing nobody wants to say: Philodemus was not Seneca. He was not Lucretius. He was not even a particularly interesting minor poet. What we just spent millions of dollars and tens of thousands of GPU-hours to recover is, very likely, a mediocre self-help manual by an also-ran.

The Author You’ve Never Heard Of (For Good Reason)

The identification of Philodemus as the author of the first fully-recovered scroll is not a surprise to classicists. About half of the known texts from the Villa dei Papiri are attributed to him, and the fragments that have been physically unrolled over the centuries—painstakingly, destructively, one flake at a time—have already given us a decent sense of his range. He wrote on ethics, on rhetoric, on music, on the management of anger. He was prolific. He was also, by the consensus of the scholars who have actually read him, a second-tier intellect in an era that produced first-tier ones.

“Philodemus is the guy you read when you’ve already read everything else,” one doctoral candidate in classical philology told me this week over Slack, after the announcement dropped. “He’s not Cicero. He’s not Vergil. He’s the person a minor Roman official would invite to a dinner party to sound cultured, and then regret it.”

The line between preserving cultural heritage and preserving the ancient equivalent of a midlist airport paperback is thinner than anyone in the digital-humanities funding pipeline wants to admit. The Vesuvius Challenge has raised its public profile on the promise of recovering lost masterpieces—Aristotle’s lost dialogues, unknown works of the tragedians, the missing books of Livy. The actual first complete scroll we recovered is a treatise on moral failings by an Epicurean who ran a modestly successful philosophy school in a resort town. It is the ancient Roman equivalent of finding a complete set of early Tony Robbins cassettes in a perfectly preserved time capsule and treating it like the Rosetta Stone.

The Real Prize Is Still Sitting in the Villa

The excitement around Wednesday’s announcement is, in one sense, entirely warranted. The technical pipeline works. The segmentation models generalize across scrolls. The ink-detection algorithms can distinguish carbonized writing from carbonized papyrus background with enough fidelity to produce continuous, readable columns of Greek prose. That is genuinely world-changing. The opportunity cost of not scaling this pipeline to the remaining 600-plus unopened scrolls in the Naples collection would be a dereliction.

But the public narrative has already slipped into a familiar groove—one in which the technology itself is the hero and the content is an afterthought. The New York Times will run a breathless feature on the AI that unlocked the scrolls. The venture capitalists who backed the prize will appear on podcasts talking about the inevitability of progress. And somewhere in the background, the actual question—what is in the rest of the library?—will get elbowed aside by the spectacle of the method.

That question matters. The Villa dei Papiri almost certainly belonged to Lucius Calpurnius Piso Caesoninus, the father-in-law of Julius Caesar and a patron of Philodemus himself. Piso was a major political figure of the late Republic, and his library was, by the standards of the time, a serious collection. The working hypothesis among papyrologists is that the villa’s shelves contained a mix: some Philodemus, some Epicurean core texts, and—crucially—some works that Piso actually valued. If the library was a vanity project for the resident philosopher, we will get a lot more Philodemus. If it was a working collection assembled by a powerful Roman senator, we may yet find the lost books that classicists have been dreaming about for two centuries.

The difference between those two scenarios is the difference between an intellectual revolution and an expensive footnote. And right now, the only data point we have—the first complete scroll—points toward the footnote.

Let the Machines Read the Boring Ones

This is not an argument against the Vesuvius Challenge. It is an argument for taking it seriously as a sorting problem, not just a reading problem. The technical pipeline that produced Wednesday’s result should now be turned loose on every scroll in the collection, as fast as the funding allows, with no expectation that every recovered text will be a lost masterpiece. Some of them will be Philodemus on ethics, part three. Some of them will be Philodemus on ethics, part four. Some of them may be a shopping list from the villa’s kitchen staff. The point is to find out.

What the AI community should resist is the temptation to treat every recovered scroll as a triumph of equal magnitude. The first complete text is a proof of concept, not a destination. The scholars who will actually spend years editing, translating, and interpreting these texts know this. “Wednesday was a dress rehearsal,” said one researcher involved in the papyrological review, speaking in a Naples hotel bar after the press conference. “We proved we can read a whole scroll. Now we need to prove we can find the ones worth reading.”

The scrolls have been silent for 1,948 years. They can wait another few months while we figure out which ones actually matter.

Sources