Bay Area techies fund $1M contest to solve ancient puzzle

0 31

One of the burnt papyrus scrolls that the Vesuvius Project’s leaders have scanned, hoping to use machine learning algorithms to find Roman, Greek or early Christian text.

One of the burnt papyrus scrolls that the Vesuvius Project’s leaders have scanned, hoping to use machine learning algorithms to find Roman, Greek or early Christian text.

Courtesy of Vesuvius Project

Silicon Valley’s wealthy investors have been busy with the side projects this year.

There’s the mysterious land grab in Solano County, apparently part of a plan to construct a Utopian city. Now, add in another potentially impossible pursuit: the restoration of an ancient library’s scrolls, burned to a crisp by the eruption of Mount Vesuvius in 79 A.D.

Bay Area entrepreneurs-turned-investors Nat Friedman and Daniel Gross, both of whom live in the Bay Area, launched the “Vesuvius Challenge” in March. It’s an open contest: Entrants have 100 more days to find passages of “continuous and plausible text” in scans of the ancient scrolls, called the Herculaneum Papyri, as well as turn in smaller discoveries like a trail of ink or a group of letters. A million dollars in prizes is up for grabs, with $700,000 for the continuous passages, according to the project’s website.

Advertisement

Article continues below this ad

Friedman, once GitHub’s chief executive, told SFGATE that when they launched the challenge, the announcement of OpenAI’s new GPT-4 product overshadowed the project. So he started drumming up more support, texting techie friends and turning to his 100,000-plus Twitter followers. The prize pool ballooned from $150,000 to about $1 million.

Friedman and Gross — whose firm Cue was bought by Apple — each chipped in $125,000, the project site says, as did Stripe co-founding duo John and Patrick Collison — who all four also bought into the Flannery Associates group behind the Solano County land grab. Bay Area investor Joseph Jacks is listed as offering $250,000 toward the scroll decoding, as is U.K.-based billionaire Alex Gerko. The site lists Shopify CEO Tobi Lütke and Notion CEO Ivan Zhao among the lower-dollar contributors.

Most of them contributed only to the grand prize, Friedman told SFGATE. “It’s an easy sell: You can be a part of solving a three-century-old archaeology puzzle and you don’t have to spend any money unless it actually happens,” he said.

The puzzle, which began with the scrolls’ discovery in 1750, is an incredibly difficult one. Hundreds of papyrus scrolls were burnt to blackened, charcoal-like logs as waves of lava washed over the Roman town of Herculaneum in 79 A.D. Their texts, which researchers think could include unknown Greek and Roman masterpieces and early Christian writings, were written with carbon-based ink, which doesn’t stand out on even the most modern X-ray scans.

Advertisement

Article continues below this ad

Because the rolls of papyrus are burnt to a crisp, they’ve survived the centuries since their writing. Researchers hope to use modern software to create text passages out of impressions in the layers of scroll. 

Because the rolls of papyrus are burnt to a crisp, they’ve survived the centuries since their writing. Researchers hope to use modern software to create text passages out of impressions in the layers of scroll. 

Courtesy of Vesuvius Project

That the pursuit feels remotely possible may be due to Kentucky scientist Brent Seales, who created the challenge with Friedman and Gross. For years, Seales has been trying to figure out what the scrolls from Herculaneum say — a project that gained more credibility in 2016, when he used X-rays and his own software to learn that an illegibly burnt scroll from 300 A.D. contained the third book of the Hebrew bible, The Guardian reported.

But that blackened log had metal-based ink, unlike the larger library lost in Vesuvius’ eruption. This time around, Seales used an X-ray machine powered by a particle accelerator to capture two of the full scrolls and several fragments, the project’s website says, with the hope that advanced software might be able to find tiny fluctuations in the interior of the logs and turn out legible script.

As part of the challenge, Seales, Friedman and Gross have made the scans public, with massive stores of data: The scans captured the scrolls’ interiors at a resolution of 4 to 8 microns, or about a twelfth of the diameter of a human hair. The project’s site offers a technical overview, links to discoveries made since the project launched (earning participants up to $10,000) and access to the large data server.

Advertisement

Article continues below this ad

The hope, Friedman said, is that by opening the project to the wider computing world, they can speed up Seales’ research. So far, efforts have centered around “segmentation,” the pulling apart of the scroll’s many layers into two-dimensional pieces. Ink won’t show up cleanly, even when and if this is done, but the project’s leaders hope that machine learning algorithms will be able to detect inked papyrus sections.

Friedman learned of Seales’ research in 2020 after he read Philip Matyszak’s “24 Hours in Ancient Rome,” according to The Information, which covered the project on Friday. He told SFGATE that he’s hoping the “Indiana Jones”-like “adventure” will unveil a great piece of literature or history, or, better yet, something that upends our understanding of the past. 

“If you could tell the ancient Romans, 2,000 years later, people would be using particle accelerators and coordinating in a global project to try to recover just a few of the words that you wrote, that would probably blow their minds,” he said, starting to laugh. “Right?”

Hear of anything happening at a Bay Area tech company? Contact tech reporter Stephen Council securely at [email protected] or on Signal at 628-204-5452.

Advertisement

Article continues below this ad

Source link

Denial of responsibility! YoursTelecast is an automatic aggregator of the all world’s media. In each content, the hyperlink to the primary source is specified. All trademarks belong to their rightful owners, all materials to their authors. If you are the owner of the content and do not want us to publish your materials, please contact us by email – [email protected]. The content will be deleted within 24 hours.
Leave A Reply

Your email address will not be published.