A 3,800-year-old puzzle may have just been solved by an amateur. Tom Di Mino, a self-taught AI engineer and hobbyist linguist from New York’s Hudson Valley, claims to have deciphered Linear A, the Bronze-Age Minoan writing system that has stumped experts for more than a century. According to Hacker News, his work is now under review by linguistics scholars at Rutgers and Cambridge, and if it holds up, it would rank among the biggest breakthroughs the field has seen in decades.
What stands out here is the tool behind the claim. Di Mino used Claude Code to build a suite of Python scripts that query and cross-reference the digitized Linear A corpus, letting him test hypotheses at a scale that would be brutal to do by hand.
What Linear A is, and why it matters
Linear A appeared around 1800 BC and was used until 1450 BC, when Mycenaean Greeks conquered Crete. The Mycenaeans borrowed the Minoan symbols and tweaked them, creating Linear B. That sister script was cracked in 1952 by Michael Ventris, another amateur, and it landed on the front page of the New York Times.
Linear A stayed locked. Here’s why:
- The two scripts share about 60 syllable signs, so experts could guess how many Linear A symbols sounded but had no idea what the words meant.
- Thirteen extra Linear A symbols had no accepted sound values at all.
- Most surviving Linear A tablets are trade inventories, which reveal almost nothing about the underlying language.
The key that turned
Hacker News reports that Di Mino’s breakthrough came on May 22, while he studied a set of Minoan prayer inscriptions that followed a fixed formula. Every word in the formula was known except the first, a verb that showed up in slightly different forms across five sanctuary sites.
Using a Linear A-only sign for “na,” he unlocked the root “nawaya,” meaning “to dwell.” That matched the three-consonant N-W-Y system found in Hebrew, Akkadian, and other Semitic languages, where the same root means “to dwell or inhabit.” His conclusion: Linear A maps to an extinct Semitic language that fed into biblical Hebrew, roughly the way Latin fed into Italian.
The Semitic idea isn’t new. Cyrus Gordon argued something similar back in 1957 and never won over the field. The difference, Di Mino says, is that his solution actually produces translations, including prayers addressed to a Goddess that echo later Hebrew prayers.
What he’s produced so far
Di Mino’s claimed results go beyond a single word:
- Proposed readings for 40 signs, including 13 whose phonetic values were previously unknown.
- Resolved sound values for 5 Linear B signs that had stayed unsolved until now.
- A lexicon of 408 Linear A terms translated into English.
- A 9-page draft manuscript, “Ya Diktu: Grammar of the Minoan Peak Sanctuary Libation Formula,” aimed at a peer-reviewed journal.
He also claims his logogram insights fix problems in some existing Linear B translations, which he treats as a check that his method is working.
Why this is worth watching
This is significant for two reasons, and only one of them is about ancient Crete. The decipherment claim still has to survive expert scrutiny, and extraordinary claims in linguistics have collapsed before. Caution is warranted until Rutgers and Cambridge weigh in.
The second reason is the method. An amateur paired domain obsession, seven years of reading, two trips to Crete, with AI-built tooling to brute-force pattern testing across a corpus. That’s a template more researchers will copy: the human supplies the hypothesis and the judgment, the AI handles the grinding scale. Ventris had Alice Kober’s statistical groundwork and index cards. Di Mino had Python scripts and an LLM.
Whether or not the translation stands, the workflow is the story for practitioners. Expect more solo researchers to attack long-cold problems this way. The full breakdown, including the inscription formula and figures, is available at the original source.