Somewhere in the second millennium before Christ, on the island of Crete, a civilisation wrote. Scribes pressed rows of small looping signs into wet clay tablets — lists of grain, counts of offerings, records of households whose names we will never know. For about four centuries they wrote this way, steadily, unremarkably. Then, around 1450 BC, their world ended in fire and ash, and the writing died with it.
Four thousand years later the tablets survive. The words on them do not.
This is the story of Linear A — the oldest script in Europe that nobody can read.
For seventy-five years, linguists and archaeologists have tried to break it. The approach is always the same: find the sound, find the language, find the key. The sibling script, Linear B, was cracked in 1952 by an architect named Michael Ventris, who proved it was an early form of Greek — the Greek of the Mycenaean palaces, written a few centuries after the Minoans. That breakthrough electrified the field. If Linear B could be read, surely Linear A, the older sister, would fall too.
It has not. Seventy years later, it still refuses.
The difference is brutal. Linear B turned out to be Greek, a language we already knew. Linear A, as far as anyone can tell, is not Greek and not any known ancestor of Greek. It may be an isolate — a language with no surviving relatives, spoken by a people who left us everything except their words. Scholars can recognise individual signs: numerals, an ideogram for wheat, perhaps a tally for wine. But the grammar, the syntax, the vocabulary — the actual voice — remains locked.
In the last few years, machine learning has entered the fight. Research teams have trained neural networks to find statistical patterns inside the Linear A corpus, hoping that structure will reveal sound. The results are intriguing: candidate readings, clusters that look like phonetic families, signs that appear to pattern like grammar. Yet no Rosetta Stone has surfaced, and every proposed decipherment has been met with suspicion. The honest position is that we have hints — and no voice.
What does a four-thousand-year-old silence have to do with a project that stores your photo and your words until 3026?
Everything.
When we seal a message for the year 3026, we assume someone will read it. That is the entire wager of the project. But Linear A is the cautionary tale told from the other side of time: its writers were understood perfectly well by everyone around them. The script did not fail. The world did. When the civilisation collapsed, the spoken language dissolved with it, the institutions that taught the script vanished, and a perfectly legible record became a corpse — readable as marks, unreadable as meaning.
Think about what that means for us. We are not preserving stone tablets; we are preserving data, which is even more fragile, because data depends on context: on file formats, on operating systems, on someone who knows how to open a folder from the 2020s. A thousand years is a very long time for a civilisation to keep its manners. The Minoans could not leave a key to their language, because they could not imagine losing it. We can, and maybe we should.
We have advantages the Minoans never did: redundancy, the ability to store the same message in a hundred formats and a hundred places. We can write self-describing records, keyed in multiple languages, with instructions — like the engraving on a phonograph cover. And we have something even more useful: the humility to admit that the reader of the far future may not know us, may not speak our languages, and may find us as foreign as we find the Minoans.
The best guarantee we can offer the year 3026 is not cleverer encryption. It is the plainest, most human gesture available: write the message down, in words that any literate stranger could follow, and place it where a civilisation — any civilisation — would think to look.
The Minoans wrote for their own generation and their words outlived them, sealed and unread. We are trying something braver: to write for a generation we will never meet. There is no guarantee the experiment succeeds. But when the future finally opens the vault and finds our clay tablets, let them not be silent.
Someone in 3026 will read what we left. We owe it to them — and to the unread voices of Crete — to make sure it can be read at all.