A security researcher named Carter Church has used GPT-6 Astra to break a Napoleonic-era cipher that had resisted decryption attempts. That’s yesterday’s news. Today’s news is a dodo.
Church posted the cipher work online, and the same researcher has now shared how they used Opus 5.5 to find a previously-unnoticed eyewitness account of dodo hunting from 1615. The account comes from a ship’s log in the Dutch East India Company archives, a vast collection known as the GLOBALISE corpus. The researcher describes the model’s search process as “epistemological weirdness.”
How Opus 5.5 Found the Log
The researcher started with a defined question: where can we find new historical evidence about dodos? They had already researched the exotic animal trade of the seventeenth century and were planning a book on animal extinctions in the early modern period, with a dodo chapter.
The next step was downloading the GLOBALISE archive and running it through an embedding model. That produced a searchable index, which Opus 5.5 then used to surface candidate passages. The model ranked those passages, and the researcher reviewed the top matches.
The process resembles how humans do research, but with a key difference. An AI agent like Opus 5.5 can spawn dozens of copies of itself to read through sources in multiple languages. Give it an API key, and it can run its own embedding searches on new sources that emerge during its work. That is exactly what Opus did here.
The 1615 Manuscript
Most of what Opus 5.5 found was already known to historians and scientists who have studied the dodo. But one manuscript stood out: a ship’s log from 1615, written by a man named Isbrant Cornelisz van Petten.
Van Petten captained a Dutch East India Company merchant vessel called Wapen van Amsterdam. The Wapen made landfall on the island of Mauritius in April 1615, where the crew collected water and food for their voyage to the East Indies. Among other things, they “caught many tortoises, dodos [dodeersen], and some geese and parrots.”
The log is available online, and the researcher transcribed and translated the relevant passages. They note that the transcription and translation are not perfect, and invite readers who work on early modern Dutch to offer corrections.
What the Passage Adds
The 1615 entry fills a gap in the dodo timeline. The dodo famously went extinct around 1681 due to overhunting, but the years between 1611 and 1616 had been relatively quiet.
The researcher read through the secondary literature, including Parrish’s 2013 book The Dodo and the Solitaire and other specialist articles. They conclude that this is a new addition to the record, even if it is not a dramatic one.
The Red Rail Find
The same search turned up another result. Opus 5.5 found a probable new reference to the red rail, another extinct bird from Mauritius.
An account from 1638, the year the Dutch first colonized the island, describes “field-hens” using the Dutch word (velthoenderen). Experts have previously identified this word as one the Dutch used for red rails. But this particular reference was apparently missed because a French scholar mistranslated it as perdrix, meaning partridges, in 1890.
Opus 5.5 was able to go back to the original manuscript source and correct the mistranslation.
The Jahangir Painting
The most intriguing possible finding remains unconfirmed. The researcher notes that Jahangir, the Mughal Emperor, apparently owned a living dodo. No one knows who gave it to him, when, or why, and Jahangir never writes about it in his memoirs.
The researcher wrote about this painting back in 2012. Two years earlier, Mansur had painted a Mauritian dodo that biologists cite as the most accurate surviving representation of the bird. In other words, Jahangir was exactly the kind of ruler a canny merchant or courtier would approach if they came across a rare specimen.
If the connection can be nailed down, it might explain one of the most mysterious paintings from the 17th century.
A Publishable Result
The researcher describes the dodo passage as a “publishable result,” especially when paired with the red rail finding. Neither discovery transforms our understanding of the dodo’s extinction, but both add detail to the historical record.
The red rail correction is particularly satisfying. It shows how an AI agent can catch errors that human translators have made over centuries.
The dodo log is less dramatic but still valuable. It is a new eyewitness account from the early 17th century, and it fits neatly into the existing timeline.
| Step | Action |
|---|---|
| 1 | Define a research question on dodo history |
| 2 | Download the GLOBALISE Dutch East India Company archive |
| 3 | Run the sources through an embedding model |
| 4 | Use semantic search to surface candidate passages |
| 5 | Ask Opus 5.5 to rank and review the matches |
| 6 | Review the top matches and confirm findings |
The process is straightforward, but the outcome is notable. A frontier model found a 1615 ship’s log that historians appear to have overlooked.
The red rail correction is a small victory. The Jahangir painting remains a mystery. But the dodo log is a genuine new source, and it is worth reading.
Source material: “Using Opus 5.5 to discover a new eyewitness record of the dodo,” resobscura.substack.com.
Get the Notebook.
The day's best stories and every fresh verdict, in plain English, in your inbox by seven. One email a day, no more.