by Joe Cerniglia
The Alan Turing Institute in the United Kingdom has published a reformulated version of the article I posted back in March of 2022. Turing Data Stories is a blog of the Institute that is dedicated to writing and publishing examples of open-sourced research, meaning that all the various threads of the author's work (programming code, methodology, and, most important, the original source data) are available to all to experiment with, critique, build upon, and understand.
The importance of having open-sourced research became obvious to me as I began to study academic articles on machine learning. Often, data from which many of these studies are drawn are not publicly available; thus, they offer no way to duplicate the work and prove that the methods elaborated in the articles are sound. The problem is common in a significant percentage of academic research that has been published. While such omissions may have been answerable if not justifiable in the pre-internet era, the age of the internet seems to have weakened if not demolished altogether any pretense of a rationale that once existed for them. Nevertheless, these omissions persist. This seemed to me a problem that someone should be working on. It was then that I discovered The Alan Turing Institute, and learned to my delight that, in the United Kingdom at least, this problem is taken very seriously at the highest levels.
I then proposed to the Turing Data Stories group that perhaps my article could be expanded as a data story to try to model some of these ideals. Working with them over the course of these past few months, and having them peer-review the work, I have been impressed by their dedication to building an A.I. infrastructure in the U.K. that will propel the country to enjoy the many benefits of a broad computer literacy. We need something similar here in the United States.
Here, then, is the latest Turing Data Story.
The connection between Amelia Earhart and the United Kingdom may seem tenuous, but in fact it is rather significant. Earhart was whisked by air to London in a driving rainstorm after her solo trans-Atlantic flight in 1932, the first made by a woman and the second made by a human being, after Lindbergh. That event so inscribed itself upon the memory of Londoners that Walter Sickert thought it worth commemorating, in a painting that now hangs in the permanent collection of London's Tate Gallery.
Additionally, the U.K. played an important part in the early research that has led some to believe Nikumaroro Island may have been the place where Earhart's world flight ended. In 1940, British subjects working as coconut planters on Nikumaroro, then Gardner Island, discovered a skull on the southeast corner of the island. Their subsequent searches in this area of the island led to the discovery of additional bones, along with a sextant box, a woman's shoe, a Benedictine bottle, and the remnants of a campsite, which to them seemed evidence of a castaway. They reported this discovery to British authorities in their chain of command, and sent the bones and artifacts to the Western Pacific High Commission in Suva, Fiji (then a British colonial possession) for further analysis. It was precisely in this area of the island where the bones were found that researchers from TIGHAR discovered the glass cosmetic jar, which is the subject of my data story, almost exactly 70 years later.
While the commanding officers in Fiji ultimately doubted they had received the remains of Amelia Earhart, their contribution to the investigation and to the various lines of evidence is noteworthy and relevant to the data story itself.
Thus, it is also noteworthy that the U.K.'s self-described national institute for data science and artificial intelligence has taken an interest in the story, both for its ability to illustrate good data science practices and simply because it is a great story.