Poe in Cyberspace: Balloons! Drones!! The Global Internet!!!

Early in the morning of April 13, 1844, a crowd formed outside the offices of the New York Sun to await the appearance of an Extra that promised details of the first Atlantic crossing by balloon. After several reports of actual shorter balloon flights, a grand transoceanic flight seemed possible, even though, as some readers still remembered, it was the Sun that had published Richard Adams Locke’s “Moon Hoax” a decade before. Edgar Allan Poe’s “Balloon Hoax” was his dramatic way of announcing his arrival in New York just seven days before. Even today, dramatic news reports of prodigious balloon accomplishments may still leave us not knowing whether to believe or to be skeptical. Thus when Google announced its scheme of a huge balloon network for global Internet coverage, some critics dubiously labeled it “crazy,” whereupon Google, flipping the hint, named the venture Project Loon.

First launched experimentally in New Zealand in 2013, Project Loon became a nationwide project in Sri Lanka in 2015. Entailing hundreds of networked high-altitude balloons, Project Loon, focusing initially on the Southern Hemisphere, will attempt to deliver 3G or LTE Internet support. If this was madness, it had a method to it, and Google did not long remain the only major entrepreneur willing to apply unconventional means to address the problem of inadequate global Internet coverage. Thus in 2015 Facebook joined the field, announcing its own project to launch a gigantic solar-powered drone named Aquila (eagle), to be supported by satellites and lasers, intended to deliver wide Internet access to poorly served populations, especially in the Southern Hemisphere.

English, the birth language of the Internet, will serve users in Australia, New Zealand, and India, but not in South America, where Spanish and Portuguese are the main languages, nor in Africa, where the three formerly colonial languages are joined by the official languages Afrikaans, Arabic, French, and Swahili, as well as a large number of unofficial languages. Incidentally, in an informal survey of Edgar Allan Poe postings on Twitter, a good many of these languages turn up, namely, Arabic, English, French, Italian, Spanish, Portuguese, and Turkish.

On the Internet, needless to say, computer languages may be as important as human languages. The battle for control of the cloud is also a battle between the languages of two popular operating systems, Google Android, found mainly on phones, and Microsoft Windows, traditionally found on many personal computers. In academic circles, however, the growing discipline of the digital humanities (formerly “humanities computing”) now requires the invention of new research tools in Unix, often running on institutional supercomputers needed to handle big data.

In this rapidly expanding field of text datasets, a leading role is being played by the Hathi Trust, a collaborative effort led by Indiana University and the University of Illinois at Urbana–Champaign, working through two of its divisions, the Hathi Trust Research Center (HTRC) and the Hathi Trust Digital Library (HTDL). At the HTRC the data worksets and algorithmic tools designed for historical research in literary texts may produce projects that are either public or private in nature; to join, scholars must be affiliated with a participating institution. Simple HTRC projects may explore tasks such as word frequency and word clusters, while advanced ventures may analyze entities such as locations and persons.

The HTRC endeavor called the Extracted Features Datasets draws data from the collection of the Hathi Trust Digital Library, described in an introductory video at https://www.hathitrust.org/htrc. Overviews of the project from varying points of view can be found at https://www.hathitrust.org/statistics_visualizations. Adepts with rsync on Unix can create their own synchronized datasets for study. Working with more than a hundred participating institutions (listed at https://www.hathitrust.org/community), the Hathi Trust Digital Library has made available for study 4.8 million volumes (equal to 1.8 billion pages or 734 billion words), derived from public domain material surveyed by Google Books, supplemented by an additional 550,000 non-Google volumes, primarily in pre-1923 English-language publications in the public domain. One unusual feature of the Hathi Trust...