Glickstein, Adina, artist.

Operational.

[Place of publication not identified] : Adina Glickstein, 2024.

1 online resource

"Operational is a typeface inspired by the evolving history of character encoding, emphasizing the gap between how humans and computers comprehend written language. Rendering the alphabet unfamiliar to the human eye, it stages a distinction between the comprehensible (i.e., human-readable) and the operational (i.e., data usable by computers) to explore how meaning functions differently across human and nonhuman use-cases. The name "Operational" is an homage to Harun Farocki, a German artist and filmmaker who coined the term "operational images" to describe images produced by and for machines. Operational images enable machines to operate autonomously for purposes like computer-vision, autonomous or remote weapons systems, and computerized industrial production. Humans may sometimes be able to see operational images, but they aren't really meant for us; we are not the primary recipients of their meaning. They are the visual language of machines talking to machines; even when they are visible to humans, their meaning is encrypted. Likewise, computers do not understand language the same way that people do. Each period in the history of computing has had to reckon with the question of how letters and words can be broken down in such a way as to be "meaningful" or functional to a machine. In the earliest years of computation, electronic data processing was carried out using punched cards. Letters were broken down into character codes comprised of a string of binary, or true/false, values, which could be represented through the negative spaces of punched holes. Later, character encoding was carried out numerically: in the 1960s, the ASCII system was developed for standardizing character codes, assigning a numerical equivalent to each letter of the alphabet. The more inclusive Unicode system was adopted in the 1980s, and remains the default character encoding convention today. With the increasing elaboration of artificial intelligence (AI), including Large Language Models (LLMs), new ways for computers to comprehend words are coming to the fore. AI companies have proprietary models for training language models to "understand" human text. OpenAI is one leading player in the field; in a similar manner to earlier eras of encoding, they assign numerical "tokens" to strings of characters. Unlike ASCII and Unicode, OpenAI's tokenization model also accounts for the context and syntax of a sentence, representing an increasingly elaborate capacity for computers to make sense of human language. Each letter in the typeface Operational is made up of a set of numbers, which correspond to that letter's numerical codes in various moments from computer history. Capital letters are three tiers high. The bottom number is that letter's character code from the EBCDIC encoding convention used by IBM punch cards; the middle number is that letter's ASCII code; and the top number is that letter's OpenAI token ID. Because the IBM character chart only accounted for capital letters, the lower-case glyphs in the Operational alphabet only consist of two numbers: the ASCII code and OpenAI token ID. This also has the pleasant visual effect of affording a taller "cap height" to the capital letters, consistent with the usual practice in type design of making capital letters taller than lower-case ones. The design for the numbers 0-9 - the atomic units of the Operational alphabet - was inspired by the history of pixel type. These numbers reference to the way that text has been rendered on computer screens. Other formal elements, including a dingbat that appears as the "space" character and illuminates certain letter glyphs, are drawn from the markings that appeared on IBM punch cards. While Operational turns text into strings of numbers, the typeface itself does not convert your words into computer-readable text. Rather, drawing together these details of computing history, it playfully renders language illegible to both humans and machines. While each glyph references, the past, present, and future of text encoding, the end result of this synthesis is strictly decorative. In this way, it is a détournement of the labor of language-processing: a refusal of the "work" of comprehension, by computers and humans alike."-- provided by distributor.

Graphic design (Typography)
Natural language processing (Computer science)
Arts graphiques.
Traitement automatique des langues naturelles.

Software.

Farocki, Harun, other.
Library Stack, distributor.
Library Stack.