Defining Digitalities II:
What’s Digital About Digital Communication?
Thomas Haigh*+, Sebastian Gießmann* 
*Siegen University, +University of Wisconsin, Milwaukee 
INFORMATION
SOURCE TRANSMITTER RECEIVER DESTINATION
SIGNAL RECEIVED
SIGNAL
MESSAGE MESSAGE
Digital Writing Digital Reading
Process Process
NOISE
SOURCE
WORKING PAPER SERIES | NO. 31 | JULY 2023
Collaborative Research Center 1187 Media of Cooperation
}
}
Working Paper Series
Collaborative Research Center 1187 Media of Cooperation
Print-ISSN  2567–2509
Online-ISSN  2567–2517
DOI  doi.org/10.25819/ubsi/10260
dspace.ub.uni-siegen.de/handle/ubsi/2453
URN  urn:nbn:de:hbz:467-24532
This  work  is  licensed  under  the  Creative 
Commons  Attribution-NonCommercial-No-
Derivatives 4.0 International License.
This Working Paper Series is edited by the Collaborative Re-
search Center Media of Cooperation and serves as a platform 
to circulate work in progress or preprints in order to encour-
age the exchange of ideas. Please contact the authors if you 
have any questions or comments. Copyright remains with the 
authors.
The Working Papers are accessible online at:  
https://www.mediacoop.uni-siegen.de/de/publikationen/
working-papers-media-of-cooperation/
Print copies can be ordered by sending an email to: 
workingpaperseries@sfb1187.uni-siegen.de
Publication is funded by the Deutsche Forschungsgemein-
schaft  (DFG,  German  Research  Foundation)  –  Project-ID 
262513311 – SFB 1187.
Cover image: Shanon 1948 diagram with reading and writing 
marked
Layout: Mattis Hunting
Universität Siegen
SFB 1187 Medien der Kooperation
Herrengarten 3
57072 Siegen, Germany
https://www.mediacoop.uni-siegen.de.sfb1187.uni-siegen.de
workingpaperseries@sfb1187.uni-siegen.de
Defining Digitalities II:
What’s Digital About Digital Communication?
Thomas Haigh*+, Sebastian Gießmann*  
*Siegen University, +University of Wisconsin, Milwaukee
Abstract: Although the distinction between digital and analog was first 
made in the context of automatic computers, the concepts were quickly 
broadened to apply to media and communication systems of all kinds. 
This working paper continues work on both fronts by looking at the his-
torical broadening of the concept of digitality to include non-numerical 
systems of representation such as those used to encode text and pic-
tures. This conception underlies the ability of computers to deal with 
things other than numbers, but it has its roots in communications the-
ory, most famously in the work of Claude Shannon. In parallel with our 
historical description of the emergence of non-numerical conceptions 
of digitality we broaden our analytical treatment of digitality to en-
compass more historical technologies and reading practices: not only 
adding machines and punched cards, but also musical boxes, weaving 
systems, movable type, and even alphabets and hand gestures.
Keywords: Claude Shannon; digital; information theory; 
Colossus;
In an earlier working paper, Haigh explored the ori- ers to deal with things other than numbers, but it has 
gins of the digital/analog divide in the discourse around its roots in communications theory, most famously in 
early automatic computers in the 1940s. Yet once the the work of Claude Shannon. In parallel with our his-
categories of digital and analog were invented, con- torical description of the emergence of non-numerical 
temporaries immediately used them to categorize ear- conceptions of digitality we broaden our analytical 
lier technologies for the representation of numbers and treatment of digitality to encompass more historical 
quantities such as slide rules and adding machines. Dig- technologies and reading practices: not only adding 
ital computers were digital in a direct, non-metaphor- machines and punched cards, but also musical boxes, 
ical sense: they mechanically or electronically encoded weaving systems, movable type, and even alphabets 
the values of digits and manipulated these encodings to and hand gestures. 
perform calculations. Building on this history, Haigh The affordances of text, of punched cards, and of 
argued for an analytical conception of digitality cen- paper tape are not identical but they all encode se-
tered on processes of reading, by which the continuous quences of symbols. This perspective demystifies the 
variation of the natural world is mapped onto one of a arrival, in the mid-1940s, of programmable computers. 
finite, and usually small, set of valid states. They embodied practices of digital reading comparable 
This working paper continues work on both fronts to those carried out by earlier machines and by humans. 
by looking at the historical broadening of the concept The addition of branching and looping capabilities, 
of digitality to include non-numerical systems of rep- while highly consequential, was a refinement of digital 
resentation such as those used to encode text and pic- control.
tures. This conception underlies the ability of comput-
4    CRC Media of Cooperation Working Paper Series No. 31 JULY 2023
INFORMATION
SOURCE TRANSMITTER RECEIVER DESTINATION
SIGNAL RECEIVED
SIGNAL
MESSAGE MESSAGE
NOISE
SOURCE
Figure 1: Shannon’s schematic representation of a “general communication system” distinguished between the message itself and the 
signal transmitted after the message had been encoded. Shannon conceptualized the message as a sequence of symbols. The task of 
the receiver was to interpret the received signal to identify each coded symbol transmitted, thus reconstructing the original message.
Digital Information Theory As with Shannon’s earlier work on switching circuits, 
his theory of information was not the transformative 
In many digital systems the signals do not represent work of a lone genius. Historians stress the extent to 
numbers at all. That marks a conceptual shift from which his work drew on both his collaborative wartime 
the origins of the digital/analog divide in discussion experience and earlier efforts to describe information 
of computers, since these machines were called digital transmission mathematically undertaken by his Bell 
precisely because they represented numbers as digits. Labs colleagues such as Ralph Hartley and Harry Ny-
Over time, however, the association of digitality quist from the 1920s onward.4 For our purposes we need 
with digits has weakened in favor of another sense of not disentangle the personal contributions of Shannon 
digitality: the engineering sense of digitality as de- to his new synthesis, merely assert that his work was 
fined by the analysis and design of digital signals that the path by which the new ideas made their way into the 
carry encoded information. This owes much to Claude broader world.
Shannon’s mathematical theory of communication, Our claim that Shannon’s landmark paper played 
or as it was more often called, information theory.1 an important role in defining a new, and much broader, 
Shannon’s work on the topic was initially published sense of non-numerical digitality may seem startling 
in his classic 1948 paper, “A Mathematical Theory of because it does not contain the words digital or analog. 
Communication.”2 In the decade since the completion Instead Shannon preferred the established mathemati-
of his master’s thesis (discussed in Haigh's previous cal terminology of discrete versus continuous functions 
working paper) Shannon had earned a Ph.D. in math- – an echo of the choice made by Stibitz two years ear-
ematics, spent a year as a visiting fellow of the Institute lier in his Moore School lecture. We believe that the 
for Advanced Studies in Princeton, and then returned to choice reflects Shannon’s knowledge that the process 
Bell Labs full time to work during the war with Stibitz he described did not necessarily involve converting the 
and Bell Labs on the NRDC’s gun direction contract and information being communicated into digits, mak-
on cryptography. His performance quickly earned him a ing symbol a more natural choice than digit and discrete 
permanent job in its mathematics research group.3 more meaningful than digital.5
1  Ronald Kline, The Cybernetics Moment, Or Why We Call 
Our Age the Information Age (Johns Hopkins University Press, 4  Statisticians also began to conceptualize information as 
2015). something quantifiable during the same period. Kline, The Cy-
2  Claude E Shannon, “A Mathematical Theory of Communi­ bernetics Moment, 22.
cation,” The Bell System Technical Journal 27 (July & October 5  As Ron Kline has pointed out to us, the term symbol was 
1948):379-423, 623-656. already been established in the mathematical theory of teleg-
3  Shannon’s career is described in Jimmy Soni and Rob Good­ raphy, notably in Harry Nyquist, “Certain Factors Affecting 
man, A Mind at Play: How Claude Shannon Invented the Infor- Telegraph Speed,” Bell System Technical Journal 3, no. 2 (April 
mation Age (New York, NY: Simon & Schuster, 2017). 1924):324-346.
Thomas Haigh, Sebastian Gießmann  5
INFORMATION
SOURCE TRANSMITTER RECEIVER DESTINATION
SIGNAL RECEIVED
SIGNAL
MESSAGE MESSAGE
Digital Writing Digital Reading
Process Process
NOISE
SOURCE
Figure 2: We conceptualize the processes on the left side of Shannon’s diagram as the act of writing digitally into a communication 
channel; we conceptualize the processes on the right side as reading digitally from the same channel.
In modern terms, or even according to definitions that tion content of an analog information source can be 
would be accepted just a few years later, Shannon’s measured only by digitizing it, or as Shannon put it, by 
paper is unmistakably a treatment of digital commu- defining the required “fidelity of recovery” and using 
nications and digitization. Shannon acknowledges in this to define “a rate, having the property that is it pos-
its first sentence that interest in a “general theory of sible, by properly encoding the information, to transmit 
communication” had been motivated by the devel- it over a channel whose capacity is equal to the rate in 
opment of pulse code modulation, the basic method question, and satisfy the fidelity requirements.”6 This 
for the digitization of audio. This technique uses high is why we call Shannon’s model of communication fun-
speed sampling to turn an audio stream into a sequence damentally digital: it can deal with an analog informa-
of numbers. Shannon’s self-proclaimed general theory tion source only by coding it into a digital signal.
of communication is often depicted as a model that en- In our conception, which we believe aligns with 
compasses techniques such as voice communication historical usage, digitality describes a class of reading 
over analog telephone lines. But the paper is unambigu- practices. From this viewpoint the act of digital read-
ously a mathematical theory of digital communication. ing, i.e. sensing something in the world and mapping 
Shannon begins by proposing the bit, a termed coined it to one of a finite number of valid states, is equivalent 
for the occasion by John Tukey, a Princeton mathemati- to the right hand side of Shannon’s widely reproduced 
cian working at Bell, as the basic unit of information. On diagram. 
the second page Shannon defines the thing being trans- The system described by Shannon, in which the 
mitted as a message consisting either of a sequence of signals read digitally by the receiver were deliberately 
letters or, essentially, as one or more functions giving encoded and placed into a channel with the intention 
numbers that change over time (in one dimension for that they be received and recoded, describes the combi-
audio, in multiple dimensions of time and space for nation of digital reading with digital writing. As Haigh 
video). After that nod to the possibility of non-textual mentioned in his previous working paper, some digital 
encoding the rest of the paper focuses squarely on text. reading practices, such as the action of a digital ther-
Shannon called the bit rate available on a given com- mometer or a digital audio recorder, apply digital read-
munication channel its bandwidth. He used the word ing to inputs that were not deliberately encoded by an 
digit thirty times in the paper when discussing methods identifiable sender.
to quantify the information content of these messages. 
The bit, which is after all a contraction of “binary digit” 
is a fundamentally digital concept. As Shannon pointed 
out, to transmit a “continuous” (analog) signal exactly 
would require infinite bandwidth. Thus the informa- 6  Claude E Shannon, “A Mathematical Theory of Communica-
tion,” 47.
}
}
6    CRC Media of Cooperation Working Paper Series No. 31 JULY 2023
Symbols Versus Numbers Indeed, the first example Shannon presented to in-
troduce his concepts was one of encoding the 32 sym-
Shannon conceptualized the message being transmit- bols used in the standard teletype alphabet. His next 
ted as a sequence of symbols, chosen from a finite set. example involved an alphabet containing only the 
While he measured the information content of this se- letters A, B, C, D and E. From this viewpoint, the chal-
quence in bits, he did not require the symbols them- lenges involved in transmitting a sequence of numbers 
selves to be numbers. This may be why Shannon, and or a sequence of letters are identical. Letters are not 
even his Bell Labs colleague Stibitz who had introduced digits, even if they the two can be interchanged with a 
the digital/analog distinction in the first place, had trivial effort, but both are symbols drawn from a finite 
come to prefer continuous/discrete instead of analog/ set. The process of measuring the information content 
digital as a description for the two approaches. Digital of a message, by converting from the appropriate base 
made sense as a description for a computer project be- for the number of symbols to base 2 (binary), hinged on 
cause the symbols being manipulated by the computers the equivalence of symbols and digits. To Shannon the 
of the period were digits. The computers were fed input interchangeability of digits and numbers was already 
digits, carried out mathematical operations on them, too obvious to explain or justify. Ciphers involving the 
and output digits. ENIAC’s card punch interface, for conversion of letters to numbers had been around for 
example, was physically incapable of punching more centuries, and the bit patterns punched in paper tape 
than one hole in each column of the card and so could for teletype transmission could be read just as easily as 
not output anything other than a single decimal digit representing the numbers 0-31 or the teletype alphabet. 
in each column. As a general term for communication, Hence symbol and message were better terms to de-
in contrast, digital left a lot to be desired because most scribe the information being transmitted than digit and 
messages did not consist entirely of digits. number. It followed that discrete was a better word than 
The ideas in Shannon’s paper were shaped by his digital to describe the encoding used to transmit the 
wartime experience in encrypted communication proj- message. Likewise, analog made sense for computers 
ects.7 These wartime projects were also important for in which specific components played roles analogous to 
the emergence of what would soon be called cybernetic quantities in the system being modelled but made less 
thinking and for the concept of communication as an area sense for describing the transmission of a regular tele-
of study. Peter Galison famously located the origins of phone call. Continuous, on the other hand, is a precise 
Nobert Wiener’s “cybernetic vision” in his experiences description of variance in the current transmitted down 
on the same gun director project that Shannon worked the wire. Shannon’s paper was hugely influential, but 
on.8 Indeed, Wiener’s famous book Cybernetics carried he did not get his way with respect to vocabulary. Com-
the alternate title “Control and Communication in the munications engineers finished up adopting the termi-
Animal and the Machine.”9 In cybernetics, “communi- nology of analog versus digital that had been introduced 
cation” turned into an operative concept that combined to distinguish between kinds of computers, while Shan-
communication and programmed control in circular non became famous as the creator of information theory 
feedback loops. despite publishing his paper as a mathematical descrip-
According to Erhard Schüttpelz the notion of com- tion of communication. The result was a redefinition of 
munication “became visible in the change from the all three terms: digital now applied to all symbols rather 
theory and practice of secret communication, from the than just digits, analog to any system of continuous 
command basis of military communication to the com- variation, and information to anything coded digitally.11
mon user basis of mass communication––between Shannon made two crucial points about the encod-
the manipulation of mass communication and its civil ing of symbols. First, their appearance in messages is 
population, and old and new promises of autonomy and not random. In English, for example, certain letters are 
democracy.” More specifically, asserted Schüttpelz, much more common than others. Beyond that, though, 
“Shannon’s famous communication diagram is both a characters tend to cluster together in fixed patterns as 
telegraphic and a one-way model––a telegram.”10 words, and even words tend to follow each other in pre-
dictable patterns. (Such insights were vital to wartime 
7  Kline, The Cybernetics Moment, 26­35 discusses the back- and Irmela Schneider (Bielefeld: transcript, 2010):109­138. We 
ground to Shannon’s theory with a particular focus on wartime should, however, acknowledge that much work being done at 
cryprography. Bell Labs during the same era focused on voice transmission 
8  Peter Galison, “The Ontology of the Enemy: Norbert Wiener including efforts that later proved foundational to audio com-
and the Cybernetic Vision,” Critical Inquiry 21, no. 1 (Autumn pression and digital voice transmission. For this reason, Mara 
1994):228-266.. Mills has argued for the central place of telephony, rather than 
9  Norbert Wiener, Cybernetics, or Control and Communica­ telegraphy, in new media history. Mara Mills, “Media and Pros-
tion in the Animal and the Machine (Cambridge, MA: Technol- thesis: The Vocoder, the Artificial Larynx, and the History of 
ogy Press, 1948). Signal Processing,” Qui Parle 21, no. 1 (Fall/Winter 2012):107­
10  Erhard Schüttpelz, “’Get the message through’: From the 149.
Channel of Communication to the Message of the Medium, 11  For a close examination of the Postwar development of 
1945–1960”, in Media, Culture, and Mediality: New Insights cybernetics its relationship to the new sense of information see 
into the Current State of Research, ed. Ludwig Jäger, Erika Linz, Kline, The Cybernetics Moment.
Thomas Haigh, Sebastian Gießmann  7
codebreaking efforts, something nodded to by Shannon and receivers thus introduced a considerable amount of 
in a reference to “certain known results in cryptogra- inefficiency into the transmission of Morse code.
phy”). In mathematical terms, the transitions from one Shannon’s other crucial point was that no commu-
symbol to the next are not random. Shannon claimed nication channel is entirely error free. A certain propor-
the redundancy of English text was about 50%, mean- tion of the symbols dispatched will be garbled in transit, 
ing that a message could usually be reconstructed ac- represented in Shannon’s diagram by the box injecting 
curately if fifty percent of its characters were deleted. noise into the channel.13 Shannon discussed ways to 
This is the animating concept behind the classic game select coding schemes to minimize this. He finished 
hangman and the long running TV gameshow Wheel of by summarizing the work of Richard Hamming, one of 
Fortune – contestants request the most common letters his colleagues at Bell Labs, who had shown that by in-
first and may attempt to guess a phrase when most of troducing redundancy into the coding of the message 
its letters remain obscured. it was possible to detect (and hence correct) these er-
Because of this redundancy if English text was en- rors. Hamming subsequently developed a comprehen-
coded using a simple method, with five bits per letter, sive treatment of error correction and detection which 
the information content would be only about half that was widely applied in digital computing.14 Messages are 
of an optimal coding mechanism. Efficiency could be split into blocks, packed with redundant information in 
improved by using shorter codes for the more com- the form of “parity bits.” Adding more information al-
mon characters or character sequences, which Shan- lows for the detection of more errors, but at the price of 
non termed compression of the message. A few years a longer sequence to be transmitted and hence a lower 
later an MIT student, David A. Huffman, came up with effective bandwidth. By making the signal sequence 
a method that he proved was optimum for coding mes- even longer, enough redundancy can be included to al-
sages where characters occur with different frequencies low the correction of errors as well as their detection. 
(assuming, unlike English, messages had no dependen- For example, one popular coding method based on 
cies from one character to the next).12 Hamming’s work, SECDED, allows correction of a single 
The idea of encoding different symbols using codes error in each block and detection of two errors. It was 
of different lengths had a long heritage in communica- deployed by IBM in 1961 to improve the reliability of the 
tions. While teleprinter codes, Shannon’s explicit ex- memory of its STRETCH supercomputer. Each 72­bit 
ample, used five bits for each character (something in- word of memory included 64-bits of data and 8-bits of 
herent to the five channel tapes used to hold messages) redundant information.15 There is always a tradeoff in 
the Morse code used in conventional telegraphy and ra- the choice of block size and the amount of redundancy: 
dio communication used shorter codes for more com- accuracy and reliability of transmission versus speed of 
mon symbols. It translated the message one character transmission. The optimal choice on the expected rate 
at a time into combinations of three symbols: dot, dash, of errors, the severity of allowing the occasional unde-
and space (used only to mark the end of each character). tected error (far more serious in a code download than 
The most common letter, E, was coded with a dot and a an audio stream, for example), and the importance of 
space. In contrast, Y, a less frequently used letter, was error correction (in many applications reliable detec-
coded as dash, dot, dash, dash, space. The codes for dig- tion of errors is enough, since the receiver can request 
its all consisted of six symbols, again ending in a space. retransmission).
In a sense two translation processes took place each 
time a message was sent in Morse: first from English 
characters into dots, dashes, and spaces, and then from 
dots, dashes, and spaces into the on/off code sent by the 13  The origin of the term noise in this context is explored in Mara Mills, “Deafening: Noise and the Engineering of Commu-
operator using a spring-loaded Morse key. If we equate nication in the Telephone System,” Grey Room, no. 43 (Spring 
1 with the depression of the key for a time interval and 2011):118-143.
0 with the key not being depressed, a dot was coded as 14  Richard W Hamming, “Error Detecting and Error Correct-
ing Codes,” Bell System Technical Journal 29, no. 2 (1950):147-
1000, a dash as 111000, and a space as 0000000. Time 160. In these working papers we are adopting a broad definition 
intervals were of course approximate, but the process of of information theory, to describe a cluster of approaches initi-
digital reading by the recipient listening to beeps on the ated at Bell Labs include Hamming’s work on error correcting and detecting codes as well as Shannon’s personal contribu-
other end of the wire could nevertheless be highly reli- tions. As Ron Kline has pointed out to us, some participants 
able because of the degree of redundancy. The gap be- argued for a narrower definition of information theory. Shan-
non’s colleague, John Pierce, later bemoaned the fact that “er-
tween characters was more than twice as long as the gap ror correction in binary signals has become strongly associated 
between symbols and a dash was supposed to be three with information theory” because Hamming’s work was tied 
times as long as a dot. The need to ensure that the three to the practicalities of “computing and switching machines.” From this viewpoint, Hamming’s work on coding is paral-
symbols were reliably differentiated by human senders lel to information theory but, despite being incorporated into 
in Shannon’s paper, not part of information theory. J R Pierce, 
“The Early Days of Information Theory,” IEEE Transactions on 
12  David Huffman, “A Method for the Construction of Information Theory 19, no. 1 (January 1973):3-8.
Minimum­Redundancy Codes,” Proceedings of the IRE 40, no. 15  Charles J Bashe et al., IBM’s Early Computers (Cambridge, 
9 (1952):1098-1101. MA: MIT Press, 1986), p. 452-3.
8    CRC Media of Cooperation Working Paper Series No. 31 JULY 2023
Colossus was not digital in the sense of a digital com-
puter, because the bits did not represent numbers. But 
it was certainly digital in the broader, Shannanonesque 
sense of a machine that read information coded dis-
cretely as a sequence of symbols. Even this language of 
bits and bitstreams is problematic with respect to Colos-
sus. The term bit was not used at Bletchley Park (and was 
still to be coined). Bletchley Park cryptographers talked 
not of 1s and 0s, or even of true and false, but of a “tele-
printer alphabet” containing just two characters: dot and 
cross. Can one responsibly speak of bits in this context 
given that the word is, as Shannon frequently reminds 
us, a contraction of binary digit? Perhaps not. The “bit-
streams” processed by Colossus contained impulses that 
were neither digits nor binary (at least in the sense of the 
binary number system). One might wish that Shannon 
had been more consistent in his efforts to avoid talk-
ing about the transmission of digits. If he had followed 
through by talking about bandwidth in terms of binary 
symbols or binary characters rather than binary digits we 
might now with a clear conscience write about bicstreams 
or measure transmission rates in bis per second.
Other scholars applied similar ideas of digital and 
Figure 3: In Nonverbal Communication (1956), Ruesch and Kees analog representations to other forms of communi-
suggested that some gestures stood in place of words, and thus cation, such as human gestures. For example, in the 
were recognized as distinct symbols. 1950s the psychiatrists Jurgen Ruesch and Weldon Kees 
drew on the concepts of analogic codification and digital 
codification to categorize different forms of nonverbal 
Non-Numerical Digitality communication. They argued that “the use of words, 
whether in speech or writing, has certain limitations 
Not all the complex electronic machines of the 1940s akin to those of digital computers: words remain iden-
dealt with encoded digits. Perhaps the most interesting tifying or typifying symbols.”17 Later in the same book 
example of a machine that is digital in terms of sym- (Figure 3) they suggested that some gestures and facial 
bol processing but not in terms of processing numbers expressions stood in place of words and hence were rec-
is the Colossus codebreaking machine (in fact a fam- ognized as coding distinct meanings.
ily of machines) employed at Bletchley Park during the Decades earlier, as Mara Mills has shown, systems 
Second World War. One of us has argued elsewhere that of lip reading had been developed around the use of 
Colossus was not, despite frequent claims to the con- photographs and drawings to illustrate discrete facial 
trary, a computer and that it could not be programmed, expressions. One system literally digitized sixteen facial 
though it could be extensively configured. 16 Instead, expressions associated with speech by assigning a nu-
Colossus could perform logical comparisons between merical code to each via a “numerical cipher method.”18
bits taken from ten bitstreams: five of them read from In this sense, giving a thumbs-up gesture in re-
paper tape, and five generated by electronic circuits sponse to the question “How are you holding up?” 
designed to mimic the encoding wheels of specialized is a digital response in that the gesture is intended to 
Lorenz teleprinter encrypting attachments. The bits be recognized as a discrete symbol (akin to a modern 
had no numerical significance: Colossus had no hard- emoji). On the other hand, if someone is asked “How 
ware capable of interpreting successive bits, or bits read big was the fish?” and responds by holding both palms 
simultaneously from multiple channels, as encoding a vertical this is an analog communication: the distance 
binary number. All it could do was to compare bits ac- between the palms represents this size of the catch. 
cording to logical functions coded on switches and a 
plugboard and tally the number of times that the condi- 17  Jurgen Ruesch  and Wheldon Kees, Nonverbal Communi­ 
tions in question were met during the reading of a tape cation: Notes on the Visual Perception of Human Relations 
(usually a tape holding an intercepted and encrypted (Berkeley, CA: University of California Press, 1956), 8. Haigh 
learned of this work though a presentation by Luke Stark at 
message). the Society for the History of Technology (SHOT) 2018 Annual 
Meeting, “After the Clinic: Jurgen Ruesch, Weldon Kees, and 
Cybernetic Non-Verbal Communication, 1950-1960,” St. Louis, 
16  Thomas Haigh and Mark Priestley, “Colossus and MO, 12 October 2018.
Programmability,” IEEE Annals of the History of Computing 18  Mills, “Media and Prosthesis: The Vocoder, the Artificial 
40, no. 4 (Oct-Dec 2018):5-17. Larynx, and the History of Signal Processing”.
Thomas Haigh, Sebastian Gießmann  9
Digitally Controlled Machines
The new non-numerical senses of digitality meant that 
new classes of machine had now become retroactively 
digital because they were controlled by media that were 
now recognized as digital even though there were no 
actual digits involved. Mechanical adding machines and 
tape-controlled relay computers had been recognized 
as digital once the concept of the digital computer was 
created. Once the Shannonesque senses of digital and 
analog communication were established, entire fami-
lies of devices that were not numerical but were con-
trolled by information encoded in discrete forms like-
wise became retroactively digital. These include player 
pianos, Jacquard looms, and musical boxes.
The music box is a hybrid of digital and analog. It can 
play a fix repertoire of notes, corresponding to the fixed 
symbols encoded on a digital channel. In this case, they 
are encoded with pins on the surface of a rotating cylin-
der or disk. Each pin is placed to strike a particular cam, 
which in turn rings a bell or vibrates a prong to produce 
a fixed note. This is discrete because the position of the 
pin to strike one or another of the reading mechanisms 
codes a discrete note. If it was analog then new notes at 
intermediate frequencies could be produced by moving 
the pins up or down a little. The theremin, for example, 
is an analog instrument because the tones it produces 
vary continuously with movement of the operator’s 
hand. On the other hand, the timing of notes is analog.  
Moving the pin forward or backwards a little will alter Figure 4: A finely detailed portrait of Jacquard woven in silk on 
the time at which the note is produced by a correspond- an automatic loom. The cards that controlled the loom con-
ing amount. Following Shannon, we might call the en- tained a digital version of the portrait in the Shannon sense of 
coding used in the music box discrete in the dimension encoding a sequence of symbols, but because the cards con-
of tone but continuous in the dimension of time. trolled weaving machinery directly rather than encoding num-
In contrast, Jacquard looms are discrete in both di- bers the looms were not digital in the original and more literal 
mensions and hence fully digital – a fact that has led numerical sense of digitality.
feminist scholars such as Sadie Plant to stress continu-
ities between weaving and programming.19 The loom music box, making the hole bigger or smaller or moving 
weaves each row by attempting to thrust control rods the hole to an intermediate position could not produce 
through a punched card. This determines which threads analogous changes in the colors woven. After one step 
will be woven in that step of the process. Elaborate de- is woven the loom resets and advances to the next card, 
signs took thousands of cards. In this case the cards thus progressing discretely in the dimension of time.
are not numerically digital in the same way as the IBM 
punched cards discussed above were: each position on 
the card controlled a separate weaving hook but there 
was no scheme equating different combinations of 
holes to numerical values. But because it reads its con-
trol information in discrete rather than continuous 
form it is certainly digital in Shannon’s broader sense 
of symbolic digitality. In each of the many cards that are 
read to weave the design a hole is either present or ab-
sent in each position, which corresponds to the threads 
attached to the corresponding hook being woven or not 
woven during that step of the process.20 Just as with the 
19  Sadie Plant, zeros + ones (New York: Doubleday, 1997).
20  Birgit Schneider, “Digitality”, in Textile Terms: A Glossary 
(Berlin: Edition Imorde, 2017):71­75.
10    CRC Media of Cooperation Working Paper Series No. 31 JULY 2023
The woven portrait (Figure 4) looks a lot like a digital Conclusion
image, essentially because it is a digital image. Jacquard 
loom data was very similar to that used in monochrome 
bitmapped displays, such as those used on the Xerox Digital computers were digital because they used dis-
Alto or early Apple Macintosh models. These machines crete methods to represent quantities numerically, that 
drove their video displays from a bank of memory chips is to say: they worked with digits. Analog computers 
known as a frame buffer. Each bit in the frame buffer represented quantities by analogy, using continuous 
corresponded directly to one pixel on the display. Inter- variations.  Although the distinction between digital 
preting patterns as digits would have been meaningless. and analog was first made in the context of automatic 
A bitmapped monochrome image is digital in the sym- computers, the concepts were quickly broadened to ap-
bolic sense, but not in the numerical sense.21 ply to media and communication systems of all kinds. 
Shannon’s approach to digitality, or as he put it the 
transmission of information over discrete channels, 
was not tied to numbers. The crucial thing was that the 
message transmitted was coded as a sequence of sym-
bols taken from a fixed and finite set. 
Many digital media meet both definitions of digi-
tality, because they turn audio or video data into se-
quences of numbers and then store the numbers. But 
not all do, and Shannon’s own examples of textual en-
coding did not rely on turning the text into numbers 
before encoding it. In electronic engineering, all sys-
tems using logic gates and switching are understood 
as digital. Once the concept of machines controlled by 
digital media was created in the 1940s, earlier mechan-
ical technologies, most notably automatic looms, were 
recognized as having similar properties
Digitality here refers not just to the literal manipu-
lation of information encoded as numbers, but works 
more broadly to describe all situations in which a part 
of the world is read by mapping inputs onto one of a 
fixed, and usually small, number of possible states. 
These states are often interpreted as symbols. More 
complex or precise information is encoded and read not 
by introducing new symbols but by arranging symbols 
in sequence. This symbolic, non-numerical digitality 
underlies today’s digital media. It is to the emergence 
of the concept of digital storage media that we turn in 
the next working paper in this series.
21  Color displays are different. Modern color displays use 24 
bits per pixel for color information, coding the intensity of red, 
green, and blue as three numbers each ranging from 1 to 255. 
Altogether that gives 16,777,216 color variations. Hence the bits 
within a color pixel do have numerical significance. In contrast, 
the Jacquard loom image was colored but the picture was cre-
ated by overlaying a series of single-color images, each coded 
by one hole position per card.
Thomas Haigh, Sebastian Gießmann  11
References 
Bashe, Charles J, Lyle R Johnson, John H Palmer, and Emerson W 
Pugh. IBM’s Early Computers. Cambridge, MA: MIT Press, 1986.
Buckland, Michael. “Information As Thing.” Journal of the American 
Society of Information Science 42, no. 5 (June 1991): 351-360.
Galison, Peter. “The Ontology of the Enemy: Norbert Wiener and the 
Cybernetic Vision.” Critical Inquiry 21, no. 1 (Autumn 1994): 228-
266.
Geoghegan, Bernard. “Information.” In Digital Keywords: A Vocabu-
lary of Information Society & Culture, edited by Benjamin Peters, 
173-183. Princeton, NJ: Princeton University Press, 2016.
Gießmann, Sebastian. The Connectivity of Things: Network Cultures 
Since 1832. Cambridge, MA: MIT Press, forthcoming.
Haigh, Thomas, and Mark Priestley. “Colossus and Programmabil-
ity.” IEEE Annals of the History of Computing 40, no. 4 (Oct-Dec 
2018): 5-17.
Hamming, Richard W. “Error Detecting and Error Correcting Codes.” 
Bell System Technical Journal 29, no. 2 1950): 147-160.
Huffman, David. “A Method for the Construction of Minimum­
Redundancy Codes.” Proceedings of the IRE 40, no. 9 1952): 1098-
1101.
Kline, Ronald. The Cybernetics Moment, Or Why We Call Our Age the In-
formation Age: Johns Hopkins University Press, 2015.
Mills, Mara. “Deafening: Noise and the Engineering of Communica-
tion in the Telephone System.” Grey Room, no. 43 (Spring 2011): 
118-143.
———. “Media and Prosthesis: The Vocoder, the Artificial Larynx, 
and the History of Signal Processing.” Qui Parle 21, no. 1 (Fall/  
Winter 2012): 107­149. Authors
Nunberg, Geoffrey. “Farewell to the Information Age.” In The Future 
of the Book, 103-138. Berkeley: University of California Press, 1997.
Nyquist, Harry. “Certain Factors Affecting Telegraph Speed.” Bell Thomas Haigh is a Professor of History and Computer 
System Technical Journal 3, no. 2 (April 1924): 324-346. Science at the University of Wisconsin—Milwaukee and 
Pierce, J R. “The Early Days of Information Theory.” IEEE Transac- visiting Comenius Professor at Siegen University. Haigh 
tions on Information Theory 19, no. 1 (January 1973): 3-8. has published extensively on many aspects of the his-
Plant, Sadie. zeros + ones. New York: Doubleday, 1997. tory of computing and won several prizes for his arti-
Ruesch , Jurgen, and Wheldon Kees. Nonverbal Communication: Notes cles. He is the primary author of A New History of Modern 
on the Visual Perception of Human Relations. Berkeley, CA: Univer- Computing (MIT, 2021) and ENIAC in Action (MIT, 2016) 
sity of California Press, 1956. and the editor of Histories of Computing (Harvard 2011) 
Schneider, Birgit. “Digitality.” In Textile Terms: A Glossary, 71-75. and Exploring the Early Digital (Springer, 2019). Learn 
Berlin: Edition Imorde, 2017. more at www.tomandmaria.com/tom.  
Schüttpelz, Erhard. “’Get the message through’: From the Channel 
of Communication to the Message of the Medium, 1945–1960.” In Sebastian Gießmann is Reader in Media Theory at the 
Media, Culture, and Mediality: New Insights into the Current State of University of Siegen. In 2023, he serves as visiting pro-
Research, edited by Ludwig Jäger, Erika Linz and Irmela Schneider, fessor for cultural techniques and history of knowledge 
109-138. Bielefeld: transcript, 2010. at Berlin’s Humboldt University. His book Connectivity 
Shannon, Claude E. “A Mathematical Theory of Communication.” of Things: Network Cultures Since 1832 is forthcoming in 
The Bell System Technical Journal 27 (July & October 1948): 379- MIT Press’s Infrastructures series. Gießmann’s work 
423, 623-656. intertwines practice theory (which he helped to estab-
Soni, Jimmy, and Rob Goodman. A Mind at Play: How Claude Shan- lish within media studies), cultural techniques, Science 
non Invented the Information Age. New York, NY: Simon & Schuster, and Technology Studies, and grounded histories of 
2017. (digital) media. He is principal investigator of a major 
Wiener, Norbert. Cybernetics, or Control and Communication in the research project on the history of network infrastruc-
Animal and the Machine. Cambridge, MA: Technology Press, 1948. tures within Media of Cooperation.