In some of Korea’s subway stations, walls are plastered with posters that resemble the aisles and shelves of a supermarket. Only one thing is different—merchandise in the posters is marked with a QR code. This is Tesco’s idea of blending grocery shopping into people’s everyday lives:
Based on the simplicity principle of marketing communication, Tesco explains the process of virtual grocery shopping with a 3-step manual.
The explanation is purely commercial, saturated with action words informing the audience “what to do” and shunning definitions of “what is”. Apparently, “to scan” is not how the virtual shopping system works. It at most serves as a trigger of a series of actions, a nexus of actants. One crucial component of this system is the little black and white tag that we scan—QR code. Ironically, this key factor in the subway shopping experience happens to be a mark that we cannot read, a mark does not make any sense to humans. Then what is it?
However, a “what is” question is sometimes complex per se. By asking “what is a QR code” we may actually refer to three different but related layers:
- Functionality: what does QR code do?
- Definition: what does QR code mean in terms of its properties as information?
- Identification: what is QR code’s identity and position in the network of hybrid actants?
This paper sets out to explore how QR code serves to connect the audience and the information, but also dives deeper to probe into the nature of this particular connection and the network of dependencies that mediates this connection. By virtue of this teardown, this paper aims at scrutinizing QR code, utilizing angles and vocabularies from mediology and Actor-Network theory. Accordingly, this paper is structured as a continuum from technical teardown to theoretical interpretation and to extension to a related contentious issue: the Internet of Things.
Connection: Structure and Substructure
QR code at first sight is a bizarre picture, sometimes thought of as a “symbol”. But unlike most symbols that are meaningful, humans gain little information from QR code, be it the signifier or the signified. What could possibly be noticed and distinguished, however, is that in the corners of QR code locates three smaller squares, which indicates that there might be a structure and therefore a rule of constructing a QR code. Thus, we start from the observable.
QR code, described in general terms, is “a matrix type symbol with a cell structure in a square”. (Tan Jin Soon, QR Code) In fact, the cells are called “modules”, and the structure comprises five major components, each of which serves different function.
1) Finder Pattern
This is a pattern used for detecting the position of the QR code. It consists of three identical square boxes located at the upper left, upper right and lower left corners of the symbol respectively. The width between each box (named Position Detection Pattern by standard) is also stipulated as “The ratio of module widths in each Position Detection Pattern is 1: 1: 3: 1: 1”. (International Standard, ISO/IEC 18004)
2) Alignment Pattern
This pattern is set for correcting possible distortion when scanning the QR code, especially effective for correcting nonlinear distortions. The number of Alignment Patterns depends on the symbol size, which will be addressed later.
3) Timing Pattern
The horizontal and vertical Timing Patterns respectively consist of a one module wide row or column of alternating dark and light modules, commencing and ending with a dark module. The horizontal Timing Pattern runs across row 6 of the symbol between the separators for the upper Position Detection Patterns; the vertical Timing Pattern similarly runs down column 6 of the symbol between the separators for the left-hand Position Detection Patterns. They enable the symbol density and version to be determined and provide datum positions for determining module coordinates. (International Standard, ISO/IEC 18004)
4) Quiet Zone
It is a margin space for easier symbol detection, especially among other signs. Quiet zone is four-module wide, and surrounds the symbol on all four sides.
5) Data Area/Encoding Region
This is area where the modules store data. Data is encoded into binary value 0 and 1 based on certain rules. Then the binary codewords are converted into black and white modules. Also, this area contains Version information and Format information as well as error correction codewords.
However, the field of how QR code connects to purchase information in Tesco’s case is still left blank, even after elucidating the QR code structure in detail. It seems that what can be observed is far less than enough to understand the procedure triggered by scanning a QR code. Indeed, as a symbol that only readable by machine, QR code is powered precisely by the invisible.
Encode procedure converts input information/data to a QR code symbol, and initially enables connection between the black-and-white blocks and purchase information. One of QR code’s strengths is the compatibility with Kanji. Because it was first developed by Toyota (precisely, Denso Wave, Toyota’s subsidiary) in Japan, the code was designed by employing The Extended Channel Interpretation (ECI) protocol, which allows usually unsupported national character sets to be used in barcode symbols. This, however, requires the input data stream to be analyzed so that the most efficient encoding mode, not necessarily the most common modes, is utilized. Apart from ECI Mode, Numeric Mode and Alphanumeric Mode, QR code supports Kanji Mode in accordance with Japanese Industrial Standard X 0208, which is a standardized 2-byte character set for information exchange. Based on these built-in encoding modes and standards, input data is converted into a bit stream consists of binary numbers. Note that each mode runs different rules, and input data analysis is conducted to identify which mode to be put in force. That is, the strangely placed black-and-white modules is merely one presentation of rules and protocols running behind the scene. For clarity, two examples are cited from ISO/IEC 18004 to illustrate the complete encode procedure.
The resulting bit stream is then divided into 8-bit-long codewords followed by error correction codewords added to the message. The codewords are called symbol character, in contrast to function pattern which stands for data concerning Version information, Format information and so on. Next comes the question of how the codewords are represented into black-and-white modules. According to ISO/IEC 18004, symbol characters are positioned in two-module wide columns commencing at the lower right corner of the symbol and running alternately upwards and downwards from the right to the left. As is illustrated in Figure 7:
Based on various demands for data size, QR codes can be generated in 40 different symbol versions, from 21 x 21 modules (version 1) to 177 x 177 modules (version 40). Each higher symbol version has 4 additional modules per side (16 additional modules per symbol), and can contain a proportionally larger amount of data. (Denso Wave, QR Code Essentials)
In fact, what is listed here is merely the essential for understanding how a QR code is constructed, as this paper does not intend to be a manual book for coding engineers. However, it would be helpful to recap the procedure from the viewpoint of a mobile phone, the ostensible “decoder”. Now that data has been input and represented into the black-and-white modules, how does the mobile phone read them?
It is not hard to notice that while tearing down the structure of QR code we touch a substructure consists of standards, protocols and prior technical implementations such as error correction codewords. By examining this substructure, we transcend the sphere of technical details about QR code into a wider “historically reconfiguring system of media type and technologies”. (Martin Irvine, Working With Mediation Theory and Actor-Network Theory) In this case, the system is called mobile tagging.
Although tagging by mobile device is relatively recent, the implementation of code tagging can be traced back to one-dimensional barcode. In 1948, Bernard Silver attempted to develop a system that “automatically read product information during checkout” for Food Fair. (Charles Fishman, The Killer App) The first thought was ultraviolet ink, which proved to be fading and expensive. However, Silver later came up with the primitive model of bar code, and filed a patent for “Classifying Apparatus and Method”. It seems that bar code has ever since become a promising solution to tagging and automatically reading information. In 1974, the Universal Product Code (UPC) appeared on the pack of Wrigley chewing gum, and first scanned in Ohio. Before UPC was designed by IBM, the National Association of Food Chains in cooperation with McKinsey & Co. deployed an 11-digit code standard (Uniform Grocery Product Code) to guide barcode development. As a result, UPC became the standardized code and literally turned “universal”. Two industries, grocery and railroad, were the first to demand barcode: checkout for groceries, and tracking train cars for railroads. This suggests that code tagging is likely a result of pursuing efficiency and accuracy.
The development of two-dimensional barcode, without doubt, is based on this history. Actually, the designation of one-dimensional barcode is out of the fact that the information contained in them is communicated only by the difference in their horizontal dimension—the width of the bars and spaces—and their position from left to right. (Denso Wave, QR Code Essentials) And two-dimensional code, of course, adds a vertical dimension.
From one-dimensional to two-dimensional, the evolvement of barcode was provoked by the demand for larger capacity of data. In fact, attempts were consequently made to increase the amount of data contained in barcodes by increasing the number of bars or creating multiple-barcode layouts. These efforts, however, resulted in a larger barcode area, complicated reading requirements and increased printing costs. To solve these problems, two-dimensional codes were developed, first as stacked barcodes, which repeat the same linear symbology vertically, and then as matrix codes, composed of small, symmetrical elements arranged in a square or rectangle. (Denso Wave, QR Code Essentials) The timeline of this development can roughly be depicted as Figure 9
Along with the barcode’s timeline, the invisible dependencies gradually manifest themselves. Similar timeline could be depicted for the technical devices for reading barcodes, from the light bulb and photomultiplier used by Silver (Kevin Roebuck, QR Code) to checkout scanners and to the mobile phone we carry everyday. Even just to answer the simple question about QR code’s functionality, it is imperative to dive into the substructure of the whole mobile tagging system. Clearly, technical details always contain much more than technical visions.
Therefore, by virtue of a series of standards, protocols and prior implementations, QR code is in essence a method to abstract and store information, which avoids human interaction for the sake of efficiency and accuracy. Frequently utilized to represent URLs, QR code is commonly known as a connection between the audience and online information. However, it is not the black-and-white symbol itself that enables the connection; rather, it is the invisible substructure of mobile tagging system that empowers the symbol.
Interface: Representation and Subprime Language
It is slightly arbitrary though, to conclude that connection is the functionality of QR code, because “connection” somehow suggests a gap between the symbol and the information, while the symbol per se could be information. Consider QR code’s usage other than representing URLs. Bus commuters pass, for instance, utilizes QR code to carry the commuter’s name, commuter fare, and destination. By scanning the code, the system automatically obtains information about the commute. Therefore, QR code is indeed information, only cannot be read by humans. But readable to humans is not a prerequisite for being information. In fact, as a commonly accepted definition of information, General Definition of Information (GDI) notes:
Information is made of data. The data is well formed, and “well formed” means that the data are rightly put together according to the rules (syntax) that govern the chosen system, code, or language being used. (Luciano Floridi, Information: A Very Short Introduction)
Syntax here does not necessarily refer to language syntax; rather, it could be, as in the case of QR code, the international standards, encoding modes and so on. From this perspective, and a relatively tenable one, QR code is information per se.
However, focusing on the distinction between communication and transmission (Régis Debray, Transmitting Culture), one can still argue that the information transmitted by QR code is essentially different from what we directly obtain from posters or on websites. In other words, QR code stands for a type of information that requires “translation”. Here we seem to naturally refer to human language as a metaphor. Interestingly, the same metaphor is used by the inventor of QR code:
Like written language, barcodes are visual representations of information. Unlike language, however, which humans can read, barcodes are designed to be read and understood (decoded) by computers, using machine-vision systems consisting of optical laser scanners or cameras and barcode-interpreting software. The rules with which a barcode is constructed (its grammar) and the character set it uses (its alphabet) are called its symbology. (Denso Wave, QR Code Essentials)
Along this thread of thinking, despite that barcodes are comparable to human natural languages, QR code is virtually a “secondary” system that comprised by representations of human languages. Thus, it can be viewed as a “subprime language”.
Recall the encoding modes available to QR code, and the encoding procedure in effect turns out to be a procedure of “translating” human natural language to machine-readable symbols, or an artificial language. Perhaps a more accurate word here is “abstract” instead of “translate”. One-dimensional barcode, for example, abstracts numeric symbols to black stripes so that they can be read and tracked efficiently and accurately. The same method can actually be traced back to Morse code. International Morse Code encodes the ISO basic Latin alphabet, some extra Latin letters, the Arabic numerals and a small set of punctuation and procedural signals as standardized sequences of short and long signals called “dots” and “dashes”. (Wikipedia)
Visual symbols of human language were thereby abstracted to audile signals while the grammar was kept. Because of its relatively simple encoding mode, Morse code can be understood by experienced listener or observer without special equipment. While barcode systems obviously have more dependency on devices, the idea of abstraction—abstract from human natural language and then apply to artificial language—remains the same. In fact, Bernard Silver who first developed barcode had this inspiration precisely from Morse code. “I just extended the dots and dashes downwards and made narrow lines and wide lines out of them,” Silver remembered. (Tony Seideman, Barcodes Sweep the World) Therefore, as information, QR code’s property of human-natural-language-abstraction must also be highlighted. Humans cannot directly obtain information from QR code; instead, an “intermediary” is required, either a decoding machine or a hyperlink.
This tailored definition of QR code helps elucidate one particularly ambiguous aspect of information transmission. When data is encoded in a QR code, and the QR code later scanned by the audience, can we say the audience receives the information? She probably has it, but not necessarily obtains it. It is because QR code is made up of abstractions as opposed to language itself that there is detour around communication and understanding. This is further illustrated by the case where QR code represents URLs. The message the sender expects to reach the receiver exists not in the symbol (QR code), but still on the webpage the URL links to. QR code here serves as a note of hyperlink, remediating the old functionality of information storage. Like written notes, it helps offload human memory, distribute memorizing effort to a symbol. But it certainly surpasses written notes as it remits the effort of writing down and the possible error accompanied. Ironically, however, this error-free technique cannot guarantee that information is truly received. Similar to information service such as “Read It Later” (currently known as Pocket), QR code tends to provoke the possession of information, but necessarily digestion of information. As the amount of information grows at exponential rate, we seem to adapt to “see” it, rather than read it. More often than not, we say “I have seen it (somewhere)” indicating a possession of information; seldom do people say “I have read it (and understood)”—getting information has become increasingly easier, but digesting it increasingly harder.
Emphasis on its properties as information leads to a more profound definition of QR code. (1) QR code is an artificial language, and the symbol stands for abstractions of human natural language. (2) The abstractions of human natural language are mediated by a range of standards and encoding rules. (3) Mechanical devices are needed to decode the symbol; that is, to restore the abstractions to its former form of human natural languages. (4) QR code is commonly adopted as a representation of URLs, in which the symbol stands for a hyperlink not the information.
From point (1) (2) and (3) we conclude that QR code in essence is a place where data are stored, ready to be interpreted by the decoder. It directly represents the information that the sender intends to emit. Point (4) presents a relatively simple situation where QR code serves as a transit spot storing the location of the desired information instead of the information itself. In both cases, however, external mediators are involved. Therefore, it seems more accurate to define QR code as an interface to information.
Back to the case of Tesco, is QR code that makes “virtual grocery shopping” possible? Defining QR code as interface to information may cast some light upon this question. In its promotional video, QR code is spotlighted since it seems to be the key factor that mediates the virtual purchase experience. Similarly framed is the 3-step manual. However, in reality, QR code is an interface to online information, a connection to Tesco’s inventory. What delivers the merchandise is still the logistics (or Tesco as a whole), not the QR code. Order information delivered to inventory, and the logistics delivers the merchandise: pretty much like an ordinary trip to grocery store. When we scan the QR code, the decoder (mobile phone in this case) obtains access to Tesco’s inventory information, which can also be obtained by browsing webpage, or by asking in person. QR code as an interface, of course, showcases a more efficient and accurate way to achieve the same goal. Is there really a virtual subway Tesco? Perhaps not. On the walls of the subway stations are symbologically represented accesses, interfaces to a long-lasting daily experience. There is no virtual Tesco. There is, a Tesco.
Actant: QR Code and More
Thinking of QR code as a “subprime language” is interesting as it opens up a stage for more cases that are being fiercely discussed, such as The Internet of Things. By virtue of abstraction, QR code is an artificial language made for machines. Along this line, The Internet of Things (IoT) can be viewed as an artificial environment for machines. By definition, IoT is a novel paradigm that is rapidly gaining ground in the scenario of modern wireless telecommunications. The basic idea of this concept is the pervasive presence around us of a variety of things or objects—such as Radio-Frequency IDentification (RFID) tags, sensors, actuators, mobile phones, etc.—which, through unique addressing schemes, are able to interact with each other and cooperate with their neighbors to reach common goals. (Luigi Atzori et al, The Internet of Things: A Survey)
IoT is extensively discussed in part because it is perhaps the closest implementation to intelligent ambience. Some of the applications listed below have already been realized.
As noted by analyst house Gartner in a very recent report, IoT will far outstrip smartphone, tablets and PCs over the coming years as the installed base is set to hit 26 billion units by 2020 as the market continues to grow at a phenomenal rate. (ITProportal)
However, increased intelligence in things has raised concerns about technology affecting human agency, privacy and autonomy. (Peter-Paul Verbeek, Moralizing Technology: Understanding and Designing the Morality of Things) The same question has been asked again: Are the intelligent things/machines/devices assistants to humans, or are they themselves active agents?
QR code might be too trivial to be involved into this discourse. However, ultimately the anxiety QR code poses is identical to that posed by IoT. Although barcodes including QR code are meant to eliminate errors and thus enhancing efficiency, the result after all turns to a void of humans—cashiers are replaced by self-checkout machines. It seems that human interactions have to be diminished in trade for efficiency. IoT apparently follows this solution, if not furthers it, by lending machine the central position of decision-making. Therefore, it is natural for humans to feel uncomfortable when coping with a situation void of human interactions. So is this the identity of QR code and its peers? The terminator of human agency?
The answer lies in the history of technology. On one hand, inscribed in QR code is a program of human actions; that is, QR code is essentially distributed human agency, rather than the terminator of human agency. On the other hand, the whole history of techniques is occupied by this sort of hybrid actants: human with techniques, and techniques with human. It is actually common for the human actor being absent, yet present simultaneously because of the existence of nonhuman actant. It might be scary when shopping at CVS with the only “human” voice coming from a machine, saying “please choose your language”. However, closer observation of technology history would reveals that It is us, the human makers that you see in those machines, those implements, us under another guise, our own hard work. (Bruno Latour, On Technical Mediation: Philosophy, Sociology, Genealogy) As opposed to being scary, we might feel instead familiar as “our delegation of action to other actants that now share our human existence is so far progressed that a program of antifetishism could only lead us to a nonhuman world, a world before the mediation of artifacts, a world of baboons”. (Bruno Latour, On Technical Mediation: Philosophy, Sociology, Genealogy) Perhaps objectivity and subjectivity have never been strictly opposed; QR code is the interface to information as well as to humans; The Internet of Things is a network of objects as well as us.
(Figure 12: What has been changed by technologies? A series of illustrations depicting our complicated relationship with technologies.)
The consumer, who is scanning the QR code printed on Tesco posters, perhaps could never have thought of the fact she is scanning an invisible nexus of dependencies, standards, and sociotechnical history. But starting from such a case, this paper closely examines the substructure of algorithm, standards and symbology, which mediates the connecting function of QR code. Functionality alone, however, cannot answer the question of “what is QR code”.
Diving into the definition of information, this paper argues that QR code is in essence an interface to information. This clarification further leads to thoughts about QR code’s identity of an actant. Although man-made “subprime” systems tend to produce a context void of human, the man-made property of these systems reminds us of their identity of distributed human agency. Even for far more complicated problems like The Internet of Things, this property does not change. In fact, we are in a ever-lasting relationship with technologies, maybe rare in the romantic sense, but it is indeed an intervolving relationship.
Ashton, Kevin. “That ‘Internet of Things’ Thing, in the Real World Things Matter More Than Ideas.” RFID Journal (June 22, 2009).
Atzori, Luigi, Antonio Iera, and Giacomo Morabito. “The Internet of Things: A Survey.” Computer Networks no. May (2010).
Debray, Régis. Transmitting Culture. New York: Columbia University Press, 2000.
Denso. “QR Code Essentials,” 2011.
Floridi, Luciano. Information: A Very Short Introduction. Oxford: Oxford University Press, 2010.
“International Standard ISO/IEC 18004,” 2000.
“Internet of Things–An Action Plan for Europe,” 2009.
Irvine, Martin. “Working With Mediation Theory and Actor-Network Theory,” (2013).
Latour, Bruno. “On Technical Mediation.” Common Knowledge, 1994.
Latour, Bruno. “Technology Is Society Made Durable.” In A Sociology of Monsters: Essays on Power, Technology and Domination, 103–131. John Law. London; New York: Routledge, 1991.
Manovich, Lev. Software Takes Command: Extending the Language of New Media. London; New York: Bloomsbury Academic, 2013.
Roebuck, Kevin. “QR Code: What You Need to Know,” n.d.
Tan, Jin Soon. “QR Code.” Synthesis Journal (2008).