Ted Nelson, the inventor of the hyperlink, wrote in 1974: “at no previous time has it been possible to create learning resources so responsive and interesting, or to give such free play to the student’s initiative as we may now” (Nelson, 1974). Half a century later, the reality we find ourselves in seems, in terms of technical possibilities, to match those expectations, yet the impression persists that we are not living in the educational paradise envisioned by the pioneers of the internet. In his book of the same name, Seymour Papert described the computer as a “Children’s Machine” (Papert, 1993). New, networked forms of representing knowledge, he argued, would finally make it possible to leave behind rote memorization, often experienced as boring. Education would become more equal, more exciting, more child-centered. Today, however, young people find themselves amid an overabundance of information and media offerings on the internet and in social networks. Floridi and colleagues (Floridi et al., 2015) describe this condition in the Onlife Manifesto as a present in which the boundaries between online and offline reality are increasingly disappearing. According to this view, the distinction between the real and the virtual, between human, machine, and nature, becomes blurred, while society shifts from a shortage of information to an overabundance of it, in which interactions and relationships become more important than the interactors themselves.

Original future visions of educational technologies were developed without awareness of this new reality. In relation to a criticized endless cycle of innovation and unmet expectations (Selwyn, 2023), researchers ask what prevents us from making meaningful use of new technical possibilities and unlocking educational potentials that have remained untapped so far. This article therefore addresses the question of how discrepancies and blind spots can be uncovered in models of educational technology that are still in use today.

The starting point for this discussion is the influence of behaviorism on learning applications. This influence is presented only to a limited extent, since later constructivist models often define themselves in opposition to this tradition. A detailed discussion of the behaviorist teaching machine can be found, for example, in Ferster (2014) and Watters (2021). This is followed by an introduction to Papert’s influential theory of constructionism, on which many current research projects are based. In the decades before that, many technology pioneers developed their own rudimentary, implicit learning theories as part of their design work, which can be linked to constructionism. The main part of this article focuses on an analysis of Nelson’s reflections on hypermedia, since today almost all educational media, whether behaviorist or constructivist, use hyperlinks. Hyperlinks, that is, cross-references that allow jumping to specific places in another (or the same) electronic document, are a central component of the World Wide Web. In this context, other media that use hyperlinks are also referred to as hypermedia. Through a critical comparison of the original concepts and present-day educational practice, this article shows how frequently occurring implicit simplifications in the design and use of educational technologies can be avoided.

In general, models of technology-supported learning can be divided into two categories. The first concerns learning through optimization and focuses on adapting learning processes so that they become more efficient. The second category aims to use new media to open up new, more “natural” forms of learning in which children construct their own learning processes.

The typical basic model of the analog teaching machine is structured as follows: inside the box-shaped machine with windows are various paper discs containing questions and answers written along their radii. One question at a time appears in the central window. The student writes an answer on a paper strip next to it and then turns the disc forward. This reveals the correct answer, but covers the student’s own response so it can no longer be changed. In other versions of the teaching machine, there are additional mechanisms that only allow turning forward if the answer is correct (Skinner, 1953, p. 40). With new devices such as laptops, PCs, and tablets, the idea of programmed learning following the principles of the teaching machine benefits from expanded fields of application and practice scenarios, since various forms of such linear and nonlinear learning programs can be implemented on a large scale at a fraction of the cost.

The basic principle of programmed learning can still be found today in many current learning and practice applications, for example in the Anton learning app (Solocode, 2024), which became widely known during the COVID-19 pandemic. In this app, children solve exercises in subjects such as German, mathematics, and general studies and can advance to the next level of difficulty depending on their success.

It is striking that Sydney Pressey, known for designing an early teaching machine in the 1920s (Pressey, 1927), criticized programmed learning because of its shortcomings with regard to medium- and long-term memory. Instead of applying learning mechanisms from animal experiments to humans, he argued for making use of specifically human abilities such as language, metacognition, and silent reading in self-directed learning (Pressey, 1963, p. 5). Nevertheless, it can be useful to take into account the learning mechanisms studied by Skinner in the preparation of learning environments. Some design principles for Skinner’s learning programs can also be used as a complement to other methods.

The rules Skinner developed for programmed learning have influenced didactic, especially media-didactic, principles (Niegemann, 2008, p. 5; cf. Hilgard & Bower, 1977, pp. 232–233; Skinner, 1958):

1. A clear, detailed, and objective description must first be developed of what it means to have mastered the subject matter.

2. A sequence of question-answer frames must be developed in which the learner is confronted with the material in gradually increasing difficulty and in which the same facts are presented repeatedly from different perspectives.

3. It must be ensured that the learner is active, for example by requiring an answer to a question or task in every frame.

4. Every answer must be followed by immediate feedback.

5. Questions should be formulated so that the correct answer has a very high probability of occurring.

6. Each learner should go through the program at their own pace.

7. Working with the program should be encouraged through additional reinforcements (rewards).

Teachers can apply these principles in their instruction. They can also serve as criteria for classifying educational media; whenever digital media take over or simplify parts of the tasks listed above, the issue is generally the optimization of an existing learning process. When learning new and complex subject matter, the challenge lies, in accordance with the second principle, in creating a gradual progression of cards that step by step expands the child’s understanding. Steering this process individually for every child, however, represents an enormous challenge. This also shows why applications for programmed learning have become established more for practicing and consolidating knowledge than for introducing new concepts.

Some authors point out that many applications implementing “personalized learning” in the form of gamified interfaces are based on principles derived from traditional behaviorism (Brass & Lynch, 2020; Manolev et al., 2019), while ignoring core criticisms such as the orientation toward behavioral control instead of intrinsic motivation. Parallels can also be drawn with the idea of cybernetic learning (Cube, 1965), computer-assisted instruction (CAI), and modern “skill-and-drill” learning applications from the edutainment sector, for which the metaphor “broccoli coated in chocolate” is often used (Laurel, 2001). According to critics, this kind of educational technology adopts an epistemological position according to which knowledge can be broken down into atomic components, usually called facts, concepts, and skills, which are then to be conveyed to children at a particular pace (Papert, 1993, p. 63). The following section takes a closer look at the perspective that attempts to provide a direct response to this objection.

If we wanted to find a theoretical foundation for the use of hypermedia in educational contexts, a reference to behaviorism and programmed instruction alone would be insufficient, even though many tutoring systems today are designed as web applications. What is still missing are perspectives that emphasize the importance of self-directed, personal learning and discovery in the learning process. In this context, situated learning describes the condition that successful learning takes place in real, authentic contexts that are meaningful to learners. Such approaches are often shaped by constructivist learning theories, which assume that individuals construct their knowledge and understanding through personal experiences and interactions with their environment.

For example, Seymour Papert argues that progressive educational approaches such as those associated with Paolo Freire, John Dewey, or Maria Montessori, despite being correct, could not be applied to schooling in general for purely practical reasons (Papert, 1993, p. 14):

With very limited means at their disposal, they were forced to rely too heavily on the specific talents of individual teachers or a specific match with a particular social context. As a result, what successes they had often could not be generalized. [...] like the sense in which Leonardo da Vinci failed in his attempts to invent an airplane [...]. His failure to make a workable airplane did not prove him wrong in his assumptions about the feasibility of flying machines.

Papert sees the missing element for an “uplifted” education, much like the missing element in da Vinci’s dream of flying, in the invention of the computer. Yet the mistake, he says, must not be made of installing a “jet engine on a horse carriage” using old pedagogy (Papert, 1993, p. 29). On the contrary, new pedagogical forms of learning made possible by digital media should be used.

A key concept Papert uses (1993, p. 17) is that of microworlds. Microworlds help learners develop a deeper understanding of a subject. As “incubators for knowledge” (Papert, 1985, p. 152), microworlds are protected spaces that simulate real phenomena or concepts and within which children can experiment freely. What matters is that they (1) allow discovery without fear, (2) make knowledge and subject matter experientially accessible through one’s own activity, and (3) are perceived by children as personally relevant and meaningful. Papert’s ideal vision of school is a set of interconnected microworlds in which children do not experience knowledge as an end in itself, but understand it as a means of acting creatively within these environments (Papert, 1985, p. 168).

Microworlds offer students the opportunity to construct and expand their knowledge through practical experience. They also enable learners to pursue their own interests and adapt their learning individually, leading to an active and engaged learning process. One example is the programming language Logo, with which the movement of a turtle could be programmed. The range of microworlds that can be realized with Logo extends from simple programming of geometric forms to the simulation of physical experiments (Papert, 1985, chapter 7).

What is striking about Logo and later graphical programming environments such as Scratch (Resnick & Robinson, 2017) is that they do not prescribe solutions to children, but instead allow them to explore their own paths to solving a problem. In this way, children are meant to develop their own interest in problems or tasks that appear authentic to them. However, exactly how this happens, and what role social interactions among children or with teachers play in bringing the significance of a problem authentically to the children, initially remains unanswered.

Computers can be used not only to explore problems in special environments, but would also enable children to have free access to all conceivable bodies of knowledge. Papert speaks of the vision of a “knowledge machine,” which he sketches in outline:

Such a system would enable a Jennifer of the future to explore a world significantly richer than what I was offered by my printed books. Using speech, touch, or gestures, she would steer the machine to the topic of interest, quickly navigating through a knowledge space much broader than the contents of any printed encyclopedia. Whether she is interested in giraffes or panthers or fleas, [...] she would be able to find her way to the relevant sounds and images she believes would help her understand what she wants to understand. (Papert, 1993, p. 5)

One cannot escape the impression that Jennifer, three decades later, does indeed have the possibility of exploring various animal species in such a knowledge space. The navigation through a knowledge space envisioned by Papert has gradually become reality. Ted Nelson had already conceived the hyperlink in the 1960s, a basic component of all web-based media today, as the connecting element of electronic documents. Apple’s HyperCard software, released in 1987, was the first widely available hypertext system, for which a wide range of educational applications were developed (Bowers & Tsai, 1990). The first version of the public HyperText Markup Language (HTML) in 1993 paved the way for the networked World Wide Web as we know it today. In 2001, Wikipedia followed as the first globally accessible collaborative online encyclopedia. Since 2014, HTML5 has natively supported the integration of audio and video, making visual context recognition in search engines (for example via Google Lens) feasible shortly afterward. Integrated applications for speech output and recognition are becoming increasingly reliable and easier to use. Teachers today are, at least technically, in a position to curate or even program constructionist microworlds or systems of programmed instruction on the basis of hypertext technologies (for example with the software Twine mentioned in the next section). How they shape these pedagogically, however, is often left up to them. The traditions outlined above provide a framework for orientation, but are often not sufficient for using these tools effectively and purposefully.

The following section therefore compares the original idea of hypertext with today’s technical possibilities. In doing so, it also brings into focus the question of what circumstances favor or hinder students in actually exploring this enriched world of the “knowledge machine.”

Ted Nelson’s work Computer Lib/Dream Machines (Nelson, 1974; Nelson, 2003) is particularly well suited for identifying blind spots in earlier future visions. In Nelson’s vision, hypertext was meant to replace traditional linear educational materials such as textbooks and lectures and allow students to navigate through a network of interconnected information and concepts. First self-published in an unusual large format, the book contains no conventional content structure, but instead sought to reproduce the interwoven nature of future hypermedia. With fanzine-style sketches and descriptions, Nelson presented his ideas for the future of computers and education. In particular, he sketched several possible educational applications to be realized with the help of hyperlinks as hypermedia.

In the section in which Nelson presents his vision of learning and teaching, different representational forms can be found. The following is a list of these proposals with a brief description:

• Discrete hypertexts: text passages and blocks connected through links

• Hypergram: a dynamic diagram that zooms and reveals more detailed descriptions as a result of user input (for example touching the screen)

• Stretchtext: a book that can be navigated not only forward and backward, but also in its level of detail. A short version is gradually supplemented with more detailed sentences depending on the setting.

• Hypermap: a zoomable, navigable world map in which specific overlays display additional information (for example population, climate, industry).

• Hypercomics: a visual, branching (nonlinear) narrative in which a story can be followed, for example, from the perspectives of different characters.

For each of these applications, several technical realizations could be named today. While every modern website is in fact a discrete hypertext, geographic information systems such as OpenStreetMap can display spatial information in the way described for the hypermap. Hypercomics are not used in textbooks, but they can be realized with systems for creating nonlinear narratives such as Twine (Interactive Fiction Technology Foundation IFTF, 2022) or Ren’Py (Ren’Py, 2024). They are also known under the name “visual novels” and enjoy great popularity in the field of independent games.

Only stretchtexts were not realized in the form described, though they could be produced from open encyclopedias such as Wikipedia with a certain amount of editorial effort. In this representation, the possibilities of other modes such as audio and video were not taken into account. Yet children and adolescents now spend most of their internet time consuming videos (Medienpädagogischer Forschungsverbund Südwest [mpfs], 2023, p. 30). It is worth noting here that videos on the internet are usually linear in structure. The interconnectedness of the hypermedium is not represented within the video itself. Only in relation to other videos, such as on social media platforms, can they be dynamically linked to one another through corresponding algorithms.

These different technical descriptions are closely tied to Nelson’s perspective on learning with media. The contrast mentioned earlier between learning through optimization and new forms of learning appears from his point of view in three different representations: classical learning with a teacher, computer-supported learning, and learning with hypermedia (Fig. 1).

Figure 1: Teaching with hypermedia, quoted from Nelson (1974, p. 101)

The figure suggests that the new, more direct forms of learning enabled by hypermedia represent progress compared with classical instruction and computer-supported learning. What is striking here is that the teacher is depicted as a wall that both conveys knowledge and shields students from unrestricted access to knowledge. In the two computer-supported variants, the teacher then disappears entirely.

The blind spots in the conception of hypertext provide an occasion to highlight relevant points of criticism in the discourse on educational technology. In contrast to the time of the teaching machine, modernization efforts before and after the turn of the millennium have been driven above all by reform ideas oriented toward constructivism (Cuban, 2001, p. 14). The criticisms therefore relate to weaknesses in both of the traditions described above. Since constructivist approaches currently predominate, the second part of this section places particular emphasis on the neglected weaknesses of this tradition.

Hypertext (Nelson), blind spotsBehaviorism (Skinner)Constructionism (Papert)How is interest generated?Rewards (positive reinforcements)Inherently givenWho mediates between children and content?Gradual progression of frames that introduce new contentLearning as reconstruction of knowledge through makingHow is information filtered?Through the selection of the programThrough the preparation of microworlds

Table 1: Comparison of theoretical differences

Table 1 summarizes the theoretical differences that emerge from the open questions surrounding hypertext. In terms of how interest is generated, behaviorism answers this through positive reinforcement. Children are motivated by external incentives that encourage desired behavior. On the constructivist side, self-determined learning is said to correspond to children’s already existing, natural interest. The excess of information is solved on the one hand through the selection of suitable programs, and on the other through the boundedness of microworlds or learning environments. This overview makes it possible to differentiate individual design patterns that respond to the three problem constellations identified above. This discussion makes it clear that these design patterns are based on simplified conceptions of technology.

Simplified conceptions of technology make its integration into teaching more difficult. The notion of digital tools as neutral instruments has been criticized in various ways. For example, critical EdTech research points to the pedagogical implications of educational technology that are often overlooked (Eynon, 2018). It must be taken into account that modern classrooms are still influenced by a “technological frame” (Bijker, 2010) whose origins lie in the historical beginnings of public schooling. Many traditional elements such as the blackboard, linear desk arrangements, and textbooks dominate our idea of how classrooms can be designed. Moreover, this technological frame of reference also influences the introduction of digital media in schools. In other words, old patterns of practice often serve as the model for new technologies. As Perrotta et al. (2020, p. 7) show in relation to Google Classroom, one of the most widespread learning management and remote teaching platforms, the established mechanisms and structures of formal education—classrooms, coursework, student submissions, and the asymmetrical relationship between teachers and students—are arranged into a predefined template that determines the possibilities of participation. Weller (2020, p. 65) draws a parallel between this general phenomenon and what Lanier (2002) referred to in software development as “software sedimentation.”

In this context, sedimentation refers to structures that emerge in progressively developed systems and are based on old ideas. These deposits, which can include components such as roadmaps, guidelines, work practices, and qualification programs, contribute to preserving old patterns. This must be taken into account when implementing new processes or media. At the same time, it is also one reason why technological solutionism (Morozov, 2013) is another source of inadequate conceptions of technology. If innovations are not tailored to their specific context, obstacles inevitably arise that are associated with precisely those sedimentations.

The problem of overly general assumptions became evident, among other places, in the One Laptop per Child (OLPC) project. The main goal of the initiative launched in 2005 was to develop a robust and affordable laptop specifically for use in schools in resource-poor regions. After some years, however, it became clear that the high expectations attached to the project could not be fulfilled. The intended cost reduction to 100 US dollars per laptop was never achieved, and the first large-scale evaluation study in Peru, covering more than 319 schools equipped with laptops, found no significant effects on attendance, mathematics, or language acquisition (Cristia et al., 2012, pp. 15–16). Ames (2019) traces how the attraction, or “charismatic authority,” of the OLPC initiative was simultaneously the reason for its failure, because implicit assumptions and ideas about how children learn were transferred uncritically from the hacker culture of MIT in the United States to the context of developing countries. Thus, instead of being shaped by sound pilot studies, fatal design decisions were largely determined by nostalgia and the childhood memories of the project’s initiators (Ames, 2019, p. 49). In addition to technodeterminist discourses on educational technology, many innovations share a problem that can be described as “historical amnesia” (Weller, 2020, p. 3), that is, the inability to learn from past mistakes and the failures of other projects.

As already noted with regard to programmed instruction, this has also been observed in certain educational software. Not without reason, Gee (2005) describes successful commercial entertainment-oriented video games as the actual “learning machines” that should be imitated. But even if one leaves aside digital skill-and-drill, further approaches still cannot be generalized to all learner groups. Ames (2018), on the other hand, highlights recurring problems in the adoption of constructivist ideas in the name of new forms of learning. This concerns, for example, the assumption that computers and digital tools have a “universal” appeal for children and that they would engage with them independently without difficulty; in fact, this often applies only to a portion of students who are capable of learning without guidance (Ames, 2018, p. 10).

Sims (2017) argues that processes of school reform often go through “cycles of disruptive fixation.” These cycles begin with a phase of “problematization and technical implementation,” in which experts and allies design and present imagined worlds that are particularly controllable with the proposed means (Sims, 2017, p. 58). At the same time, however, these worlds contain many implicit assumptions that later prove problematic in implementation. Thus, gaps in the vision and planning also function as aids to implementation, because they help generate sufficient support at the beginning of projects.

In other words, only those projects that exaggerate their advantages while downplaying risks and complexities will attract enough supporters and resources to become reality. This raises the question whether this applies not only to individual projects, but to a certain extent also to learning theories in general. Pressey’s criticism of operant conditioning, cited above, includes the point that children’s mental functions are equated with those of animals, without making use of the creative, playful, and narrative potentials of human learning processes. Constructionism, on the other hand, presupposes an inherent curiosity and interest in technological novelty on the part of children. While one theory tends to underestimate children, the other tends to overestimate them. In both cases, this type of simplification serves the purpose of portraying learning processes either as controllable or as indefinitely expandable.

To avoid this contradiction, the implicit assumptions often contained in the underlying theories must be questioned. What type of learner can really benefit from this or that digital form of learning? Further, the question is also what digital tools are being used for.

The initial positioning in the “onlife” or in digitality intensifies the tensions named above. If the boundary between the real and the virtual becomes blurred, this simultaneously raises the question under what different conditions real or virtual objects are suitable for generating interest. The fading boundary between human, machine, and nature makes it more difficult to locate precisely the reconstruction of knowledge or the learning process. At the same time, the overabundance of information and stimuli confronts teachers with the task of finding suitable methods of selection and filtering that are adapted to the different needs of individual students.

Teachers can respond differently to the aspects mentioned above. In reference to Nelson’s depiction, these are summarized in Figure 2.

Teachers play a decisive role in bringing students to new knowledge. By asking exciting questions, introducing interesting topics, and establishing relevant connections to everyday life, they can stimulate students’ curiosity and encourage their willingness to seek further information.

With their contextual knowledge and familiarity with their students, they can make a preliminary selection of online resources, interactive learning platforms, and multimedia tools to support the learning process. In this way, students can be relieved of the burden of having to find the learning strategy most suitable for them.

With regard to the third aspect, teachers act in the field of media pedagogy. They can teach information literacy skills that enable students to find, evaluate, and use relevant information. By teaching strategies of critical reflection and responsible engagement with digital media, teachers can help students find their way in a world full of information and make well-founded decisions.

In addition to the tasks of teachers referred to in the educational technology models discussed here, a new responsibility emerges that expands the teacher’s media-didactic action into a broader form of media education: instead of moving students to learn through incentives and motivation, the aim is to minimize the manipulability of students by digital applications and, for example, to raise awareness of dark patterns.

The expanded role of the teacher in mediating the theoretical gaps outlined above constitutes a necessary response to the challenges of the digital world. But it also carries the danger of remaining trapped in traditional ideas of school and instruction. The task here is not to reduce pedagogy to the metaphor of knowledge transfer between teacher and student, but to give students, through the use of digital media, sufficient space for initiative, collaboration, and design processes that they themselves experience as important and meaningful.

This article has explored the question of what educational potentials were originally embedded in different conceptions of educational technology and what obstacles impaired their implementation. Simplified conceptions of technology make their integration into teaching more difficult. Reality often diverges from original technological expectations. The two traditions sketched here—those aiming on the one hand to optimize learning processes and on the other to promote constructivist forms of learning—are marked by different points of criticism.

Overall, the use of digital media in education is characterized by a tendency for sedimentations to restrict conceptions of what can realistically be integrated. On the other hand, a form of historical forgetting is also evident, since problematic approaches for particular contexts are often repeated in new form. Frequently, implicit assumptions are made about students that hinder the differentiated and effective use of digital media.

To make these premises explicit, however, it is useful to analyze historical representations of original predictions concerning educational technology. Here, the choice fell on Ted Nelson’s Computer Lib/Dream Machines, which formulates especially vividly the expected educational potentials of hypermedia. Even though most of the innovations presented there were realized in one way or another, blind spots can still be identified to which interdisciplinary educational research can and should respond.

These include the question of how interest is generated, the mediating link between content and student, and saturation through information offerings. A differentiated perspective, critical engagement, and ongoing research are necessary, with research needs lying more strongly on the user side than on the technical side and the expansion of technical possibilities. This is connected with new conceptions of roles and responsibilities for teachers. It remains an important task to shape the integration of digital media into education carefully, while keeping the needs of students in view.

Ames, M. G. (2018). Hackers, Computers, and Cooperation: A Critical History of Logo and Constructionist Learning. Proceedings of the ACM on Human-Computer Interaction, 2(CSCW), 1–19. Ames, M. G. (2019). The charisma machine: The life, death, and legacy of One Laptop per Child. MIT Press. Biesta, G. (2020). Risking Ourselves in Education: Qualification, Socialization, and Subjectification Revisited. Educational Theory, 70(1), 89–104. https://doi.org/10.1111/edth.12411 Bijker, W. E. (2010). How is technology made?—That is the question! Cambridge Journal of Economics, 34(1), 63–76. Bolter, J. D. & Grusin, R. (1998). Remediation: Understanding New Media. MIT Press. Bowers, D. & Tsai, C. (1990). HyperCard in educational research: An introduction and case study. Educational Technology, 30(2), 19–24. Brass, J. & Lynch, T. L. (2020). Personalized learning: A history of the present. Journal of Curriculum Theorizing, 35(2), 3–21. Burbules, N. C. & Callister, T. (1996). Knowledge at the Crossroads: Some alternative futures of hypertext learning environments. Educational Theory, 46, 23–50. Cristia, J., Ibarrarán, P., Cueto, S., Santiago, A. & Severín, E. (2012). Technology and child development: Evidence from the One Laptop per Child program. Inter-American Development Bank. http://dx.doi.org/10.18235/0012202 Cuban, L. (2001). Oversold and Underused: Computers in Classrooms. Harvard University Press. Cube, F. v. (1965). Kybernetische Grundlagen des Lernens und Lehrens. Stuttgart. Di Geronimo, L., Braz, L., Fregnan, E., Palomba, F. & Bacchelli, A. (2020). UI Dark Patterns and Where to Find Them: A Study on Mobile Applications and User Perception. Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems. https://doi.org/10.1145/3313831.3376600 Eynon, R. (2018). Into the mainstream: where next for Critical Ed Tech research? Learning, Media and Technology, 43(3), 217–218. https://doi.org/10.1080/17439884.2018.1506976 Ferster, B. (2014). Teaching machines: Learning from the intersection of education and technology. JHU Press. Floridi, L. (2015). The Onlife Manifesto: Being human in a hyperconnected era. SpringerOpen. https://doi.org/10.1007/978-3-319-04093-6 Gee, J. P. (2005). Learning by Design: Good Video Games as Learning Machines. E-Learning and Digital Media, 2(1), 5–16. https://doi.org/10.2304/elea.2005.2.1.5 Interactive Fiction Technology Foundation. (2022). Twine 2.5.1. https://twinery.org/ Hilgard, E. R. & Bower, G. H. (1977). Theories of Learning (4th ed.). Prentice-Hall. Hof, Barbara (2021). The turtle and the mouse: how constructivist learning theory shaped artificial intelligence and educational technology in the 1960s. History of Education, 50(1), 93–111. https://doi.org/10.1080/0046760X.2020.1826053 Lanier, J. (2002). The complexity ceiling. In J. Brockman (Ed.), The Next Fifty Years: Science in the First Half of the Twenty-first Century. Vintage Press. Laurel, B. (2001). Utopian Entrepreneur. MIT Press. Luguri, J. & Strahilevitz, L. J. (2021). Shining a light on dark patterns. Journal of Legal Analysis, 13(1), 43–109. Mangen, A. & van der Weel, A. (2017). Why don’t we read hypertext novels? Convergence: The International Journal of Research into New Media Technologies, 23(2), 166–181. https://doi.org/10.1177/1354856515586042 Manolev, J., Sullivan, A. & Slee, R. (2019). The datafication of discipline: ClassDojo, surveillance and a performative classroom culture. Learning, Media and Technology, 44(1), 36–51. https://doi.org/10.1080/17439884.2018.1558237 Medienpädagogischer Forschungsverbund Südwest. (2023). KIM-Studie 2022 Kindheit, Internet, Medien (technical report). Medienpädagogischer Forschungsverbund Südwest. https://www.mpfs.de/studien/kim-studie/2022/ Morozov, E. (2013). To save everything, click here: The folly of technological solutionism. Public Affairs. Murray, J. H. (2011). Inventing the medium: principles of interaction design as a cultural practice. MIT Press. Nelson, T. H. (1974). Computer lib: you can and must understand computers now (1st ed.). Nelson. Nelson, T. H. (2003). From Computer Lib/Dream Machines, 1974. In N. Montfort & N. Wardrip-Fruin (Eds.), The New Media Reader. MIT Press. Niegemann, H. M. (2008). Die Suche nach der Lehrmaschine: Von der Buchstabiermaschine über den Programmierten Unterricht zum E-Learning. In Kompendium multimediales Lernen (pp. 3–16). Springer. Papert, S. (1985). Kinder, Computer und Neues Lernen. Springer Basel AG. Papert, S. (1993). The children’s machine: Rethinking school in the age of the computer. Basic Books. Perrotta, C., Gulson, K. N., Williamson, B. & Witzenberger, K. (2020). Automation, APIs and the distributed labour of platform pedagogies in Google Classroom. Critical Studies in Education, 1–17. https://doi.org/10.1080/17508487.2020.1855597 Postman, N. (1995). The End of Education: Redefining the Value of School. Vintage. Pressey, S. L. (1927). A machine for automatic teaching of drill material. School & Society, (25), 549–552. Pressey, S. L. (1963). Teaching machine (and learning theory) crisis. Journal of Applied Psychology, 47(1), 1–6. https://doi.org/10.1037/h0047740 Ren’Py (7) [computer program]. (2024). https://www.renpy.org/ Resnick, M. & Robinson, K. (2017). Lifelong kindergarten: Cultivating creativity through projects, passion, peers, and play. MIT Press. Rixen, J. O., Meinhardt, L.-M., Glöckler, M., Ziegenbein, M.-L., Schlothauer, A., Colley, M., Rukzio, E. & Gugenheimer, J. (2023). The Loop and Reasons to Break It: Investigating Infinite Scrolling Behaviour in Social Media Applications and Reasons to Stop. Proceedings of the ACM on Human-Computer Interaction, 7(MHCI), 1–22. https://doi.org/10.1145/3604275 Rohs, M. & Ganz, M. (2015). MOOCs and the claim of education for all: A disillusion by empirical data. International Review of Research in Open and Distributed Learning, 16(6), 1–19. Selwyn, N. (2023). Lessons to Be Learnt? Education, Technological Solutionism and Sustainable Development. In Sætra (Ed.), Technology and Sustainable Development. Routledge. Sims, C. (2017). Disruptive fixation: School reform and the pitfalls of techno-idealism. Princeton University Press. Skinner, B. F. (1958). Teaching Machines: From the experimental study of learning come devices which arrange optimal conditions for self-instruction. Science, 128(3330), 969–977. Skinner, B. (1968). The Technology of Teaching. Appleton. Solocode (2024). Anton app [online learning software]. https://anton.app van Dijck, J., Poell, T. & de Waal, M. (2018). The Platform Society. Oxford University Press. https://doi.org/10.1093/oso/9780190889760.001.0001 Watters, A. (2021). Teaching Machines. MIT Press. Weller, M. (2020). 25 Years of Ed Tech. AU Press.

Michael Schlauch, Dr. studied industrial engineering at TU Dresden and sociology at the University of Trento. He completed his doctorate in educational sciences at the Free University of Bozen-Bolzano. He is currently a research associate at the Institute of Educational Sciences at Humboldt University of Berlin in the area of school pedagogy. His work focuses on general educational science, media pedagogy, design-oriented research, and inquiry-based learning in teacher education. Email: michael.schlauch@hu-berlin.de