In one of my lectures, “Digital Communication,” I always give examples of the three farsighted contributors to technological development way behind the actual innovations that took place. Notoriously, Jules Verne and Alan Turing were the pioneers of this genre of literature and science; Octave Uzanne, Vannevar Bush, and Cahit Arf were more accurate in what they had thought as will becoming reality. To me, technology has its own consciousness embedded in homofaber, free from all social, economic, and political occurrences. Technology is the determinant of humanity. Uzanne, Bush, and Arf are the visioneers.
(1)
In 1894, A French Writer Predicted the End of Books & the Rise of Portable Audiobooks and Podcasts Books, History, Technology | January 10th, 2025
The end of the nineteenth century is still widely referred to as the fin de siècle, a French term that evokes great, looming cultural, social, and technological changes. According to at least one French mind active at the time, among those changes would be a fin des livres as humanity then knew them. “I do not believe (and the progress of electricity and modern mechanism forbids me to believe) that Gutenberg’s invention can do otherwise than sooner or later fall into desuetude,” says the character at the center of the 1894 story “The End of Books.” “Printing, which since 1436 has reigned despotically over the mind of man, is, in my opinion, threatened with death by the various devices for registering sound which have lately been invented, and which little by little will go on to perfection.”
First published in an issue of Scribner’s Magazine (viewable at the Internet Archive or this web page), “The End of Books” relates a conversation among a group of men belonging to various disciplines, all of them fired up to speculate on the future after hearing it proclaimed at London’s Royal Institute that the end of the world was “mathematically certain to occur in precisely ten million years.” The participant foretelling the end of books is, somewhat ironically, called the Bibliophile; but then, the story’s author Octave Uzanne was famous for just such enthusiasms himself. Believing that “the success of everything which will favor and encourage the indolence and selfishness of men,” the Bibliophile asserts that sound recording will put an end to print just as “the elevator has done away with the toilsome climbing of stairs.”
These 130 or so years later, anyone who’s been to Paris knows that the elevator has yet to finish that job, but much of what the Bibliophile predicts has indeed come true in the form of audiobooks. “Certain Narrators will be sought out for their fine address, their contagious sympathy, their thrilling warmth, and the perfect accuracy, the fine punctuation of their voice,” he says. “Authors who are not sensitive to vocal harmonies, or who lack the flexibility of voice necessary to a fine utterance, will avail themselves of the services of hired actors or singers to warehouse their work in the accommodating cylinder.” We may no longer use cylinders, but Uzanne’s description of a “pocket apparatus” that can be “kept in a simple opera-glass case” will surely remind us of the Walkman, the iPod, or any other portable audio device we’ve used.
All this should also bring to mind another twenty-first century phenomenon: podcasts. “At home, walking, sightseeing,” says the Bibliophile, “fortunate hearers will experience the ineffable delight of reconciling hygiene with instruction; of nourishing their minds while exercising their muscles.” This will also transform journalism, for “in all newspaper offices there will be Speaking Halls where the editors will record in a clear voice the news received by telephonic despatch.” But how to satisfy man’s addiction to the image, well in evidence even then? “Upon large white screens in our own homes,” a “kinetograph” (which we today would call a television) will project scenes fictional and factual involving “famous men, criminals, beautiful women. It will not be art, it is true, but at least it will be life.” Yet however striking his prescience in other respects, the Bibliophile didn’t know – though Uzanne may have — that books would persist through it all.
(2)
Vannevar Bush’s Visionary Essay: “As We May Think”
The essay, written in 1945 by Vannevar Bush, was published in the journal Atlantic Monthly. It introduced science to a new way of thinking, making it clear that for years, all inventions had only taken the extent of humankind’s physical powers into consideration, rather than the power of their minds.
Vannevar Bush was an American engineer, born on the 11th of March 1890. He was the Vice President of MIT, and held many patents of analogue computers, as well as being the inventor of the differential analyzer.
The essay “As We May Think”
Written in 1945, the essay titled “As We May Think” was published in the journal Atlantic Monthly in the July, as well as a condensed version being republished in the September of the same year, practically straddling the Hiroshima and Nagasaki bombings.
Cover page of the essay “As We May Think”
Bush expressed his concern about scientific efforts moving in the direction of mass destruction, rather than understanding. He explained the desire and the need of a type of “collective memory machine”, with an enormous potential to make knowledge more accessible for all, as well as answering the question:
“How can technology contribute to the wellbeing of humanity?”
The essay bases its reasoning on a premise: Human knowledge is a set of connected knowledge, and has a universal dimension that cannot be limited to the life of an individual.
Knowledge is the result of a continuous process, built due to a fruitful collaboration among scientists, and includes the wealth of all human knowledge, where access to scientific information is a necessary condition for the growth of mankind.
In fact, the essay did not just pose a mere technical question, the subject was mostly a strong philosophical and political reflection upon how knowledge is produced and communicated.
The Concept of Memex
Bush, in the sixth section of the eight-part essay, presents the “memex”, a sort of extension of human memory in the form of a mechanical desk containing a microfilm archive inside. This would have been the storage method for books, registers, and documents, in order to subsequently be able to reproduce them and associate them with each other.
The memex takes its name from the words “memory extender”, and its purpose was considered essential by Bush, saying that
“The human mind […] operates by association. With one item in its grasp, it snaps instantly to the next that is suggested by the association of thoughts, in accordance with some intricate web of trails carried by the cells of the brain.”
In fact, the memex helps a researcher to remember things more quickly, in addition to being able to quickly see their archives of information by typing the code of the register and the book, in a way that it can be called and immediately viewed on one of its screens.
The interface was simple, consisting of buttons and simple levers.
Pushing the lever to the right moves to the next page of a document, and pushing it even further to the right causes the memex to scroll 10 pages at a time. Logically, moving the lever to the left has the same functionality but in the opposite direction, while a special button allows a complete shift, repositioning to the first page of the text.
Moreover, the user of the memex would have been able to have opened several books at the same time, perhaps on the same subject, creating a path and various links between the information. This would in fact define a new book, as well as including note-taking thanks to dry photography technology.
Bush, with this essay, strongly anticipates what we will see a few decades later on the world wide web. He also revolutionises the concept of the workspace, introducing page-by-page navigation via links and connections.
This essay inspired the subsequent innovation of the “oN-line-System” in a fundamental way, the operating system created by Douglas Engelbart, presented in the 1968 “Mother of all demos” which we discussed in a previous article on Red Hot Cyber.
This article is written by Author: Massimiliano Brolli Original Publication Date: 15/11/2021 Translator: Tara Lie: Cyber Security analyst from Perth, Western Australia, focused on governance, risk quantification and compliance. Graduate of cyber security and pure mathematics, with a second-major in Italian Studies. Tara has earned a Master's degree in Cyber Security, and has a great passion for quantum-preparedness.
(3) Cahit Arf
Can machines think and how can they think? [1]
Cahit Arf 1958/59, Turkish mathematician
Can machines think and how? If someone told about that idea approximately 60 years ago, most probably only a few people would believe that, and it did. But today we all know the fact that machines can think. Although it might be a frightening thought for some people, at the same time for some it is the salvation, “learning” of machines continue to grow without slowing down day by day, even while I am writing this article.
One of the first papers, which declares the fact that machines can think, is written by Alan. M. Turing, who is the encoder of “Enigma” [2]. For the first time, he proposed that the so-called two unconnected words such as “machine” and “think”, which might look like they are far away from each other, can have meaning together and he told about “the imitation game”. Basically, a machine can take the place of a person to send a message, and someone should decide whether it is a machine or a person, somehow. Even today it is hard to imagine those mindblowing concepts, they proposed and showed them 60–70 years ago.
Introduction to Machine Learning
Now we know that machines can think. The second question was “how”. So how can machines think? In this regard, the “machine learning” concept is introduced. Machine learning is a subfield of artificial intelligence (AI) and it is a concept that a machine can decide it’s on behalf by means of some steps. In order to apply Machine Learning concept, there are three main points:
· Data
· Pattern
· No mathematical formulation
Data:
This is the main part of the machine learning. How do we learn in real life? We experience, right? Actually, we collect data. With a simple example, we might have learned that we should not wear a winter jacket in the summer by experiencing or somebody might have given us that knowledge. The same concept is valid for machines as well. If you don’t have data, then machine learning can not be applied. From simple to complicated examples, no machine learning algorithm can be applied if there is no data. Actually, this is one of the reasons why data science gained surplus importance particularly in the last decade, since it is the main and fundamental source of machine learning.
No mathematical formulation:
Let’s start with such an idea. Let’s say, I ask a question and I want a solution not only for one case but in general. Basically, what should be done is to find a function or a formulation that describes all the cases of the question. It is possible to find such a solution, right? In machine learning, one can not find such a function. This is the concept of the “target function” in machine learning. The target function is on the top of everything and it is unknown but the aim is to find the best function which is the closest to “unknow target function”. If there is a mathematical formulation of the problem at the very beginning, it does not make sense to apply machine learning algorithms. It might work, which means a solution can be found, but it is not going to worth it. As a result, the main aim of machine learning is to find the best approximation (hypothesis) since the target function will be always an unknown.
Pattern:
The last point is that there should be a pattern in the data. What does it mean? If the data is completely random, machine learning can not learn “efficiently”. Just imagine, you are trying to teach a child how to speak German by talking in all languages randomly. The child can speak, this is the learning part, but she/he can not speak German in a correct way. Thus, in the data, there should be a pattern so that the machine can learn efficiently.
Well, we told about three main parts of machine learning. Which one is the most important? Is there any guess? You guessed it, right. The answer is data. You can learn without a pattern and if there is a mathematical formulation of the problem. It might not be the best solution, but still you can learn. But without data, machine learning can not be performed.
Main Types of Machine Learning
Let’s explain machine learning types with an easy example in order to understand the concept.
Assume that you have a business that makes profits by selling cars in a small town close to the sea. Every morning before coming to work, you take a walk on the seashore and drink some coffee. Then you watch… Stop stop! We are losing our path. Let’s say you are selling cars. You sold many cars beforehand and keep the information of your customers, such as their age, salary, job, where they live, how often they visit and buy cars and maybe they did not buy any car and so on. In this way, you have a database consisting of the customers, which labels them whether she/he bought a car or not. And a new customer already came! So the question is that the new customer is going to buy a new car or not? We might guess whether the new customer will buy a new car or not with a probability by applying a machine learning model since we have the data of the previous customers. If the features of the new customer are mostly matching with the public profile of the customers who bought cars, then she/he most probably is going to buy the car, in vice versa, she/he is not. This is called “supervised learning”. You train your model according to previous experiences and labels. Therefore, the machine learning model knows the profile of the people who are prone to buy a new car or not.
Let’s talk about another scenario. It is the same business. However, in this case, you just started your business, so you don’t have the database in such a way that you don’t have the information about the customers who bought cars or not. You only have their features, such as their age, salary, job, where they live and so on, but you don’t know whether they buy any car or not. What can you do at this point? You can cluster the customers according to their features to find hidden patterns or groupings in the data. This is called “unsupervised learning”. You don’t have labels so that they can supervise you, but you can find the pattern by clustering the data or summarizing the distribution of the data.
Well, we know the difference between “supervised ” and “unsupervised” learning. The final main machine learning type is “reinforcement learning”. Let’s try to explain reinforcement learning with the same example. Similarly, you don’t know whether the customers bought cars or not, but their features are known, like unsupervised learning. In this case, you send a “salesperson”, called “agent” in the machine learning terminology, to find the potential people who can buy a new car. The agent picks up a customer that might buy a new car and come to you. You talk to the customer and say to the agent: “yes it is the right customer or no it is the wrong customer”. Basically, you reward or punish the agent so that the agent can learn to choose the right customer. This is the concept of “reinforcement learning”. The agent will learn according to rewarding or punishing.
There are other machine learning types as well, such as semi-supervised, inductive learning, ensemble learning, and so on. But only there main types of machine learning models are investigated in this article.
Conclusion
As a conclusion, over the past years, humankind gained a powerful momentum to go further regarding artificial intelligence and machine learning. Some people think these trendy topics will bring the end, while others think they are salvation. In my opinion, there is a thin line between end and salvation. Therefore we have to be ready for the new era, not only the machines we must “learn“, too.
Reference 1: https://www.mbkaya.com/makine-dusunebilir-mi-cahit-arf/
Reference 2: http://csee.umbc.edu/courses/471/papers/turing.pdf
Reference 3: https://mc.ai/supervised-versus-unsupervised-learning-2/
Can a Machine Think?
Alan Turing, who first came up with the ideas of artificial intelligence and thinking machines, and Cahit Arf, who talked about these machines for the first time in Turkey’s history, analyze thinking machines.
Alan M. Turing proposes in his paper Computing Machinery and Intelligence to consider the question “Can machines think?”. In his approach, he replaces the question with another which is “expressed in relatively unambiguous words” to avoid the vagueness of the ordinary concepts of “machine” and “think”. He attempts to describe the new form of the problem in terms of a game (“imitation game”). Nowadays, the game is known as the Turing test.
According to Turing, the original question “Can machine think?” can be replaced by a question, which is similar to the following: Is the interrogator able to distinguish between the machine and the person? In explicit terms, the criterion determining that a machine can think is the interrogator’s inability to tell the difference.
Arf presented the concrete indicator of thought as different reactions to different effects and emphasized that people react with different words to different words spoken to them or to different effects they are exposed to and that these reactions should be taken as proof of their thinking. Thus, Arf establishes a link between thought and behavior by referring to the effect and reaction relationship and deals with this link depending on the language phenomenon.
The first thing to note is that Arf does not touch upon the subject of “learning” for thinking machines. According to Turing, on the other hand, the main point is what he calls learning, and according to Turing, it is necessary to turn to this learning business for thinking machines to be possible. One similar way of thinking, both Turing and Arf, draws attention to and emphasizes the importance of machine memories.
Another similar inquiry is to point out that machines are designed to solve a specific problem. While Arf explains this subject by making an analogy between the human brain and the machine; Turing analyzes this issue in the context of Lovelace’s thesis and does not accept Lovelace’s idea, rejecting it with the seed/plant analogy.
By looking at the correspondence between the machine that does not see each other and the human, what Turing means by thinking machine emphasizes that if the human cannot understand that the other person is a machine, that machine will be a thinking machine, Arf does not put forward such a condition. Since it is the uncertainty principle, it says that it is possible if these subatomic parts direct machines that think like this.
Finally, Turing has full faith in thinking machines and has full hope that these machines will emerge by the end of his century. Arf, on the other hand, stands at a more pessimistic point and thinks that it may never be done for many years.
References
- Turing, Alan M. (1950). Computing Machinery and Intelligence, Mind, 59/433–460.
- Arf, C. (1959).Makine Düşünebilir Mi ve Nasıl Düşünebilir?, Atatürk Üniversitesi - Üniversite Çalışmalarını Muhite Yayma ve Halk Eğitimi Yayınları Konferanslar Serisi No: 1, Erzurum, s. 91–103
- Sarı, F. (2021). Cahit Arf’in “Makine Düşünebilir mi ve Nasıl Düşünebilir?” Adlı Makalesi Üzerine Bir Çalışma . TRT Akademi , 6 (13) , 812–833.
