Evolution of Computing

James R. Larus

Abstract Computers and computing emerged within a lifetime and completely

changed our world. Although their history is brief, the change they precipitated
has been rapid and constant. Today’s world would be unimaginably different
without these machines. Not necessarily worse, but certainly slower, static, discon-
nected, and poorer. One has to look back to the steam engine in the nineteenth
century or electricity in the early twentieth century to find technologies with similar
rapid and far-reaching effects. This chapter briefly describes the evolution of com-
puting and highlights how its growth is closely tied to concerns of digital humanism.

1 Introduction

Electronic digital computers have existed for only 75 years. Computer science—or
informatics, if you prefer—is roughly a decade older. Computer science is the
expanding discipline of understanding, developing, and applying computers and
computation. Its intellectual roots were planted in the 1930s, but it only emerged in
the 1940s when commercial computers became available.
Today’s world would be unimaginably different without these machines. Not
necessarily worse (computers emerged during but played little role in the world’s
deadliest conflict), but certainly slower, static, disconnected, and poorer. Over three-
quarters of a century, computers went from rare, expensive machines used only by
wealthy businesses and governments to devices that most people on earth could not
live without. The technical details of this revolution are a fascinating story of
millions of peoples’ efforts, but equally compelling are the connections between
technology and society.
Like the emergence of a new animal or virus, the growth of computing has serious
and far-reaching consequences on its environment—the focus of this book. In seven
decades, computing completely changed the human environment—business,

J. R. Larus (✉)
School of Computer and Communication Sciences, EPFL, Lausanne, Switzerland
e-mail: james.larus@epfl.ch

© The Author(s) 2024 31

H. Werthner et al. (eds.), Introduction to Digital Humanism,
https://doi.org/10.1007/978-3-031-45304-5_3
32 J. R. Larus

finance, social relations, government, and society, to name a few—through its

seminal advances such as personal computers, the Internet, the World Wide Web,
mobile computing, machine learning, and artificial intelligence. One has to look
back to the steam engine in the nineteenth century or electricity in the early twentieth
century to find technologies with similar rapid and far-reaching effects.
This chapter offers a brief overview of the evolution of computing and its
connection to the concerns of digital humanism. The velocity and broad impact of
computing’s emergence discussed in this chapter partly explain why the humanism
implications discussed in the rest of this book are among society’s prominent and
concerning challenges.

2 Prehistory

In most people’s opinion, computer science started in 1936 when Alan Turing, a
student at Cambridge, published his paper “On Computable Numbers, with an
Application to the Entscheidungsproblem” (Turing, 1937). This paper settled a
fundamental open question in mathematics by showing that a general technique
does not exist to decide whether a theorem is true or false.
More significantly for this history, Turing’s paper introduced the concept of a
universal computer (the Turing Machine) and postulated that it could execute any
algorithm (a procedure precisely described by a series of explicit actions). The idea
of a computing machine—a device capable of performing a computation—had
several predecessors. Turing’s innovation was to treat the instructions controlling
the computer (its program) as data, thereby creating the infinitely malleable device
known as a stored program computer. This innovation made computers into univer-
sal computing devices, capable of executing any computation (within the limits of
their resources). Even today, no other field of human invention has created a single
device capable of doing everything. Before computers, humans were the sole
universal “machines” capable of being taught new activities.
In addition, by making computer programs into explicit entities and formally
describing their semantics, Turing’s paper also created the rich fields of program and
algorithm analysis, the techniques for reasoning about computations’ characteristics,
which underlie much of computer science.
A Turing Machine, however, is a mathematical abstraction, not a practical
computer. The first electronic computers were built less than a decade later, during
World War II, to solve pressing problems of computing artillery tables and breaking
codes. Not surprisingly, Turing was central to the British effort at Bletchley Park to
break the German Enigma codes. These early computers were electronic, not
mechanical like their immediate predecessors, but they did not follow Turing’s
path and treat programs as data; rather they were programmed by rewiring their
circuits.
However, soon after the war, the Hungarian-American mathematician John von
Neuman, building on many people’s works, wrote a paper unifying Turing’s insight
Evolution of Computing 33

Fig. 1 ENIAC (1947). (Public domain) In Wikipedia. https://en.wikipedia.org/wiki/ENIAC

with practical engineering. It described an architecture for stored-program com-

puters, which laid the computer industry’s foundation. This so-called von Neuman
architecture still is the blueprint for today’s computers. Figure 1 shows a picture of
ENIAC, the first general-purpose electronic computer.

3 Computers as Calculators

The first applications of computers were as calculators, both for the government and
industry. The early computers were expensive, slow, and limited machines. For
example, IBM rented its 701 computers for $15,000/month for an 8-h work day
(in 2023 terms, $169,000) (na, 2023a). This computer could perform approximately
16,000 additions per second and hold 82,000 digits in its memory (na, 2003). While
the 701’s performance was unimaginably slower than today’s computers, the
701 was far faster and more reliable than the alternative, a room full of clerks with
mechanical calculators.
The challenge of building the first computers and convincing businesses to buy
them meant that the computer industry started slowly. Still, as we will see, progress
accelerated geometrically. The societal impact of early computers was also initially
small, except perhaps to diminish the job market for “calculators,” typically women
who performed scientific calculations by hand or mechanical adding machines, and
clerks with mechanical calculators.
At the same time, there was considerable intellectual excitement about the
potential of these “thinking machines.” In his third seminal contribution, Alan
34 J. R. Larus

Turing posed the question of whether a machine could “think” with his famous
Turing Test, which stipulated that a machine could be considered to share this
attribute of human intelligence when people could not distinguish whether they
were conversing with a machine or another human (Turing, 1950). Seventy years
later, with the advent of ChatGPT, Turing’s formulation is still insightful and now
increasingly relevant.

4 Computers and Communications

Computers would only be slightly more exciting than today’s calculators if they
were only capable of mathematical calculations. But it quickly became apparent that
computers can exchange information and coordinate with other computers, allowing
them (and people) to communicate and collaborate as well as compute. The
far-reaching consequences of computing, the focus of this book, are due as much
to computers’ ability to communicate as to compute, although the latter attribute is
more closely identified with the field.
Among the most ambitious early applications of computers were collections of
devices and computers linked through the telephone system. SAGE, deployed in
1957, was a computer-controlled early warning system for missile attacks on the
United States (na, 2023b). In 1960, American Airlines deployed Sabre, the first
online reservation and ticketing system, which accepted requests and printed tickets
on terminals worldwide (Campbell-Kelly, Martin, 2004). The significance of both
systems went far beyond their engineering and technical challenges. Both directly
linked the real world—World War II and commercial transactions—to computers
without significant human intermediation. People did not come to computers;
computers came to people. Starting with systems like these, these machines have
increasingly intruded into everyday life.
Businesses using computers, e.g., American Airlines, quickly accumulated large
quantities of data about their finances, operations, and customers. Their need to
efficiently store and index this information led to the development of database
systems and mass storage devices such as disk drives. Around this time, the
implications of computers on people’s privacy emerged as a general concern as
the capacity of computers to collect and retrieve information rapidly increased. At
that time, perhaps because of its traditional role, attention was focused more on
government information collection than private industry (na, 2973).
Another fundamental innovation of that period was the ARPANET, the Internet’s
direct intellectual and practical predecessor. The US Department of Defense created
the ARPANET in the late 1960 and early 1970 as a communication system that
could survive a nuclear attack on the USA (Waldrop, 2001). The ARPANET’s
fundamental technical innovation was packet switching, which splits a message
between two computers into smaller pieces that could be routed independently
along multiple paths and resent if they did not reach their destination. Before,
communication relied on a direct connection between computers (think of a
Evolution of Computing 35

telephone wire, the technology used at the time). These connections, called circuits,
could not have grown to accommodate a worldwide network like today’s Internet.
Moreover, the engineering of the ARPANET was extraordinary. The network grew
from a few hundred computers in the 1970s to tens of billions of computers today in
a smooth evolution that maintained its overall structure and many of its communi-
cation protocols, even as new technologies, such as fiber optics and mobile phones,
emerged to support or use the Internet (Mccauley et al., 2023).

5 Computing as a Science

In the 1960s and 1970s, the theory underlying computer science emerged as a
discipline on its own that offered an increasingly nuanced perspective on what is
practically computable. Three decades earlier, Turing hypothesized that stored
program computers were universal computing devices capable of executing any
algorithm—though not solving any problem, as he proved that no algorithm could
decide whether any algorithm would terminate. Turing’s research ignored the
running time of a computation (its cost), which held no relevance to his impossibility
results but was of first-order importance to solving real-world problems.
The study of these costs, the field of computational complexity, started in the
1960s to analyze the running time of algorithms to find more efficient solutions to
problems. It quickly became obvious that many fundamental problems, for example,
sorting a list of numbers, had many possible algorithms, some much quicker than
others.
Theoreticians also realized that the problems themselves could be classified by
the running cost of their best possible solution. Many problems were practically
solvable by algorithms whose running time grew slowly with increasingly large
amounts of data. Other problems had no algorithm other than exploring an expo-
nential number of possible answers, and so could only be precisely solved for small
instances. The first group of problems was called P (for polynomial time) and the
second NP (nondeterministic polynomial time). For 50 years, whether P = NP has
been a fundamental unanswered question in computer science (Fortnow, 2021).
Although its outcome is still unknown, remarkable progress has been made in
developing efficient algorithms for many problems in P and efficient, approximate
algorithms for problems in NP.
Moreover, computer science’s approach of considering computation as a formal
and analyzable process influenced other fields of education and science through a
movement called “computation thinking” (Wing, 2006). For centuries, scientific and
technical accomplishments (and ordinary life—think food recipes) offered informal,
natural language descriptions of how to accomplish a task. Computer science
brought rigor and formalism to describing solutions as algorithms. Moreover, it
recognized that not all solutions are equally good. Analyzing algorithms to under-
stand their inherent costs is a major intellectual step forward with broad applicability
beyond computers.
36 J. R. Larus

6 Hardware “Laws”

Computer science was extremely fortunate to ride on the back of an extraordinary

and unprecedented improvement in silicon semiconductors, the underlying technol-
ogy used to construct computers. The earliest computers were built from mechanical
relays, which could switch on or off roughly 20 times per second. They were quickly
succeeded by vacuum tubes, which could switch millions of times per second, but
were large, hot, and unreliable. In the 1960s, transistors replaced tubes with much
smaller, more reliable switches. More importantly, many transistors could be fabri-
cated and wired together on a small piece of silicon called a “chip,” which offered
compelling size, speed, and cost advantages. In 1965, Gordon Moore noted that the
number of transistors on a chip doubled every year, an observation that came to be
called “Moore’s law.” This geometric increase in capacity has continued for over
four decades, albeit at a slower pace. A decade after Moore, Robert Dennard
published rules for IC design, which quantified how the smaller, denser transistors
resulting from Moore’s Law could run faster without consuming more power.
Figure 2 illustrates this remarkable progress. Moore’s law and Dennard scaling
led to three decades of computers whose running speed doubled every other year, a
remarkable period of innovation that ended around 2005, when electrical consider-
ations made it impossible to continue running computers faster, even though the
number of transistors on a chip continued to double. From the 1970s to the early
2000s, computers dropped rapidly in cost at the same time as their performance
increased, which hastened the birth of the software industry (discussed below) and
made possible increasingly ambitious uses of computers.

Fig. 2 Moore’s law and Dennard scaling. The number of transistors on a chip has doubled every
other year for 50 years. For the first half of this period, each generation of chips also doubled in
speed. That improvement ended around 2005. From Karl Rupp, CC BY 4.0
Evolution of Computing 37

Another important observation, called Kryder’s law, was that the amount of data
that could be stored in a square centimeter also grew geometrically at a faster rate
than Moore’s law. This progress has also slowed as technology approaches physical
limits. Still, storage cost fell from $82 million/Gigabyte (billion bytes) for the first
disk drive in 1957 to 2 cents in 2018 (both in 2018 prices). This amazing improve-
ment not only made richer and more voluminous media such as photos and video
affordable, but it also made possible the collection and retention of unprecedented
amounts of data on individuals.

7 Personal Computers

In the mid-to-late 1970s, the increasing density of integrated circuits made it possible
to put a “computer on a chip” by fabricating the entire processing component on a
single piece of silicon (memory and connections to the outside world required many
other chips). This accomplishment rapidly changed the computer from an expensive,
difficult-to-construct piece of business machinery into a small, inexpensive com-
modity that entrepreneurs could exploit to build innovative products. These com-
puters, named microprocessors, initially replaced inflexible mechanical or electric
mechanisms in many machines. As programmable controllers, computers were
capable of nuanced responses and often were less expensive than the mechanisms
they replaced.
More significantly, microprocessors made it economically practical to build a
personal computer that was small and inexpensive enough that an employee or
student could have one use to write and edit documents, exchange messages, run
line-of-business software, play games, and do countless other activities.
With the rapidly increasing number of computers, software became a profitable,
independent business, surpassing computer hardware in creativity and innovation.
Before the microprocessor, software was the less profitable, weak sibling of hard-
ware, which computer companies viewed as their product and revenue source. The
dominant computer company IBM gave away software with its computers until the
US government’s antitrust lawsuit in the early 1970s forced it to “unbundle” its
software from hardware. Bill Gates, a cofounder of Microsoft, was among the
earliest to realize that commodity microprocessors dramatically shifted computing’s
value from the computers to the software that accomplished tasks. IBM accelerated
this shift by building its iconic PC using commodity components (a processor from
Intel and an operating system from Microsoft) and not preventing other companies
from building “IBM-compatible” computers. Many companies sprung up to build
PCs, providing consumer choice and driving down prices, which benefited the
emerging software industry.
Moreover, the widespread adoption of powerful personal computers (doubling in
performance every 2 years) created a technically literate segment of the population
and laid the foundation for the next major turning point in technology, the Internet.
38 J. R. Larus

8 Natural Interfaces

Interaction with the early computers was textual. A program, the instructions
directing a computer’s operation, was written in a programming language, a highly
restricted and regularized subset of English, and a computer was directed to run it
using textual commands. Though small and precise, most people found these
languages difficult to understand, limiting early machines’ use. In the late 1960
and 1970s, graphical user interfaces (GUIs) were initially developed, most notably at
Xerox PARC (Michael A. Hiltzik, 1999). They became widespread with the intro-
duction of the Apple Macintosh computer in the early 1980s. These interfaces
provided pictural metaphor-oriented interfaces directly manipulated through a
mouse. This user interface change made computers accessible and useful to many
more people.
The graph aspect of GUIs enabled computers to display and manipulate images,
though initially, software treated them as collections of pixels and could not discern
or recognize their content. This capability only came later, with the advent of
powerful machine-learning techniques that enabled computers to recognize entities
in images. In addition, the early computers were severely constrained in computing
power and storage capacity, which limited the use of images and video, which is far
larger than a single image.
Computers also adopted other human mechanisms. Voice recognition and speech
generation are long-established techniques for interaction. Recently, machine learn-
ing has greatly improved the generality and accuracy of human-like speech and
dialog, so it is not unusual to command a smartphone or other device by speaking
to it.
Most computers do not exist as autonomous, self-contained entities, like PCs or
smartphones with their own user interface. They are instead incorporated into
another device and interact through its features and functionality. Mark Weiser
called this “ubiquitous computing” (Weiser, 1991), where computing fades into
the background, so no one is aware of its presence. Many of these computers,
however, are accessible through the Internet, raising vast maintenance, security,
and privacy challenges.

9 The Internet

The Internet started as a US government research project and infrastructure in the

1970s. Access was initially limited to the military, universities, and a few
government-related businesses. In the early 1990s, two important events occurred.
The US government agency managing the public Internet, the National Science
Foundation (NSF), decided it was time to transition from a government-led project
into a commercial product. In a little-heralded but enormously successful effort, it
Evolution of Computing 39

turned the Internet over to the technical community that built it and the private
companies that operate the individual networks that comprise today’s Internet.
The other crucial change was the emergence of the World Wide Web (the “Web”)
as the Internet’s “killer app,” which caused it to gain vast public interest and financial
investment. While working at CERN, a physics research lab in Switzerland, Tim
Berners-Lee developed a networked hypertext system he optimistically called the
“World Wide Web (WWW).” CERN released his design and software to the public
in 1991. A few years later, the University of Illinois’s Mosaic browser made
Berners-Lee’s invention easier to use and more visually appealing on many types
of computers. The academic community, already familiar with the Internet, rapidly
jumped on the Web. Then, remarkably, both inventions made a rare leap into the
public eye and widespread adoption. In a remarkably short time, businesses started
creating websites, and the general population started to buy personal computers to
gain access to “cyberspace.”
Other chapters of this book discuss a remarkable spectrum of societal and
personal changes in the past three decades. Underlying all of them are the Internet
and the Web, which made it possible to find information, conduct commerce, and
communicate everywhere at nearly zero cost. Before these inventions, there were
two ways to communicate.
First, you could speak to another person. If the person was distant, you used a
telephone or radio. However, both alternatives were expensive, particularly as
distance increased, because the technical structure of telephone systems allocated a
resource (called a circuit) to each communication and charged money to use it
throughout the conversation. By contrast, the Internet used packet switching,
which only consumed resources when data was transferred, dramatically lowering
costs. In fact, users pay a flat rate in most parts of the Internet, independent of their
usage, because finer-grained billing is neither necessary nor practical. In addition, for
historical reasons, telephone companies were regulated as “natural” monopolies,
which allowed them to keep their prices high. The Internet, in reaction, sought
multiple connections between parties and resisted centralization and
monopolization.
The second alternative, of course, was to engrave, write, or print a message on a
stone tablet or piece of paper and physically convey the object to the recipient,
incurring substantial costs for the materials, printing, and delivery. Moreover, paper
has a low information density, requiring considerable volume to store large amounts
of data. In addition, finding information stored on paper, even if well organized,
takes time and physical effort.
Computing and the Internet completely changed all of this. A message, even a
large one, can be delivered nearly instantaneously (at no cost). And data, stored
electronically at rapidly decreasing cost, can be quickly retrieved. This is the
dematerialization of information, which no longer needs a physical presence to be
saved, shared, and used. This change, as much as any, is behind the “creative
destruction” of existing industries such as newspapers, magazines, classified adver-
tising, postal mail, and others that conveyed information in a tangible, physical form.
40 J. R. Larus

Another fundamental computer science innovation is public key cryptography,

which made private communication and safe online commerce possible, enabling
businesses to communicate and interact through cyberspace rather than the physical
world. Cryptography hides the contents of messages so that only the sender and
receiver can read them, even if they traverse the public Internet. And cryptographic
protocols added functionality such as making it possible for two parties to identify
each other online and authenticate a transaction.

10 Mobile Computing

The next important and radical change was mobile computing, which became
practical when computers became sufficiently power-efficient (another consequence
of Moore’s law) to be packaged as smartphones. The defining moment for mobile
computing was Apple’s introduction of the iPhone in 2007 (Isaacson, 2011). It
combined in a pocket-sized package, a touchscreen interface appropriate for a
small device without a keyboard or mouse, and continuous connectivity through
the wireless telephone network. For most of the world’s population, smartphones are
the access point to the Internet and computing. “Personal” computers never shrank
smaller than a notebook and remained better suited to an office than as a constant
companion. In less than a decade, the smartphone became an object that most people
always carry.
Smartphones also changed the nature of computing by attaching cameras and
GPS receivers to computers. Smartphone cameras dramatically increased the num-
ber of photos and videos created and let everyone be a photographer and videogra-
pher. They also exploited the vast computational power of smartphones to improve
the quality of photos and videos to a level comparable with much larger and optically
sophisticated cameras operated by skilled photographers. Their GPSs introduced
location as an input to computation by continuously tracking a phone’s location in
the physical world. Location, like many features, is a two-edged sword that offers
sophisticated maps and navigation and enables tracking of people by advertisers and
malefactors.
Perhaps the most far-reaching consequence of smartphones is that they “democ-
ratized” computing in a form whose low cost and remarkably new functionality was
quickly adopted by most people worldwide. Earlier computers were concentrated in
the developed world, but smartphones are ubiquitous, with a high adoption even in
less developed countries. The deployment of wireless networks in these countries
brought the citizens of these countries to a nearly equal footing in terms of infor-
mation access and communications.
Evolution of Computing 41

11 Machine Learning

A recent and extremely significant advance in computing is machine learning (ML),

the process of automatically inferring features in data collections and applying this
inference to make predictions from and take actions on other, unseen data. For
example, an ML system can be trained on a large collection of labeled photographs,
e.g., a bird’s photo might be labeled “bird, Herring Gull Larus.” An ML model
trained on a large collection of photographs could then analyze other photos, even of
bird species not included in the training set, and recognize that the images contain
birds. Several years ago, ML systems reached a human performance level in this
computer vision classification task (Shankar et al., 2021).
Beyond image recognition, ML has been trained to mimic many human skills,
such as computer vision, speech recognition, language translation, grammar correc-
tion, question answering, game playing, and others. In most cases, the key enabling
factor is a large training set of labeled data. For example, large language models
(LLMs) are often trained on tens or hundreds of billions of documents from the Web.
OpenAI’s ChatGPT was trained this way and can respond to general questions with
articulate, well-formed responses and conduct a realistic dialog, albeit with many
grievous lapses, reflecting a complete lack of understanding of the underlying
meaning.
ML represents a fundamental change in how computers are programmed. For the
first seven decades, programmers wrote explicit instructions to direct a computer to
solve a task. ML shifted the perspective from “teaching” a computer to having the
computer “learn” how to accomplish a task by observing past examples. This new
approach has proven very successful in developing human-like skills for computers,
which programmers found difficult or impossible to describe fully in a program.
However, the shift leaves some people concerned that computers are becoming
“intelligent” and might soon surpass human abilities (Bostrom, 2014).
These topics are discussed in more detail in the chapter by Woltran and
Heitzinger in this volume.

12 Big Data and Cloud Computing

Underlying these advances in machine learning, and many other fields, is the ability
to collect and analyze vast amounts of data, known as “Big Data.” The hardware and
software infrastructure for storing and processing this data was originally developed
for Web applications such as search engines, which harness warehouses full of tens
of thousands of computers to index most Internet pages and rapidly respond to user
queries (Barroso et al., 2013). Each search triggers coordinated activity across
thousands of computers, a challenging form of computation called parallel
computing.
42 J. R. Larus

Internet search was made possible by advances in many fields of computer

science, including computer design, high-bandwidth networking, inexpensive stor-
age devices, and research on using multiple computers to solve a single task. The
infrastructure was originally the proprietary asset of a few companies. Amazon
democratized this form of computing with a product that came to be called “cloud
computing.” It comprises low-cost computer access in Internet-connected data
centers (the cloud) and sophisticated software for building reliable and scalable
systems on collections of computers. Before this, a company needed to buy and
manage its own computers, which limited the tasks that most companies could
accomplish with their limited capital and expertise.
Cloud computing effectively removed the barrier to constructing large-scale
computing systems. This has made it possible to collect, store, and analyze vast
amounts of data, now a routine business practice. Many websites record every user
interaction in detail, and these records are retained to provide the raw material to
train machine-learned systems. This practice has serious privacy implications but is
routine because data [aka the “new oil” (na, 2017)] promises to be the raw material to
build profitable new businesses.
More benignly, the ability to collect and analyze large amounts of data is
changing how other fields of science and engineering conduct research. Jim Gray
called Big Data the fourth paradigm of scientific discovery (after observation,
theory, and modeling) (Hey et al., 2009).

13 Security and Privacy

Because computers contain valuable information and control important devices and
activities, they have long been the target of malicious and criminal attempts to steal
data or disable their functions. The Internet greatly worsened these problems by
making nearly every computer accessible worldwide.
Computer science has failed to develop a software engineering discipline that
enables us to construct robust software and systems. Every nontrivial program (with
a handful of exceptions) contains software defects (“bugs”), some of which would
allow an attacker to gain access to a computer system. The arms race between the
attackers and developers is very one-sided since an attacker only needs to find one
usable flaw, but the developer must eliminate all flaws. Like security in general,
mitigations—updating software to fix bugs, watching for attacks, and encrypting
information—are essential.
Privacy is typically grouped with security because the two fields are closely
related. Privacy entails personal control of your information: what you do, what
you say, where you go, whom you meet, etc. However, privacy differs in a crucial
aspect from security since the owners and designers of systems abuse privacy
because this personal information has significant value that can be exploited. See
the chapter by Weippl and Sebastian in this volume.
Evolution of Computing 43

14 Conclusions

A natural question is whether computing’s rapid growth and evolution can continue.
As Niels Bohr said, “Prediction is very difficult, especially about the future.” I
believe computing will continue to grow and evolve, albeit in different but still
exciting directions. New techniques to perform computation, for example, based on
biology or quantum phenomena, may provide solutions to problems that are intrac-
table today. At the same time, new inventions and improved engineering will
continue to advance general-purpose computing. However, the enjoyable decades
of exponential improvement are certainly finished. Computing will become similar
to other fields in which improvement is slow and continuous.
The separate questions of whether computing’s rapid growth was good or bad and
whether its likely demise is good or bad can be evaluated in the context of the rest of
this book. In many ways, this question is like asking whether the printing press was
good or bad. Its introduction allowed the widespread printing of vernacular bibles,
which supported the Protestant Reformation and led to decades of religious and civil
war. Was that too large a cost to spread literacy beyond a few monks and royalty?
Computing has also disrupted our world and will likely continue to do so. But these
disruptions must be balanced against the many ways it has improved our life and
brought knowledge and communication to the world’s entire population.
Discussion Questions for Students and Their Teachers
1. Computers have grown cheaper, smaller, faster, and more ubiquitous. As such,
they have become more embedded throughout our daily life, making it possible to
collect vast amounts of information on our activities and interests. What apps or
services would you stop using to regain privacy and autonomy? Do you see any
alternatives to these apps and services?
2. Many aspects of computing work better at a large scale. For instance, an Internet
search engine needs to index the full Web to be useful, and machine learning
needs large data sets and expensive training to get good accuracy. Once these
enormous startup costs are paid, it is relatively inexpensive to service another
customer. What are the consequences of this scale for business and international
competition?
3. Moore’s law is coming to an end soon, and without new technological develop-
ments, the number of transistors on a chip will increase slowly, if at all. What are
the consequences of this change for the tech industry and society in general?
4. Climate change is an existential threat to humanity. Because of their ubiquity and
large power consumption, computers are sometimes seen as a major contributor
to this challenge. On the other hand, our understanding of climate change comes
from computer modeling, and computers can replace less efficient alternatives,
such as using a videoconference instead of travel. What is the actual contribution
of computing to global warming, and what can be done about it?
44 J. R. Larus

Learning Resources for Students

Many technical books and research papers describe the technical innovations men-
tioned above in great detail. They can easily be found with a search engine.
1. Lewis, H.R. (Ed.), 2021. Ideas that Created the Future: Classic Papers of
Computer Science. MIT Press, Cambridge, MA.
For convenience, many classic papers are collected in the volume edited by
Harry Lewis.
2. Barroso, L.A., Clidaras, J., Hölzle, U., 2013. The Datacenter as a Computer: An
Introduction to the Design of Warehouse-Scale Machines, 2nd ed, Synthesis
Lectures on Computer Architecture. Morgan & Claypool.
However, Lewis’s collection misses recent papers and those concerned with
practical aspects, such as building Internet-scale computer systems, which
Barroso et al. cover well.
3. In addition, the tech field is caught in the public eye, and many excellent,
accessible books talk about its history and technical aspects.
4. Waldrop, M.M., 2001. The Dream Machine: J. C. R. Licklider and the Revolution
that Made Computing Personal. Viking.
Waldrop’s book on J.C.R. Licklider is a biography of the remarkable psychol-
ogist who led the development of interactive computing and the Internet
at ARPA.
5. Hiltzik, M.A., 1999. Dealers of Lightning: Xerox PARC and the Dawn of the
Computer Age. Harper-Collins.
Hiltzik’s book followed these ideas as they were incubated at Xerox PARC, a
remarkable industrial lab.
6. Isaacson, W., 2011. Steve Jobs. Simon & Schuster.
Isaacson’s biography of Steve Jobs provides the other half of the story by
showing how he made these ideas into two products that changed the world, the
Apple Mac and iPhone.
7. Gleick, J., 2021. The Information. Vintage.
Gleick’s book dives into communication and information theory, the opposite
side of the computational coin.

References

Barroso, L. A., Clidaras, J., & Hölzle, U. (2013). The datacenter as a computer: An introduction to
the design of warehouse-scale machines (Synthesis Lectures on Computer Architecture) (2nd
ed.). Morgan & Claypool.
Bostrom, N. (2014). Superintelligence. Oxford University Press.
Campbell-Kelly, M. (2004). From airline reservations to sonic the Hedgehog. MIT Press.
Fortnow, L. (2021). Fifty years of P vs. NP and the possibility of the impossible. Communications
of the ACM, 65, 76–85. https://doi.org/10.1145/3460351
Gleick, J. (2021). The information. Vintage.
Hey, T., Tansley, S., Tolle, K., & Gray, J. (Eds.). (2009). The fourth paradigm: Data-intensive
scientific discovery. Microsoft Research.
Evolution of Computing 45

Hiltzik, M. A. (1999). Dealers of lightning: Xerox PARC and the dawn of the computer age.
Harper-Collins.
Isaacson, W. (2011). Steve Jobs. Simon & Schuster.
Lewis, H. R. (Ed.). (2021). Ideas that created the future: Classic papers of computer science. MIT
Press.
Mccauley, J., Shenker, S., & Varghese, G. (2023). Extracting the essential simplicity of the Internet.
Communications of the ACM, 66, 64–74. https://doi.org/10.1145/3547137
na. 2973. Records, Computers and the Rights of Citizens (No. DHEW NO. 73-94). US Department
of Health, Education, and Welfare.
na. (2023a). IBM 701. Wikipedia.
na. (2023b). Semi-automatic ground environment. Wikipedia.
na. (2017). The world’s most valuable resource. The Economist.
na. (2003). IBM Archives: 701 Feeds and speeds [WWW Document]. Accessed 17 February
17, 2023, from www-apache-app.cwm.gtm.ibm.net/ibm/history/exhibits/701/701_feeds.html
Shankar, V., Roelofs, R., Mania, H., Fang, A., Recht, B., & Schmidt, L. (2021). Evaluating machine
accuracy on ImageNet. NeurIPS 2021 Workshop on ImageNet: Past, Present, and Future.
Turing, A. M. (1950). Computing machinery and intelligence. Mind, LIX, 433–460. https://doi.org/
10.1093/mind/LIX.236.433
Turing, A. M. (1937). On computable numbers, with an application to the Entscheidungsproblem.
Proceedings of the London Mathematical Society, s2-42, 230–265. https://doi.org/10.1112/
plms/s2-42.1.230
Waldrop, M. M. (2001). The dream machine: J. C. R. Licklider and the revolution that made
computing personal. Viking.
Weiser, M. (1991). The computer for the 21st century. Scientific American, 265, 94–105.
Wing, J. M. (2006). Computational thinking. Communications of the ACM, 49, 33–35. https://doi.
org/10.1145/1118178.1118215

Open Access This chapter is licensed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as long as you give appropriate
credit to the original author(s) and the source, provide a link to the Creative Commons license and
indicate if changes were made.
The images or other third party material in this chapter are included in the chapter's Creative
Commons license, unless indicated otherwise in a credit line to the material. If material is not
included in the chapter's Creative Commons license and your intended use is not permitted by
statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder.