Software is a set of programmed instructions stored in the memory of stored-program digital computers for execution by the processor. Software is a recent development in human history and is fundamental to the Information Age.
Ada Lovelace's programs for Charles Babbage's analytical engine in the 19th century are often considered the founder of the discipline. However, the mathematician's efforts remained theoretical only, as the technology of Lovelace and Babbage's day proved insufficient to build his computer. Alan Turing is credited with being the first person to come up with a theory for software in 1935, which led to the two academic fields of computer science and software engineering.
The first generation of software for early stored-program digital computers in the late 1940s had its instructions written directly in binary code, generally for mainframe computers. Later, the development of modern programming languages alongside the advancement of the home computer would greatly widen the scope and breadth of available software, beginning with assembly language, and continuing through functional programming and object-oriented programming paradigms.
Before stored-program digital computers Origins of computer science
Computing as a concept goes back to ancient times, with devices such as the abacus, the Antikythera mechanism, astrolabes, mechanical astronomical clocks and mechanical calculators. The Antikythera mechanism is an example for a highly complex ancient mechanical astronomical device.
However, these devices were pure hardware and had no software - their computing powers were directly tied to their specific form and engineering.
Software requires the concept of a general-purpose processor - what is now described as a Turing machine - as well as computer memory in which reusable sets of routines and mathematical functions comprising programs can be stored, started, and stopped individually, and only appears recently in human history.
The first known computer algorithm was written by Ada Lovelace in the 19th century for the analytical engine, to translate Luigi Menabrea's work on Bernoulli numbers for machine instruction. However, this remained theoretical only - the lesser state of engineering in the lifetime of these two mathematicians proved insufficient to construct the analytical engine.
The first modern theory of software was proposed by Alan Turing in his 1935 essay Computable numbers with an application to the Entscheidungsproblem (decision problem).
This eventually led to the creation of the twin academic fields of computer science and software engineering, which both study software and its creation. Computer science is more theoretical (Turing's essay is an example of computer science), whereas software engineering is focused on more practical concerns.
However, prior to 1946, software as we now understand it programs stored in the memory of stored-program digital computers did not yet exist. The very first electronic computing devices were instead rewired in order to "reprogram" them. The ENIAC, one of the first electronic computers, was programmed largely by women who had been previously working as human computers. Engineers would give the programmers blueprints of the ENIAC wiring and expected them to figure out how to program the machine. The women who worked as programmers prepped the ENIAC for its first public reveal, wiring the patch panels together for the demonstrations. Kathleen Booth developed assembly language in 1950 to make it easier to program the computers she worked on at Birkbeck College.
Grace Hopper worked as one of the first programmers of the Harvard Mark I. She later created a 500-page manual for the computer. Hopper is often falsely credited with coining the terms "bug" and "debugging", when she found a moth in the Mark II, causing a malfunction; however, the term was in fact already in use when she found the moth. Hopper developed the first compiler and brought her idea from working on the Mark computers to working on UNIVAC in the 1950s. Hopper also developed the programming language FLOW-MATIC to program the UNIVAC. Frances E. Holberton, also working at UNIVAC, developed a code, C-10, which let programmers use keyboard inputs and created the Sort-Merge Generator in 1951. Adele Mildred Koss and Hopper also created the precursor to a report generator.
Early days of computer software (1948–1979) In his manuscript "A Mathematical Theory of Communication", Claude Shannon (1916–2001) provided an outline for how binary logic could be implemented to program a computer. Subsequently, the first computer programmers used binary code to instruct computers to perform various tasks. Nevertheless, the process was very arduous. Computer programmers had to provide long strings of binary code to tell the computer what kind of data it should store. Code and data had to be loaded onto computers using various tedious mechanisms, including flicking switches or punching holes at predefined positions in cards and loading these punched cards into a computer. With such methods, if a mistake was made, the whole program might have to be loaded again from the beginning.
The very first time a stored-program computer held a piece of software in electronic memory and executed it successfully, was 11 am 21 June 1948, at the University of Manchester, on the Manchester Baby computer. It was written by Tom Kilburn, and calculated the highest factor of the integer 2^18 (262,144). Starting with a large trial divisor, it performed a division of 262,144 by repeated subtraction and then checked if the remainder was zero. If not, it decremented the trial divisor by one and repeated the process. Google released a tribute to the Manchester Baby, celebrating it as the "birth of software".
FORTRAN was developed by a team led by John Backus at IBM in the 1950s. The first compiler was released in 1957. The language proved so popular for scientific and technical computing that by 1963 all major manufacturers had implemented or announced FORTRAN for their computers.Jean E. Sammet (1969). Programming Languages: History and Fundamentals, Prentice Hall, Englewood Cliffs, New Jersey.R.W. Bemer (1969). A politico-social history of Algol, Annual Review in Automatic Programming, pp 151-237. Pergamon Press, Oxford.
COBOL was first conceived of when Mary K. Hawes convened a meeting (which included Grace Hopper) in 1959 to discuss how to create a computer language to be shared between businesses. Hopper's innovation with COBOL was developing a new symbolic way to write programming. Her programming was self-documenting. Betty Holberton helped edit the language which was submitted to the Government Printing Office in 1960. FORMAC was developed by Jean E. Sammet in the 1960s. Her book, Programming Languages: History and Fundamentals (1969), became an influential text.
Apollo Mission
The Apollo Mission to the moon depended on software to program the computers in the landing modules. The computers were programmed with a language called "Basic" (no relation to the BASIC programming language developed at Dartmouth at about the same time). The software also had an interpreter which was made up of a series of routines and an executive (like a modern-day operating system), which specified which programs to run and when. Both were designed by Hal Laning. Margaret Hamilton, who had previously been involved with software reliability issues when working on the US SAGE air defense system, was also part of the Apollo software team. Hamilton was in charge of the onboard flight software for the Apollo computers. Hamilton felt that software operations were not just part of the machine, but also intricately involved with the people who operated the software. Hamilton also coined the term "software engineering" while she was working at NASA.
The actual "software" for the computers in the Apollo missions was made up of wires that were threaded through magnetic cores. Where the wire went through a magnetic core, that represented a "1" and where the wire went around the core, that represented a "0." Each core stored 64 bits of information. Hamilton and others would create the software by punching holes in punch cards, which were then later processed on a Honeywell mainframe where the software could be simulated. When the code was "solid," then it was sent to be woven into the magnetic cores at Raytheon, where women known as "Little Old Ladies" worked on the wires. The program itself was "indestructible" and could even withstand lightning strikes, which happened to Apollo 12. Wiring the computers took several weeks to do, freezing software development during that time.
While using the simulators to test the programming, Hamilton discovered ways that code could produce dangerous errors when human mistakes were made while using it. NASA believed that the astronauts would not make mistakes due to their training. Hamilton was not allowed to program code to prevent errors that would lead to system crash, so she annotated the code in the program documentation. Her idea to add error-checking code was rejected as "excessive." However, exactly what Hamilton predicted would happen occurred on the Apollo 8 flight, when human error caused the computer to wipe out all of the navigational data.
Bundling of software with hardware and its legal issues Later, software was sold to multiple customers by being bundled with the hardware by original equipment manufacturers (OEMs) such as Data General, Digital Equipment and IBM. When a customer bought a minicomputer, at that time the smallest computer on the market, the computer did not come with pre-installed software, but needed to be installed by engineers employed by the OEM.
This bundling attracted the attention of US antitrust regulators, who sued IBM for improper "tying" in 1969, alleging that it was an antitrust violation that customers who wanted to obtain its software had to also buy or lease its hardware in order to do so. However, the case was dropped by the US Justice Department, after many years of attrition, as it concluded it was "without merit".
Data General also encountered legal problems related to bundling although in this case, it was due to a civil suit from a would-be competitor. When Data General introduced the Data General Nova, a company called Digidyne wanted to use its RDOS operating system on its own hardware clone. Data General refused to license their software and claimed their "bundling rights". The US Supreme Court set a precedent called Digidyne v. Data General in 1985 by letting a 9th circuit appeal court decision on the case stand, and Data General was eventually forced into licensing the operating system because it was ruled that restricting the license to only DG hardware was an illegal tying arrangement. Even though the District Court noted that "no reasonable juror could find that within this large and dynamic market with much larger competitors", Data General "had the market power to restrain trade through an illegal tie-in arrangement", the tying of the operating system to the hardware was ruled as per se illegal on appeal.