There are thousands of programming languages (strictly speaking, so say the people who consider them). The classification of such quantities of their varieties is very complex and is a source of controversy. However, the classification is confusing and often self-contradictory. I propose to limit the four branches at the top level. I tied each branch to a language that was first in my family in chronological order:
- imperative (1956, Fortran);
- functional (1959, Lisp);
- object-oriented (1972, SmallTalk);
- logical (1974, Prologue).
In my reasoning, I will be repelled by the fact that three families of four feel not bad, in the sense that they have many successors. In relation to these three, the Prologue is extremely backward. But in the early 1980s it was different. Then Prologue competed with Lisp for the minds of those who were engaged in non-IBM programming (let's call it so to avoid the ambiguous term AI).
Let's return to the article title. The killer (or deliverer, depending on how you feel about the question), I consider the Japanese project of Fifth Generation Computational Systems, which existed from 1982 to 1992. Even those who knew about the project should refresh their fascinating history in the context of time. This article is a review, on the one hand, and a theory, on the other, of how the Prologue was killed, and how Lisp was able to avoid a similar fate.
It’s worth starting since 1982. For a long time, the focus of attention was the military and political confrontation of the USSR and the USA, but now it is being superseded by industrial and commercial rivalry with Japan. Japan, destroyed and impoverished in 1945, was transformed, so far no one saw, into a brilliant model of society, creating all the envy of the modern industrial world. The Japanese have become the best not only in watches, cameras and other consumer electronics, but also in high-speed trains, industrial robots, cars, steelmaking and mainframes (which were surprisingly fully compatible with IBM systems).
Although Japan’s commercial success was frightening, there was a conviction that its main role was an imitator, not an innovator. Japan was credited with unfair competition, parasitic on foreign studies, especially on American ones. Japanese competition was also considered dishonest because Japanese companies (especially Keiretsu) could create alliances prohibited by antitrust law for their American rivals. Moreover, MITI (Ministry of Foreign Trade and Industry of Japan) was suspected of managing Keiretsu. Of course, such competition was considered dishonorable - it is so non-American.
Chalmers Johnson’s book, which is worth a look at, if you don’t read, “MITI and the Japanese Miracle: The Politics of Industrial Growth, 1925-1975,” was published in 1982. Not ten years earlier, not ten years later, such a book could not attract attention . In 1982, she became a bestseller.
Try to imagine the consequences of such a situation led to the news that MITI is launching a project to develop fundamentally new computer systems. From the program development side, the project included almost everything related to AI research. The hardware was completely parallel.
IBM marketers have taught everyone to evaluate the development of computers in terms of generations. It was argued that the lamps gave birth to the computers of the First Generation, transistors - the Second Generation. Therefore, when the IBM 360 was released, it was not just a new type of computer, it was a new generation - Third! In the 1970s, everything got confused, and there was no clear criterion for defining Fourth Generation computers. So, in 1982, MITI financed the launch of an institute called ICOT (Institute for New Generation Computer Technology), the purpose of which is to create the Fifth Generation Computing Systems.
The project was associated with two words unnerving the west: MITI and AI. MITI - for the reasons described above. AI - because it is simply impossible to listen to him quietly: for too long it was quiet in this area. Periods of calm attitudes towards AI alternate with times when unrealistic breakthroughs are expected of him. The beginning of the next such surge occurred in 1982. It seemed that Japan was changing its policy, which was already a terrible force, towards innovations in order to remove all remaining barriers and finally take over the world.
Across the world began a fuss in high offices. In the minds of the ministers responsible for production, science, commerce, economics, education and everything else, the question was itchy: What to Contrast? The Thatcher government decided that the answer would be the Alvey program; The European Union launched ESPRIT (European Strategic Program for Research in Information Technology). In the US, the situation was different; the government could not simply allocate a budget and give it to researchers describing themselves as worthy recipients of money. As a result, countermeasures in the United States were more interesting. If the government cannot counteract, then maybe the industry will form a consortium that will provide US leadership in the Fifth Generation Computing System? It will not form - it is prohibited by antitrust laws. But the passions were such that in 1984 the government approved the “Act on Collaborative National Studies”. No lobbying would be enough to pass such a law. It seems to me that the adoption of the book "The Fifth Generation" by Edward Feigenbaum and Pamela McCormick, published in 1983. Although Faygenban was an academic, very respected and standing at the origins of expert systems, the book is beautifully written. She is eloquent and exciting, like an article in The Times magazine, which has announced that it’s not someone who has been named Man of the Year, but Computer.
After explaining how expert systems will affect the development of the Knowledge Industry and how Knowledge itself will become a national treasure, Fegenbaum and McCormick continue:
To realize this vision, the Japanese have both a strategy and a tactic. Their strategy is simple and wise: to avoid a head-on collision with American companies that dominate the market now, to get ahead in the 1990s, where there is a huge scope for growth of economic potential that myopic and complacent American companies do not notice and quickly seize these spaces. The tactics are determined by the impressive MITI plan, which is called the “fifth generation”.
The Japanese plan is heavy and forward looking. It is unlikely that it will be fully implemented in the next ten years. Nevertheless, it is a very big mistake to consider it just a “hype”, as some leaders of American industry believe. Even partially implemented, but perfectly executed concepts can have enormous economic value: to prepare the market and allow the Japanese to dominate it.
In the warm atmosphere of the reception of this book, small lobbying was enough to pass an act that sufficiently weakened the antitrust laws so that the creation of a response consortium would become legitimate. As the head of the project, a suitable admiral was found, apparently under the impression of the Manhetenna Project, which was headed by a general. This admiral was Bobby Ray Inman, former director of the National Security Agency and deputy director of the Central Intelligence Agency. The consortium was named “Microelectronics and Computational Technologies Corporation” and located in Austin, Texas.
Project countering computers of the fifth generation had many opponents. The main argument was the statement that the project of computers of the fifth generation should not be taken seriously. It was suggested that the cunning Japanese created a sensation to make competitors waste their resources and weaken the west. Some said that the systems of the fifth generation are extremely hype, because directed towards AI. And even if the Japanese begin to do something related to AI, they will be engaged in Lisp machines, and not to replace Lisp with ... how is it there ... well, this ... yes, for sure - the Prologue.
Nevertheless, the Prologue choice was voiced firsthand at the Fifth Generation International Computer System Conference, held in Tokyo October 19-21, 1981. A collection of reports, edited by T. Moto-Oka, is still kept in many libraries. The conference gave the official launch of the project. Many reports of the organizing committee describe and explain how breakthroughs in AI, software and hardware will allow computers to take society to a new level of prosperity and harmony. But there were also reports by scholars, the reports by F. Fuci (who later became the director of ICOT) and K. Furukawa (who became the head of the direction at ICOT) deserve special attention. If the executive committee talked about choosing between Lisp and Prolog as the language and the functional and logical approach in hardware, then Fuchi and Furukawa clearly defined the concepts: Prolog is a language, logic programming is a methodology. Parallelism was seen as the main direction of hardware development, and Prolog (being associated with it at a low level) seemed to be the most appropriate tool. Consequently, the systems of the fifth generation will be parallel.
Fast forward to 1992. The world has changed. In 1990, the Nikkei index was growing a record long time from the very beginning of the fifth generation system project and was about to break the bar at 40,000. But instead of continuing to grow, it began to fall, and by 1992 it was already half as large as its peak value. Most of the Toyota and Honda, driving around the states, were made in the states. If anyone remembered MITI, it was only in textbooks telling that MITI never financed a single successful project and never helped the industry, but only prevented it from intervening. The 1992 book, “The End of History and the Last Man,” by Francis Fukuyama, sings praises to the spread of the American way of life throughout the world. The companies involved in Lisp machines disappeared, and those that remained turned into something very different from the manufacturers of Lisp machines. Intel everywhere demonstrated Moore's law on its traditional processors, which executed Lisp programs on ordinary PCs faster than Lisp machines. The rapid increase in speeds of conventional processors killed the interest in parallelism, which is much more difficult to put into practice. The prologue version of the parallel Lisp machine, so tempting and promising in 1982, has become a relic.
Meanwhile, Tokyo’s fifth-generation Computational Systems 1992 conference is taking place, marking the end of this project. Some reports were published in the March issue of Communications, published by ACM. The project participants could not cope with people who only thought that the project would create a new generation of systems that would lead society to a new level of prosperity and harmony, and who forgot about the very smallness of parallel computers. Fucci’s first report was a veiled apology, with the excuse “but we have a wonderful parallel computing machine.” The second report, written by Robert Kowalski, is clearer and even contained a topic in bold type: “What went wrong?”
That's all. Just "hype" in the end. Fifth-generation computing systems went into oblivion, taking logic programming and Prolog with them. The fatal connection between logic programming and fifth-generation systems was formed only because Fuchi and Furukawa fell in love with Prolog. The moral is that if you omit the chatter of organizational committees, people choose between technologies, and they choose the ones they can love. In my next article, I plan to talk about how the Prologue appeared on the radars of the Japanese, when the sky glittered from Lisp shots. The shots made by people who love Lisp. I will try to describe as well as possible how people fall in love with Lisp, and how the same thing happens with the Prologue.
Continued: " Fatal choice