The first life and death of Italian computing (1954 – 1964) – The Wargaming Scribe

National beginnings: Spain – Argentina – Uruguay – France (1939-1955, 1955-1975, 1975-1985), Italy

Recently, El Explorador de RPG covered the Italian RPG Avventura 1, which drew my attention to two Italian games in my domain: Avventura 2 (a Star Trek-inspired game) and Battaglia d’Inghilterra – which means it’s time to look into the Italian ecosystem. I don’t think it’s controversial to state that, like the rest of Europe, Italy is not the first country you think of when you hear about state-of-the-art computing. For a very short time however, Italy was at the cutting edge of computing, so much so that it provided some of the computing power used in the Apollo program. It did not last, and despite its technical excellence Italian computing was defeated by Italy’s age-old short-termism and technocratic incompetence. Let’s go through this story of rags-to-riches-to-rags.

I. The defeated March of Picone

As in France, the First World War revealed to the Italians the need for new ways to automate calculations. In 1917, the mathematician Mauro Picone, serving with the VI Armata (which included two Corps of Truppe Altipiani defending the Asiago plateau), was confronted among other tasks with the problem of calculating firing tables for heavy mountain artillery. Computing a firing solution in the Alps was far harder than on the plains of Belgium: air density varied with altitude, and guns had to engage targets that could sit above or below them. To create the firing tables, Picone turned the VI Armata‘s Firing Office into a computing centre, equipping it with mechanical calculators borrowed from universities across the country. After the war, convinced that the demand for calculation would only keep growing, he founded a calculation research centre in Naples in 1927. In 1932, the centre moved to Rome and was renamed the Istituto Nazionale per le Applicazioni del Calcolo, or INAC.

Picone and the INAC fully understood the limitations of mechanical calculation and, like everyone else, they started investigating electro-mechanical and electronic computing. When Picone learned about the analogue Mallock machine in Cambridge, he sent a student to test it with a set of nine equations with nine unknowns to solve (something theoretically within the capabilities of the machine), but either the student failed to use the machine or it was poorly calibrated that day, since the student reported that the machine was quick but its results “totally wrong” – had he been given the choice, Picone would have preferred the opposite. While the INAC tried to build a working version of the Mallock machine, another researcher named Alessandro Boni proposed a totally different solution: a punched-tape-operated electromechanical device with magnetic storage – a promising design, similar in principle to that of the Harvard Mark I. A patent was filed in 1940, and there were plans to build that contraption with the help of Olivetti and the Italian army, but the project was eventually stopped without a working prototype when Italy signed the armistice in September 1943. Like France, Italy came out of the war with promising designs but no working computer.

After the war, Picone kept abreast of American computing thanks to contacts with both US officials and Italian émigrés and resumed his project to build a computer. Initially, the Americans refused to provide one of theirs and Picone partnered with the French, in particular with Louis Couffignal’s CNRS computing project – if you read the French article you can guess how it ended. The situation finally cleared after 1948, once the Americans became keen on supporting their new NATO partner. In 1950, the Americans welcomed in their labs several Italian scientists from both the INAC (including Picone himself) and Olivetti. The INAC was finally ready to build its own computer.

It did not unfold as it should have. By 1952, the Consiglio Nazionale delle Ricerche (CNR) – which was overseeing the INAC but also a huge chunk of Italian public research – had waited for a computer for too long: now it cared more about having a computer than about building one. Its various laboratories needed computing power, and so while it wasn’t a straight “no” the CNR was unwilling to fund any risky undertaking that might delay access to computing any further. This left the funding to Olivetti, which was adamant it wanted to build a computer to develop the technical expertise, but also unwilling to carry the financial risk alone. Eventually, Olivetti pulled back from the project altogether, leaving the INAC so underfunded it was not even able to buy the American computer it wanted; instead, the INAC ordered a British Ferranti Mk1* in 1954. By then, the Centro di Calcoli Numerici in Milan had ordered a CRC 102-A. The CRC 102-A was the budget option of the early 1950s computing: the Ferranti was ten times faster and had more memory than the CRC 102-A, but the latter arrived first in 1954, with the Ferranti only inaugurated in 1955. On its inauguration, the Ferranti was set to play the Triumphal March from Aida, but for Picone there was nothing triumphal about the arrival of the Ferranti: not only was it not an Italian computer, but it was also only the second computer in Italy.

For our purposes, the CRC 102-A is a milestone, as the first Italian game was programmed on it in late 1954. It was a simple player-versus-computer game called DADI [“Dice“], in which both sides tried to reach a target number by “playing” a number between 1 and 6, with the added twist compared to Nim that you could not play the same number as your opponent, or the number on the opposite face of the die (e.g. no 1 if your opponent played a 6). As simple as it was, this game would find an unexpected audience a decade later, but let’s not get ahead of our story.

The Milanese CRC 102-A, the Roman Ferranti and the imported computers that followed provided hands-on training for future engineers, but did not build up any domestic capability. This would change when the University of Pisa entered the arena.

II. The last gift from Fermi

In 1953, the University of Pisa raised substantial regional grants to build the first Italian electrosynchrotron. However, in April 1954 and at the last possible moment, the accelerator was assigned to the city of Frascati, leaving the University of Pisa with promises it could no longer keep and a lot of money. Having too much money and no project is a rare situation for a university – usually it is the opposite – and so Pisa sent representatives to the famous Italian physicist and Pisa alumnus Enrico Fermi, asking him which prestigious scientific project Pisa could undertake. Fermi’s answer, later confirmed in a letter to the Rector, was that Pisa should build a large electronic computer: “An electronic computer would constitute a research instrument from which all science and research activities would profit, in such a manner that is currently inestimable.” Fermi passed away fewer than four months later, and this letter has sometimes been described as “the last gift Italy inherited from Fermi.” Yet, if Fermi supplied the idea, two others would turn it into an Italian industry: Mario Tchou and Adriano Olivetti.

I could write a whole article and more on the many lives of Adriano Olivetti (1901-1960), engineer, entrepreneur, anti-Fascist, social theorist and, maybe more than anything else, idealist. Suffice to say for our purposes that he was the son of Camillo Olivetti, founder in 1908 of the typewriter manufacturer that bore their name. Adriano was an avid student of new ways to do things, everything, whether artistic, organisational, industrial, scientific, commercial or political. As his father’s right-hand man and then as president of Olivetti, he steered the family firm into becoming both a national champion and the showcase for a new social model called Comunità – the “Industrial City” of Ivrea he built is now a UNESCO World Heritage Site. By the 1950s, Olivetti had expanded far beyond Italy, or even Europe, and it had become one of the world leaders in small-size mechanical equipment: calculators, typewriters and the like.

Like Picone, Adriano kept close tabs on the progress of computing, of which Fermi himself had made him aware as early as 1949. Before almost anyone else, he realised that electronic computers would eventually transform the market in which he operated, not only by being better, but by being part of integrated systems that would make his current business obsolete. Adriano was particularly worried about IBM, and in 1950 he sent his engineer Michele Canepa to learn modern computing first-hand from Howard Aiken at Harvard. After learning this new trade, Canepa helped establish Olivetti’s first electronic research lab in New Canaan in 1952. As we know, Adriano’s plan to build an Italian computer with the INAC didn’t pan out, but the Pisan project was a new and perhaps better opportunity. So, early in 1955, Adriano offered the University of Pisa a deal it could not refuse, because, frankly, it had no real drawback: Olivetti would add its own financial support to the project, provided that the team included some of Olivetti’s engineers, in particular one of Adriano’s young protégés: Mario Tchou.

Mario Tchou (1924-1961) was the son of a Chinese diplomat to the Holy See. Protected from the Fascist regime by his family and his nationality, he studied at the prestigious Liceo Tasso in Rome before leaving for the United States in 1945. He completed his studies at the Polytechnic Institute of New York in 1949 and then, thanks again to Fermi who apparently helped every Italian he met, was eventually given a position in Columbia University’s industrial electronics department. This is where he met Adriano in 1954. Tchou – energetic, charismatic and, rarest of all, perfectly knowledgeable about computers – made an impression on Adriano. Adriano knew he had found the technical lead he needed; he offered Tchou a chance to return to Italy, work for the University of Pisa, build a team and then take charge of a new Olivetti lab: the Laboratorio di Ricerche Elettroniche (LRE), to be established within walking distance of the University.

The joint venture between Olivetti and the University of Pisa – a public-private partnership common in the United States but rare at the time in Europe – worked perfectly. Tchou recruited the best engineers that money and the prestige of Pisa could afford (including some from the USA and the UK), and those engineers could work with parts designed specifically by Olivetti. Even the most complex components would not resist Adriano and Tchou’s systematic approach: they would be supplied either by the facility in New Canaan (notably the magnetic drums) or by a new Italian semiconductor manufacturer created by Olivetti for the occasion: the Società Generale Semiconduttori. All those resources moved seamlessly between the University of Pisa and the Olivetti LRE, with Adriano’s exceptional drive and influence handling any organisational or bureaucratic hurdle that might have arisen.

The University of Pisa got what it wanted: the experimental Macchina Ridotta [the Smaller Machine] in July 1957 and then the superb Calcolatrice Elettronica Pisana (CEP) in November 1961. However, Olivetti had effectively been a senior partner in this alliance, so it inaugurated its computer first: the fully functional vacuum-tube Elea 9001 (ELaboratore Elettronico Aritmetico), unveiled in spring 1957. While not ready for commercial use, the 9001 paved the way for the rest of the Elea series. The first commercial computer model would be the Elea 9003 (November 1959) – at its inauguration one of the world’s few fully transistorised computers. Olivetti had an answer to IBM.

III. Death, debt and General Electric

With the creation of the Elea, our company has not merely extended its traditional production to a new sector with immense possibilities, but has attained a goal in which what I think is the inalienable, noblest end that an industry must set itself will directly come to pass, that’s to say, not only the success of its own name and work, but the common progress – economic, social, ethical – of the entire community.

Adriano Olivetti, November 1959

With the Elea 9003, Adriano had fulfilled his vision. Starting in March 1960, the computer would be sold or leased to more than thirty customers, including giants such as ENI or FIAT. However, Adriano would never see that: he died of a heart attack on the 27th of February 1960. Not only did Adriano have no opportunity to properly prepare his succession, but his death came at the worst possible moment for Olivetti: in addition to the massive investment in computing, Olivetti was under financial strain because it had begun buying its American competitor Underwood in 1959. Underwood lost an ungodly amount of money for years, and the problem would become the central focus of the new head of Olivetti Giuseppe Pero, a company veteran and close friend of Adriano, but with none of his social or leadership skills. Pero left Tchou free to continue developing Olivetti’s computing, and in 1961 Olivetti released the (relatively) low-cost Elea 6001. More than 100 Elea 6001s were built, mostly for the domestic market.

Like Adriano, Tchou would not see this success: in November 1961, he died in a car accident. His death, so soon after Adriano’s, is still the subject of conspiracy theories in Italy. The computer division was taken over by Roberto, relocated near Milan and divided between a commercial Divisione Elettronica and the research-focused LRE. The business was growing, and Olivetti was by then Italy’s second-largest computer supplier, though far behind IBM. However, the Divisione Elettronica and the LRE were still losing between 0.6 and 1.4 billion lire (₤) in 1963, depending on whether you asked Olivetti or its banker: the Italian market was small, R&D costs were massive and the computers were mostly leased, which means that profit was slow to arrive. Meanwhile, the traditional mechanical activity had generated ₤12.1 billion in profit in 1963. This discrepancy fed resentment and even hatred within Olivetti: mechanical and electronic engineers had unbridgeable cultural differences to start with, but now the mechanical engineers saw that their work was being used to fund the technology that would make their skills obsolete – Roberto would find no internal support for the survival of his division.

The computer activity and the acquisition of Underwood had been financed by opening up Olivetti’s capital with the help of Mediobanca, an operation approved by the shareholders a few days before Adriano’s death. The new capital structure was complex as Olivetti was then owned by six family branches (Adriano had five siblings) and three “allied” families: to keep the company under family control, the new shares were divided between the ordinary shares (voting, non-publicly traded) and the privileged shares (generally non-voting and publicly traded) – and of course the family held a majority of the former. As the whole Underwood fiasco dragged on, eventually swallowing 100 million dollars, Pero had raised more capital in 1961 and 1962, forcing the family shareholders to borrow money in order to contribute, with their debt secured against their ordinary shares. This detail would be fatal for Olivetti’s computing.

Pero died after a short illness in November 1963 and to replace him the shareholders chose in murky circumstances Bruno Visentini, the vice-president of the state-owned Istituto per la Ricostruzione Industriale (IRI) and as such an outsider. What follows is extremely controversial, with major historiographical disagreements about the events and the motivations of the people involved. To recount what I would call the central consensus, Olivetti’s core business was still profitable in 1964 and the debt level was, depending on whom you read, either healthy or on the verge of a liquidity crisis – although we know that 1964 closed in the black. On the other hand, the shareholders’ situation was clear: the value of the shares they had put up as collateral for their debt had collapsed to ₤1500, partly due to Olivetti’s financial difficulties and partly due to a general Italian market slump. As the collateral was not covering the debt anymore, the family shareholders had to bridge the gap; otherwise the banks threatened to simply seize the shares.

To address the Olivetti shareholders’ debt crisis while pretending to address the Olivetti crisis, Visentini assembled a Gruppo d’Intervento: IMI, FIAT, Mediobanca, Pirelli and La Centrale. The Gruppo was willing to refinance Olivetti and buy 5 million of the family’s ordinary shares, but there was one condition – one condition that could not be negotiated. As the head of FIAT explained to his own shareholders in April 1964: “Over [Olivetti’s] future there is a looming threat, a blemish to be removed: its entry into the electronics sector, which requires investments that no Italian company can afford“. By chance, the Gruppo d’Intervento had found a buyer: General Electric (GE). As we know from the article on French computing, GE was at the time trying to enter the computer market and had started talking to the Gruppo as early as February 1964. Most of the shareholders on the other hand had no specific vision for the future of the Divisione Elettronica, except that they were being told that it was an obstacle to receiving dividends again. In May 1964, the principles of the rescue plan:

• the sale to the Gruppo of some of the ordinary shares at ₤1000 each (at least one third less than what they were worth at that time),
• a ₤20 billion line of credit that Olivetti barely used,
• the sale of 75% of Olivetti’s computer division to GE (announced in August 1964) for an estimated ₤8 billion.

As Fortune would write: “Never have saviours obtained so much while risking so little”. Just like that, GE had snatched up another European national computing champion.

Where France and Italy differ, of course, is that despite Roberto’s efforts the government never saw Olivetti as a national champion of computing. Olivetti’s computers were not prioritised by the public bodies that bought computers, and consequently there was no attempt to save the Divisione Elettronica from GE. Given the behaviour of the Gruppo, some authors reckon that FIAT, Mediobanca or maybe both had an ulterior motive (improving relations with GE or with “the Americans”) in handing the computing activity of Olivetti to GE. However, most authors prefer to apply Hanlon’s razor: Vittorio Valletta of FIAT was an 80-year-old man who did not understand electronics, the head of Mediobanca was deeply averse to risk, and generally the political and economic elite of Italy was ignorant of what was at stake with electronics. More than anything else maybe, the Italian establishment doubted that Italy could or even should compete with the Americans on innovation – on this certainly the Italians were the very opposite of the French.

IV. Olivetti’s last electronic ember

The sale of Olivetti’s electronic activity to GE marked the end of the first phase of independent Italian electronic computing. The sale however left a speck of electronic computing with Olivetti – one that ended up more prestigious and innovative than anything the new Olivetti-General Electric division would ever do: the Programma 101. In 1957, Pier Giorgio Perotto had joined the newly created LRE in Pisa and was initially assigned to work on the console of the Elea 9003. When Tchou caught Perotto finishing the paintwork on the computer’s casing, he stated it was a poor use of his skills and asked him to design a machine that would convert the paper tapes produced by the mechanical accounting machines into punched cards. Tchou’s objective was to buy some goodwill from the mechanical engineers: it failed superbly at that, but the convertitore banda schede was nonetheless Olivetti’s first commercial electronic product (autumn 1958).

Tchou was satisfied, and Perotto continued designing small electronic devices for Olivetti. In spring 1962, Roberto gave Perotto a new project, the most ambitious thus far. The engine of Olivetti’s success in the 1950s was the incredibly profitable MC-24 series of mechanical calculators: one source states that the production cost of one of those calculators was ₤39 000 and the selling price ₤390 000 – almost the price of a small car. Roberto wanted the Divisione Elettronica to have its own ultra-profitable calculator, ideally in a higher price range to avoid cannibalising the existing line-up, but I cannot help believing that Roberto would not have minded too much if such a calculator had shifted the profits from the mechanical division to the electronic one.

The requirement was for the calculator to be smallish (the same size as existing calculators), be fully electronic, be usable by anyone without programming skills and have a memory – an extremely complex set of constraints given that nothing similar existed. It took several years, but Perotto was helped by his charisma and the non-hierarchical nature of the LRE in which you could easily “borrow” specialists if you could convince them your project was interesting. After a lot of trial and error, the first prototype was ready in January 1964, right in the middle of the shareholder crisis. As for the final prototype (nicknamed the Perottina), it was delivered in October 1964, in theory after the Divisione Elettronica had been bought by GE.

Except that GE had not bought all of the Divisione Elettronica: it actually had bought the Divisione Elettronica, minus the “teleprinters, accounting machines, numerical control and the peripheral and subsidiary equipments of large computers“; in other words, GE was after the computers and nothing else. When Perotto visited GE, he became absolutely certain he did not want to work for it; furthermore he understood that GE’s vision of time-sharing a mainframe’s computing power went against the distributed computing philosophy of the Perottina. Many accounts state that Perotto spent days removing all references to “computer” from his project in order to move outside of the scope of the acquisition. Perotto gives another version: “I assumed an attitude of bellicose opposition to the agreement at all the tables to which I had the good fortune of being called. […] The conclusion was that the Americans let Roberto Olivetti know, not even all that discreetly, that they would not be displeased if the engineer Perotto, while remaining part of the [Milanese] electronics laboratory being handed over to them, stayed behind with Olivetti. And so it actually happened.”

The calculator was named the Programma 101 (according to its hardware engineer De Sandre: “it reflected its most important feature well, and it also sounded good in English with an American accent“) and shown for the first time at the Business Equipment Manufacturers Association (BEMA) show in October 1965. It was presented as a minor niche product, while the star of the Olivetti line-up was the new mechanical calculator Logos 27. Needless to say that no one remembers the Logos 27, but then it was not a fair fight: the Programma 101 was a powerful calculator to begin with: 22 digits of precision and a square-root key. It could also be operated without all the cranking and shifting of a mechanical machine. But more importantly, as the name implies, one could also program directly on the calculator and then save the program on a magnetic card and share it with a colleague. At its first demonstration in front of journalists, the Programma 101 received a round of applause when it calculated the elliptical trajectory of a satellite; later some people even asked to check whether the device was secretly plugged into a computer. But the crowd-favourite software of the Programma 101, and the one that made Olivetti the most popular booth of the show, was a game called Angela whose description matches the CRC 102-A’s DADI perfectly: the old Milanese test game of 1954 had crossed the ocean!

The Programma 101 had almost no competition for two years and was a resounding success: more than 40 000 units sold for around $3200 in the USA (where most of the sales were made) and ₤2.0 million in Italy. Ten units were famously bought by NASA, making it the only non-American computer used in the Apollo program. More units were infamously bought by the US Air Force to help with its bombing campaign in Vietnam.

The success of the Programma 101 helped Olivetti recover financially. Together with the failure of the Logos 27 and the drop in sales of the MC-24, it convinced the management that, maybe, they had been too hasty in selling the Divisione Elettronica. Things began to change, and even the mechanical division began to leverage electronics. Roberto had been sidelined by the “rescue” and left in charge of the rump of the former Divisione Elettronica; thanks to the Programma 101 he became the company’s rising star. In 1967 he was promoted to (co-)managing director of Olivetti, while Perotto took charge of R&D. Like Adriano and Tchou before them, Roberto and Perotto were now Olivetti’s power couple.

The deal with GE included a provision that Olivetti would no longer build “large computers”, so it had to stay out of this specific market. For the rest, it went all-in: in addition to variations on the Programma 101, it converted all its mechanical products to electronics. As Perotto remembers it, by the early 1970s Olivetti had become an electronic company with cutting-edge products. An electronic company without “large computers”, of course, but – well – the era of microcomputers was coming.

Sources:

If you need to read only one, google translate and read Programma 101 by Pier Giorgio Perotto.

On the early machines

• Storia dell’informatica italiana, Maristella Agosti and Virginio Cantoni, 2004 – if you want to know everything about the earliest machines. Includes the pre-war efforts.

• Early computers of Italy, Giuseppe de Marco, Giovanni Mainetto, Serena Pisani, Pasquale Savino, 1999

• Archivi Olivetti: US activities & Missed Opportunity

• Retracing and Assessing the CEP Project, Giovanni A. Cignoni, Fabio Gadducci, 2015(?)

• Museo Virtuale dell’informatica: un esempio emblematico, Paolo Paolini, 2009, with some notes on the Milanese computer.

• Il pc dell’Olivetti (che spaventò l’America) e il mistero della morte di Mario Tchou, Corriere della Sera, 2020 – on the life and death of Mario Tchou

• Storia e memoria del personal computer – il caso italiano, Marcello Zane, 2008; includes a lot of hard data for market penetration of mainframes (how many units, when, which model, which owner and so on)

• Oral History: Ottorino Beltrami on the Engineering and Technology History Wiki, which I discovered as I wrote this article.

On the sale of the Divisione Elettronica

• When the Italians Made Computers, Giorgio Garuzzo, 2015.
  Autobiography by an Olivetti insider. It gives fascinating accounts of the day-to-day operations of an Olivetti (mechanographic) factory (how to handle unions! how to handle affairs! how to handle terrible managers!) and of the infighting after the death of Adriano Olivetti and Mario Tchou.

• Mediobanca e il salvataggio Olivetti, Giampietro Morreale, 2019.
  A 2019 Mediobanca assessment of the situation, so of course it’s the family shareholders’ fault and Olivetti was on the brink of a liquidity crisis. On the other hand, it includes a lot of documents from the era, including the minutes of all the (official?) meetings. Great for the details of the financial history of Olivetti.

• Olivetti. La crisi del 1964 e la perdita progressiva dei suoi valori fondativi, Giuseppe Silmo, 2022
  A summary of all the assessments done on Olivetti’s financial situation in 1964 – some of them I could not check personally.

• L’Olivetti dalla macchina da scrivere alla telefonia, Elsa Boscia, 2000. It states that Olivetti was not going well at all, but doesn’t take a side on whose fault it was.

• Normalizzare l’innovazione. Le vicende dell’elettronica e dell’informatica da Adriano a Roberto Olivetti, Giuliana Gemelli, 2013

• A few disjointed sources on the Elea 4115 and the Honeywell computers that followed. They were not the core of the article so I did not record them properly 🙂

Programma 101

• Programma 101, Pier Giorgio Perotto, 1995
  An account by the lead of the Programma 101 development. Note that, surprisingly given his background, Perotto believes the Divisione Elettronica could not be made profitable.

• Microstoria del progetto della Programma 101, Giovanni De Sandre, 2016
  An account by the hardware specialist of the Programma 101 development