Une hallucination des Intelligences Artificielles

Leave a reply

I made a mistake and published a French article in the English part. I will not correct it but add a French version elsewhere!

Je reste partagé par l’impact qu’aura l’intelligence artificielle dans nos vies. L’internet ne m’avait pas laissé aussi circonspect. Je ne doute pas que l’impact sera considérable, mais en lisant deux articles scientifiques assez passionnants, j’en suis arrivé à faire une brève expérience que je décris plus bas et que je trouve amusante.

Le premier papier est celui de Michael I. Jordan A Collectivist, Economic Perspective on AI. Ce bref extrait m’a beaucoup plu : “Alors qu’un LLM peut apparaître comme une « entité » unique à l’apparence humaine, il est tout aussi pertinent de le considérer comme un artefact « collectiviste ». En effet, interagir avec un LLM revient à interagir implicitement avec un grand nombre d’individus ayant contribué, via Internet et d’autres médias, à des données, opinions, constructions linguistiques et œuvres créatives à un niveau micro. Lorsque ces contributions humaines convergent de diverses manières, le LLM est capable de transformer cette convergence en abstractions utiles, renforçant ainsi l’illusion de personnalité. Si l’analogie entre un LLM et une personne semble irrésistible, l’analogie avec une culture est tout aussi valable. Les cultures sont des réservoirs de récits, d’opinions et d’abstractions. Les cultures ont une personnalité.” (traduit avec Google depuis “whereas an LLM may appear to be a single “entity” that is human-like, it is equally well understood as a “collectivist” artifact. Indeed, in interacting with an LLM, one is interacting implicitly with a vast number of humans who have contributed micro-level data, opinions, linguistic constructions, and creative works to the LLM via the Internet and other media. When these human contributions agree in various ways, the LLM is able to promote that agreement into abstractions that are useful and that strengthen the illusion of personhood. But, while the analogy of an LLM to a person seems irresistible, an analogy of an LLM to a culture is equally valid. Cultures are repositories of narratives, opinions, and abstractions. Cultures have personalities.”)

Le second papier est celui de Stéphan-Eloïse Gras et Gaël Varoquaux Connaître avec les modèles de langage : une rupture paradigmatique. L’article y aborde notre rapport à la connaissance et à la rationnalité et un simple exemple est fascinant : “Il est possible, selon la manière dont on pose la question, de pousser le robot conversationnel à dire de véritables stupidités, notamment en matière de géographie. L’exemple le plus connu, et très marquant car il a été repris par Google Search Generative Experience concerne la question “Savais-tu qu’il n’y a pas de pays en Afrique commençant par la lettre K?”. Le robot, programmé pour répondre “oui, et”, à une question sur l’étendue de ses connaissances, se trouve pris au piège ; le comble de l’absurde est atteint quand, du fait sans doute de la faible qualité des sources géographiques, sa réponse confirme à l’utilisateur qu’il n’existe en effet pas de pays commençant par la lettre K.”

Alors j’ai refait l’expérience avec cet exemple mais aussi bien chatGPT que Claude, Gemini et LeChat ont répondu correctement. J’ai refait l’expérience avec la question “Quel(s) département(s) français ont un nom qui commence par la lettre Y ?” et voici les réponses :

selon ChatGPT de openAI :

selon Gemini de Google :

selon Claude de Anthropic :

selon Le Chat de Mistral :

Dans un second temps j’ai repris la question et cela a donné :

Heureusement mon moteur de recherche préféré n’a pas perdu la mémoire !

PS: Dans un commentaire au post que j’ai fait sur LinkedIn sur cet article, Hugues Séverac m’indique que “le mode de fonctionnement des LLMs consistant à découper les textes en token c-a-d autre chose que des lettres, les rend particulièrement fragiles pour traiter des requêtes concernant des lettres”. J’aurais presque dû y penser et le mentionnant, il ajouter ensuite “C’est un piège connu depuis un moment par les gens du domaine, mais ça bouge pas vite 😏”.

Je me souviens que le moteur de recherche de Google a lui aussi ses inexactitudes mais au fond on s’en moque un peu car donner un lien pu précis voire incorrect n’est pas de même nature d’une information incorrecte. Dans leur article sur les LLMs Gras et Varoquaux notent de manière un peu surprenante “[qu’]une connaissance est une croyance qui a été générée par un processus fiable“. Jordan, dans le second papier cité, décrit les IAs comme un marché “bottom-up self-organizazation will be the dominant paradigm for growth of learning-catalyzed markets. But such growth need not be uncontrolled or outside our comprenhension.” (L’auto-organisation ascendante deviendra le paradigme dominant de la croissance des marchés catalysés par l’apprentissage. Mais cette croissance ne doit pas nécessairement être incontrôlée ni hors de notre compréhension.)

Collective and individual again: David Graeber, Cynthia Fleury

Leave a reply

About ten years ago I discovered the work and thought of Cynthia Fleury ici.

MesLivres-Cynthia-Fleury

About six years ago, I translated, without really having the right to do so, an interview that I entitled « To be Brave is sometimes to Endure, sometimes to Break up »

About three years ago I discovered apodcast on Radio France by Cynthia Fleury about Vladimir Jankélévitch. All of this is somewhere on my blog.

This is in part why I’m so thrilled of my privilege to be at the @letheatredelatelier last Friday to hear two great actresses, Isabelle Adjani and Laure Calamy, in the adaptation of @lafinducourage by @fleuryperkinsc.

And here’s what I have noted: “Eight actresses, six duos for the face-to-face encounter between an author and a journalist in an engaged and stimulating dialogue that acutely and self-deprecatingly questions what it means to ‘hold on’ in a world under tension. A diversity of voices and perspectives, reflecting the cherished idea of ​​philosopher Cynthia Fleury, according to which ‘courage is not proclaimed: it is rebuilt, together.'”

But I can’t stop there this morning: Thanks to a post LinkedIn by the excellent Olivier Alexandre, often mentioned here, I discovered a fascinating exchange between Peter Thiel, who needs no introduction, and David Graeber, whom everyone should know. Unfortunately, I’ve quoted Thiel more often than Graeber. As I commented on the post, “This fascinating exchange reveals a pessimistic view of humankind that trusts only some “Happy Few”, leading to authoritarianism and illiberalism, and an optimistic view that trusts and advocates decentralization and democracy. I want to believe that it’s not just a matter of belief, even though social sciences, and science in general, are not purely mechanical, so unpredictability persists despite science.”

The World of Startups according to Marion Flécher – final post : the sociology, & not coolness

Leave a reply

I just finished reading *The World of Startups* by Marion Flécher, and I can confirm that it’s an excellent book, even if it’s a bit depressing at times — I’ll come back to that. In a previous post, I described her comparative work between French and “Silicon” startups. But the core of her research deals, on the one hand, with a sociology of entrepreneurs that seems to shatter the myth of the self-made man, and on the other hand, with the internal workings of startups, which are far removed from the sweet and colorful “coolness” of these so-called liberated companies.

A sociology of entrepreneurs

Marion Flécher criticizes the idea that entrepreneurs owe their success solely to their own merit [page 83]. She illustrates this point with the education level of this population. 92% hold a Master’s or Doctoral degree (compared to 27% of traditional business founders). She also illustrates this point by noting that a large majority held salaried positions before launching their businesses [page 87].

Furthermore, all these entrepreneurs benefited from a state-supported ecosystem, notably through the numerous initiatives of BPIFrance (for training, networking, and access to subsidies, without which only those with significant financial resources would launch their businesses). This analysis, reminiscent of Bourdieu, demonstrates that entrepreneurs possess economic, social, and cultural capital.

She convincingly returns to an analogy with the art world: just as an artist doesn’t create a work of art alone in their studio, startup founders don’t create their companies in isolation [page 97].

As a result, the social inequalities of Western societies are amplified here. Gender discrimination, of course: Manon Flécher uses the expression “Entrepreneurship + Technological Innovation = Sexism Squared” [page 109]. But also geographical discrimination: the startup world is very urban and bourgeois. Entrepreneurs from the suburbs, lacking the necessary codes, information, and networks, are rare.

Ultimately, nothing is really new. I experienced similar things in my school and professional life (at a time when social mobility was a reality, a more optimistic and enthusiastic era), and indeed, technological entrepreneurship comes at the end of the path of scientific training and management career.

Entry modalities for entrepreneurship

Her multiple correspondence analysis of startup founders resonated strongly with my own experience in this world. She classifies them into three groups: independent or career entrepreneurs, salaried entrepreneurs or entrepreneurs by opportunity, and young startup founders or novice entrepreneurs [pages 128-131].

The book is essential reading for a nuanced description of this population. Some individuals “seek to rediscover meaning and autonomy; a logic of seeking prestige and social distinction; and a strategic logic that leads them to seize (or not) the opportunities that arise” [page 133]. Others create based on a novel idea with which they hope to “change the world,” which is reminiscent of the Schumpeterian entrepreneur [page 136].

A long personal note, or rather a few notes, at this point in my synthesi of the book. For years I’ve been studying entrepreneurs in my own way. I’ve built a very personal sociology of them, as scientific as possible:
1- It’s undoubtedly an elitist world, even more so if you focus on tech startups. It’s difficult to launch a business without a solid education, often acquired through a doctorate, and self-taught entrepreneurs are very rare. (However, we shouldn’t forget this group: school dropouts who decided to interrupt their studies to embark on this adventure, and Steve Jobs, Mark Zuckerberg, and Dylan Field are a few examples. But be aware, they would undoubtedly have had brilliant academic careers at the best universities had it not been for this interruption.) In fact, my main study on the subject concerns entrepreneurs from Stanford University. Could it get any more elitist?
2- I’ve finally discovered a categorization of entrepreneurs that groups together, on the one hand, novices under 30 and, on the other, more seasoned entrepreneurs aged 30 to 50. It’s quite rare to find such an analysis, and I’ve too often read that entrepreneurs are generally experienced, with an average age of 39. This reminds me of my analysis of “serial entrepreneurs” and, on the other hand, the age of entrepreneurs. Whereas Marion Flécher seems to show (and I hope I’m not mistaken) that in France, experience is an advantage, I’ve tried to demonstrate that value creation is correlated with the inexperience of the founders (who, of course, are not alone as they progress through their ventures).
3- Societal discrimination is a crucial issue. I love mentioning the Lost Einsteins, the Marie Curies lost in Morbihan. I don’t have many more solutions than others to remedy this; I can only observe.

Fortunes and misfortunes in the world of startups

I now come to what I consider the most depressing part of the book. Few innovative companies manage to become sustainable — according to some studies, 90% of startups end up closing or filing for bankruptcy before their tenth year — and even fewer reach the expected growth level to become “unicorns”: only 1% of startups created in the United States achieve this, and only 25 of the 15,000 created in France, or 0.1% [page 147].

The author’s descriptions of fundraising (page 155), growing startups (page 182), and even startups in their early stages (page 204) are sometimes chilling. It’s difficult to deny these realities, even if they are not the only ones. It contains expressions such as fundraising as a test of strength (page 160), co-optation logics unfavorable to women (page 165), making a virtue of misfortune, a strategy of dominant people (page 172), from well-being to control: when moments of conviviality lead to over-investment of workers (page 188), the creation phase: a “joyful mess” not always so joyful (page 205), interns abandoned in the face of demanding work (page 207), a work environment not conducive to the emergence of collective mobilizations (page 228) so that the choice is reduced to “leave or stay” (page 232).

Marion Flécher also notes general advice from the ecosystem that has always struck me as wrong, not to say toxic, at the risk of diminishing the quality and scale of potential successes:
– “In the startup world, failure is both a common and symbolically valued phenomenon” [page 172]; however, I encourage you to read a different perspective from the founder of Zendesk, who doesn’t celebrate failure at all;
– “The ability to pivot is a condition for the success of entrepreneurial projects” [page 173];
– “Startup founders are advised to build teams with complementary profiles; above all, you mustn’t duplicate skills” [page 99]; here again, a different vision of entrepreneurship from Charles Geschke, co-founder of Adobe.

The conclusion: Start-ups, the new face of capitalism

The conclusion is also a bit depressing, so I’ll add a few more personal notes!

Startups are a unique model of companies due to at least three fundamental characteristics (pages 237-243):
– economic, which is not geared towards profitability but towards strong and rapid growth through innovation, made possible by external funding;
– organizational, which utilizes horizontal structures, cooperation, autonomy, and well-being;
– ideological, which values ​​entrepreneurship as the main driver of economic and social progress and merit as the ultimate principle of success and social justice.

But the author’s ambition was to deconstruct these ideals and promises by confronting them with reality. Deconstructing
– the myth of risk and merit, which serves to justify the enrichment of a few by obscuring the decisive role of the State
– the meritocratic myth of the self-made man, which renders inequalities of access invisible
– the myth of the liberated company, which in reality perpetuates, in renewed forms, logics of exploitation, control, and segmentation of the workforce.

More personal comments: the power law and the exceptional nature of this world?

The analysis is correct, but isn’t the picture painted too bleakly? Everyone should form their own opinion, and the facts should help. I’m even surprised that Marion Flécher didn’t cite Mariana Mazzucato and her book, *The Entrepreneurial State*.

The debate surrounding this undoubtedly abnormal world, like the art world for that matter, is not surprising. It’s full of exceptions rather than averages, to the point that some believe Gaussian statistics don’t apply. Power law should be used instead.

We’re closer to a (more or less absolute) monarchy than a democracy. Founders are practically kings. Additional note: in my data on nearly 1,000 startups (961 to be precise, as of today), 60% of CEOs are founders, and even 70% in the software and internet sectors. And only 40 CEOs are women… (and even worse, 88 female founders out of 1,733 male founders).

In fact, neither entrepreneurs nor investors behave in a completely conventional way. Take, for example, the founders of Apple: The two of them did not make a good impression on people. They were bearded. They didn’t smell good. They dressed funny. Young, naive. But Woz had designed a really wonderful, wonderful computer. […] And I came to the conclusion that we could build a Fortune 500 company in less than five years. I said I’d put up the money that was needed.

This is an important book about startups. It’s quite rare in French, and anyone interested in the subject should read *Le Monde des Start-up*.

The World of Startups according to Marion Flécher (Part 2)

Leave a reply

In my post of November 8th, I promised to “read Marion Flécher’s book with great interest and will write to you in a future post to see if I find cause for resignation or optimism regarding the startup world.” I’m going to say both! The book is indeed excellent and perfectly describes the differences between France (which tried to claim to be a Startup Nation) and Silicon Valley (which never felt the need for such a claim).

The Difficulty of Defining the Word “Startup”

The heart of the book is not the comparison between the two regions, but rather what the French startup scene is like; I will return to this point. In her introduction, she explains the difficulty in defining the word, to the point of writing: “Making startups my subject of study was not a given. […] I was encouraged to use alternative terms such as innovative companies, technology companies, or high-growth companies” [page 16]. With a striking analogy, she adds, “as with the art world, in which players spend their time trying to determine what is art and what is not, it is by observing how a world makes these distinctions, and not by trying to make them ourselves, that we begin to understand what is happening in that world” [page 20]. Without naming Steve Blank, through a survey of entrepreneurs who gave her a multitude of definitions, she almost mentions my favorite definition: “a temporary organization in search of a repeatable and scalable business model”.

Silicon Valley, Heart of Startups

In her first chapter, Marion Flécher describes Silicon Valley, the region that gave rise to the semiconductor, the microprocessor, the microcomputer, software, the internet, social networks (and ultimately artificial intelligence, which had not yet emerged when she conducted her research). She demonstrates that the region was at the heart of a revolution that goes beyond technological innovation. This innovation was “accompanied by organizational and managerial innovations that led to a profound ideological redefinition of the entreprise” [page 52].

Yet, she shows that pinpointing the moment of their emergence is just as difficult as defining startups. “For many historians and ethnologists specializing in Silicon Valley, it is Intel’s invention of the microprocessor that constitutes the true starting point of the region’s technological boom and rise to power” [page 39]. She does not, however, neglect to note the importance of Hewlett-Packard (founded in 1939) and Fairchild (founded in 1957) in this dual revolution, including the development of project-based work in small teams, the upheaval of dress codes, and the implementation of new incentive structures [pages 52-53]. The complexity of the region’s origins also stems from the existence of other significant influences such as free software and the “Do It Yourself” culture [pages 55-56], and from a diversity of major players including venture capital funds, the federal government, and the companies themselves. A world, as the author describes it, an ecosystem.

Another difficulty addressed by Marion Flécher, at least for a Silicon Valley enthusiast like myself, is: when did the word “startup” first appear? “The term ‘startup’ seems to have been used for the first time in 1976, in a Forbes magazine article, to refer to young technology companies in Silicon Valley” [Author’s note: The Unfashionable Business of Investing in Startups in the Electronic Data Processing Field]. “The term ‘startup company’, which associates the startup with a specific type of business, appeared a year later in an article entitled ‘An Incubator for Startup Companies, Especially in the Fast-growth, High Technology Fields,’ published in Business Week in 1977.” The author adds that the term acquired its current meaning and spread worldwide during the 1990s, even though this new business model can be traced back to the 1940s. NB: I confirm this in a blog post: When was the word “start-up” first used?


Scan of Figure 2 [page 43]: Timeline of the main Silicon Valley technology companies. I circled two things by hand while reading. My surprise at seeing only one founder for eBay; I thought Jeff Skoll was a co-founder, but apparently he may have only been the first employee. And my other surprise at seeing three co-founders for Apple, which is rarely mentioned. Ronald Wayne is often forgotten. And then a rather unimportant note for Marion Flécher: Wo[lk]zniak is misspelled on page 41!

Risk and Uncertainty

In a brief but equally excellent section on venture capital, Marion Flécher explains that “unlike risk, which refers to a situation that can be predicted by probability and in which actors can reason rationally […] uncertainty refers to a situation in which the degree of singularity is such that it cannot be compared to any other. By developing disruptive innovations, Silicon Valley entrepreneurs create situations of radical uncertainty in which actors can only make speculative judgments” [page 46]. This explains why the term [ad]venture capital is very different from the term capital-risque in France (which is undoubtedly indicative of dissimilar worlds). “Nevertheless, these players have resources that allow them to transform the uncertainty of a situation into a quantifiable risk. Their activity requires, first and foremost, a sound knowledge of the technical environment, which enables them to assess the growth prospects of projects. Most investors […] are thus often former engineers or entrepreneurs” [page 47]. This is undoubtedly another major difference between Silicon Valley and France.

Since I mentioned personal surprises in the commentary on the figure above, I take this opportunity to add a few more personal comments (for myself, the author, and any potential reader!):
– Without a doubt, the startup world is a new illustration of capitalism, and this has undoubtedly been misunderstood. Startups have never been liberated companies; trade unions are very rare, not to say unwelcome. I already mentioned this point in my first post.
– Marion Flécher emphasizes intellectual property (Microsoft’s proprietary software, Intel’s microprocessor patent), which creates near-monopolies favored by a state that “seems to have implicitly supported market concentration” [page 49]. Yet it was the state that forced Bell Labs to grant licenses for the transistor, for which the company held the patent. Intel certainly had a near-monopoly on the microprocessor, even though IBM and AMD were genuine competitors. But competition in new sectors has made Intel a declining player in recent years (telecommunications, artificial intelligence). I would say that the US government is protecting the country at a macroeconomic level by defending its technological lead rather than protecting any individual company. OpenAI might replace Google, which could have replaced Microsoft, just as Nvidia or even AMD might replace Intel. The same goes for mobile phones. The US remains the leader.
– Another minor point of doubt: “Between 1998 and 1999, venture capital almost doubled, going from $3.2 billion to $6.1 billion” [page 48]. I have the impression, and I could be wrong, that the amounts were about ten times larger and that these figures correspond more closely to the 1980s.
– Finally, I see confirmation of a personal impression regarding the decrease in the number of IPOs: Silicon Valley had 417 IPOs in 2000 compared to only 14 in 2021 [page 51]. Indeed, for years I’ve been compiling market capitalization tables and dreamed of quickly reaching 1,000, but the IPO drought is slowing my ambition… On the other hand, M&A acquisitions still seem as prosperous as ever, since Marion Flécher mentions more than 90 acquisitions by Facebook since its creation (see my posts on Cisco and Google). A startup may not be destined to become a long-term business, but it’s also possible that the monopolistic concentration mentioned above is at its peak…

France, a nation of startups?

That’s the title of Chapter 2. And the second page of this chapter includes the following figure. It’s easy to see that the French press began to take an interest in the subject during the dot-com bubble and then again from 2012 onwards. Manon Flécher explains that this second period is significant, coinciding with the arrival of Uber and Airbnb in France, but also with the creation of BPIFrance and the French Tech initiative. Equally interesting, the author reminds us that General de Gaulle visited San Francisco in 1960, Georges Pompidou 10 years later, and François Mitterrand in 1984. The author doesn’t mention the creation of Sophia Antipolis in 1969, which Paul Graham mocks, more or less gently (see my post from 2011). Presidents Hollande and then Macron have been apparently much more involved in the topic.

Is France a startup nation? The answer is clear if you’ve read the preceding paragraphs. But the debate runs deeper, as I indicated in *Politics vs. Economics: A Country Is Not a Startup*, translating the article Non, la France ne doit pas devenir une start-up. I didn’t know, or had forgotten, that Emmanuel Macron had then used the term “hyper-innovation.” But the doomsayers are always ignored, and hyper-communication too often trumps reality and facts.

Marion Flécher also addresses this by stating that “despite this growth, the State remains the primary funder of startups in France” [page 71]. Her footnote on page 69 is revealing. “In 2015, business angels reportedly invested a total of €41 million, which was still half the amount in the United Kingdom and 2.5 times less than in Germany. […] In 2023, the United Kingdom continued to lead other European countries with €307.4 million invested by business angels, compared to €198.5 million for Germany and €142.5 million for France.” BPIFrance represents two billion euros in direct investments in company capital [page 72].

My post is already too long, which is convenient since I’m at this point in my reading of Le Monde des startups. Yet I haven’t even started on the main topic: the sociology of startup founders. A sequel coming soon!

Post-scriptum : On a related note, I just bought *Palo Alto: A History of California, Capitalism, and the World*, one review of which says, The most comprehensive — and incendiary — history of the place that we’re ever likely to get. A sweeping and unsparing critique, it’s also well written, frequently surprising and, because history tends to rhyme, increasingly urgent. You may never think about Stanford, iconic tech companies like Hewlett Packard or, indeed, the Valley itself the same way again. I won’t.” LOS ANGELES TIMES

The introduction already haunts me: the author promises to explain that the inhabitants of California, Silicon Valley and Palo Alto have not all forgotten the ghosts that surround them and without which the region could not have become what it is… To be continued!

The World of Startups by Marion Flécher

Leave a reply

The startup world has been the core of my professional life for almost 30 years. I know it perhaps too well, and I also read almost everything I can get my hands on. So I feel compelled to read a book that takes this subject as its title. Marion Flécher is a lecturer in sociology at the University of Paris Nanterre, and her book, *The World of Startups* (Le monde des startup) is based on her doctoral thesis, entitled *The Startup World: The New Face of Capitalism? An Investigation into the Modes of Creation and Organization of Startups in France and the United States*. (Le monde des start-up, le nouveau visage du capitalisme ? Enquête sur les modes de création et d’organisation des start-up en France et aux Etats-Unis).

Marion Flécher just post an article on LinkedIn pand she was alos the guest of Guillaume Erner yesterday morning on France Culture.

As you may have guessed, I haven’t read the book yet, but I will as soon as I receive it. I’m always a bit ambivalent when I read critical analyses of this world, and of Silicon Valley in particular. Here’s what the publisher says on the book’s page: “The book compares the promises made by startups (success stories, meritocracy, flexibility) with the reality on the ground (failure rates, mass layoffs, pressure to perform, burnout). For the past fifteen years or so, young, innovative companies grouped under the term “startups” have occupied an increasingly prominent place in the media and political arena. Symbols of modernity and Capitalism 2.0, they are being held up as a genuine economic and organizational model. How can we understand this enthusiasm? (…) Dismissing the mythical figure of the self-made man, she sheds light on the reality of this social world and defines the profile of those, and more rarely those women, who launch startups and claim this label. While they claim to break with the hierarchical rigidity of traditional firms, startups offer a new face of capitalism, one whose gentler and more colorful aspects have allowed it to recover from its critics. In the digital age and the era of new technologies, what do they foreshadow for the future of work?”

I am ambivalent for two reasons :

– The first reason is that it seems to me France has gone from a profound ignorance of this world to a critique of its excesses that is undoubtedly justified but harsh.

About ten years ago, Professor Libero Zuppiroli brilliantly criticized the unfulfilled promises of innovation in *The Utopias of the 21st Century*. A few years earlier, he had teased me about the fact that startups didn’t represent much at all. That was the subject of another post.

He was right, at least for Europe, to the point that neither France nor Europe has generated any real success stories. My previous post last week celebrated an EPFL’s success story at $3 billion, and Criteo, Mistral, and others are all modest in size compared to the behemoths that the GAFAMs have become. SAP, Spotify, ARM, and only a few others in Europe can be compared to American successes. I am therefore sometimes torn between comparing and criticizing two continents that have little in common. France has moved very quickly, not to say too quickly, from a lack of understanding to a criticism of this world without grasping the reasons for its appeal and success.

– The second reason is that this criticism has existed for a long time, and I’m saddened that communication takes precedence over information. Startups have never been a paradise; they are rarely liberated companies, so storytelling has transformed a few exceptions into a universal model.

For example in 1986 in Startup fever, one could read “The Silicon Valley has been called “one of the last great bastions of male dominance” by the local Peninsula Times Tribune. […] They are under-represented in management and administration. Few women have technical or engineering backgrounds. […] Why there are few women in positioning of responsability in Silicon Valley is complex and puzzling. Until recently, the overwhelming majority of engineering graduates were men. […] Scientific and engineering professionals in the finance community and in start-ups are likely to be men: these power-brokers rely exclusively on their personal networks. […] Twenty of the largest publicly held Silicon Valley firms listed a total of 209 persons as corporate officers in 1980; only 4 were women. The board of directors of these 20 firms include 150 directors. Only one was a woman: Shirley Hufstedler, serving on the board of Hewlett-Packard.” But the authors are optimistic: they explain that any woman with a technical background or an interest in high-tech has opportunities: “A Martian with three heads could find a job in Silicon Valley. So for women with a technical background, it’s terrific. […] An exception to masculine dominance is Sandy Kurtzig. “I wanted to start in a garage like HP, but I didn’t have one. So I started in a second bedroom of my apartment.” At first, Kurtzig did sales, bookkeeping and management of her start-up. As long as she had only five or six employees, they worked out of her apartment. It went into rapid growth and had annual sales in 1982.”

And what about Anna-Lee Saxenian in 1999 : “In 1979, I was a graduate student at Berkeley and I was one of the first scholars to study Silicon Valley. I culminated my master’s program by writing a thesis in which I confidently predicted that Silicon Valley would stop growing. I argued that housing and labor were too expensive and the roads were too congested, and while corporate headquarters and research might remain, I was convinced that the region had reached its physical limits and that innovation and job growth would occur elsewhere during the 1980s. As it turns out I was wrong.” Returning to the earlier statement (France has moved very quickly, not to say too quickly, from a lack of understanding to a criticism of this world without grasping the reasons for its appeal and success), I believe it is essential to read and reread the remarkable works of Professor Saxenian (Regional Advantage, The New Argonauts). She has been able to explain the reasons for these attractions and successes, which persist today despite the excesses.

I’ve arrived at these compromises, which the great Bernard Stiegler helped me develop:
– In a 2016 post, “Has the World Gone Mad? Perhaps…,” I wrote that Stiegler’s main thesis is that capitalism has gone mad and that the absence of regulation and controls can lead to madness. “Disruption” can be beneficial when followed by a period of stabilization.
– That same year, in “Disruption and Madness According to Bernard Stiegler”, I noted that in ancient Greece, the term Pharmakon referred to both remedy, poison, and scapegoat. Medicine becomes harmful if consumed in excess…

I will therefore read Marion Flécher’s book with great interest and will write to you in a future post to see if I find cause for resignation or optimism regarding the startup world.

Post-scriptum: I regularly wonder about the nature of French society’s relationship with science and technology. All the pitfalls of technology are clear, and I regularly mention them here; for example, I’m thinking of François Jarrige’s *Technocritiques*. I’m not sure that the criticism is as severe for science or mathematics, as if they were more neutral. We hear less often about the fact that very few women have been Fields Medalists: two female laureates in 2014 and 2022, compared to 80 laureates in total since 1936.

Science in the media is neglected in France, generalist journalists and even French intellectuals have a rather deplorable scientific culture. Science doesn’t attract many girls, or at least they are discouraged from pursuing it for complex reasons. This wasn’t the case in the former Eastern Bloc countries (whatever the reasons may be). The relationship with technology, and therefore with innovation and startups, is perhaps a consequence of this, or simply a correlation. This might explain some of my sadness or frustration, even though I haven’t lost my enthusiasm. And in those moments, Churchill’s quote comes back to me: “Success is going from failure to failure without losing your enthusiasm.”

Two references in French.

– Mathématiques : deux femmes récompensées depuis 1936 : https://www.lemonde.fr/les-decodeurs/article/2019/03/20/mathematiques-deux-femmes-recompensees-depuis-1936_5438858_4355770.html

– Maryna Viazovska, deuxième femme lauréate après Maryam Mirzakhani : https://information.tv5monde.com/terriennes/medaille-fields-maryna-viazovska-deuxieme-femme-laureate-apres-maryam-mirzakhani-2933

Nexthink, a Lausanne-based startup, acquired for 3 billion dollars

Leave a reply

This doesn’t happen that often in a lifetime [1], so it deserves a pause: Nexthink, a spin-off from EPFL, has just been acquired for $3 billion [2] — you read that right, $3 billion!! Admittedly, Nexthink isn’t exactly young anymore; it was co-founded in September 2004, by Pedro Bados, who came (I believe, to do a master’s internship as part of the Erasmus program) in the artificial intelligence lab. AI wasn’t what it is today. His project was based on Bayesian methods… As luck would have it, I met Pedro then to analyze the intellectual property he had generated and what could be done with it. Pedro wanted to sell the patent that EPFL had filed, and when he discovered it wasn’t generating as much interest as he had imagined, I suggested, “Why not a startup?” That became my main role. I also helped him structure the initial deal, I mean who he would work with, within EPFL and outside, and made some introductions at the time. The rest is history !

Pedro has become a discreet but essential figure in the Swiss ecosystem [3]. He was our guest for ventureideas as early as 2006, along with Jordi Montserrat.

Here are notes taken at the time from his présentation (pdf).
NEXThink notes

He also gave a beautiful interview ten years ago to the newspaper Le Temps (in French): Nobody is ready to be an entrepreneur

I’ve had some wonderful experiences in my professional life, in scientific research first between Paris and Stanford University, then in venture capital with Index Ventures, finally at EPFL for 15 years, and for the last 6 years at Inria. Nexthink will certainly remain one of the best ones, and I hope to experience a few more. As I said 15 years ago [4], my professional passion is to encourage these kinds of ventures (see the Stanford interview): “A love of entrepreneurship, a passion that I took back home to Europe after studying here [at Stanford University]. I want to see a stronger entrepreneurial culture there, and I am working in more than one way to effect that change.”

A few notes:

[1] I was struck some time ago by this post about Index Ventures and its stratospheric performance :

8 startup with an exit above $10B : Figma – $59B, Revolut – $75B, Adyen – $44B, Robinhood – $82B, Scale – $14.9B, Wiz – $32B, Datadog – $46B, Roblox – $86B. So what does $3B represent for Index? And I don’t have a list of their exits above a billion. I remember Virata, Numerical Technologies, The Fantastic Corporation, and Skype before 2005. I might ask them!

[2] Some more news and archives – What I found online about Nexthink acquisition, mostly LinkedIn posts by its founders and investors as well as press articles
Nexthink News autour acquisition


Neil Rimer (Index ventures) and Pedro Bados I am not sure where and when

[3] A short presentation of the EPFL ecosystem to its alumni before 2010 that mentions Nexthink (pdf)

a3angels10-herve_lebert-innogrants

[4] Thanks to my favorite search engine, an exchange about what I thought of startups, Silicon Valley in 2008 where I mention Nexthink. I am not sure things have changed that much…

Stanford HL

Why is Silicon Valley Silicon Valley ? A debate initiated by AnnaLee Saxenian

Leave a reply

I should go back to Silicon Valley to ask AnnaLee Saxenian what she thinks of Silicon Valley today and, more importantly, if her analysis is still valid today (which I think it is !)

I fall by accident on a short 6-page article that she wrote in 1999 entitled Comment on Kenney and von Burg,‘Technology, Entrepreneurship and Path Dependence: Industrial Clustering in Silicon Valley and Route 128’ that you can easily find on the net, for example on Researchgate.

Why do I care about a 25-year old article about innovation ? Because I often have the feeling Europe and France in particular miss the point about what matters. Saxenian’s argument can be summarized with an extract from that article: It is precisely the openness, multiplicity and diversity of interconnections in Silicon Valley that allows economic actors to continually scan the environment for new opportunities and to invest in novel technologies, markets and applications with unprecedented speed. The autarkic structure of firms and institutions in Route 128, by contrast, have historically discouraged precisely the decentralized flows of information, skill and capital that encourage such technological experimentation. [Page 5]

The analysis is subtle because one might get the impression that Europe has understood the role of collaboration, but I fear that it has remained institutional and not informal. Has there ever been a Wagon Wheel Bar in Europe?

Again the article is short and worth reading. The debate is also about the role of big firms in technology clusters. The figure which follows is a perfect illustration:

Source: J. Zhang (2003) compiled data from VentureOne based on VC-backed spin-offs from 1992 to 2001

Zhang had worked with Saxenian and the table shows that from an academic standpoint, entrepreneurship was as strong at MIT/Harvard as Stanford/Berkeley. But if we just look at IBM West vs. IBM East, the difference is striking. “Individuals in Silicon Valley regularly leave jobs at established firms to join start-ups, and vice versa, all the while maintaining networks of cross-cutting personal and professional relationships. Nor is it uncommon for individual engineers working at large firms to make angel investments in promising start-ups — or for established firms to organize venture divisions to support new ventures in related industries.” [Page 4]

I have been using it since I discovered it years ago, particularly when I was giving classes about startup and Silicon Valley. With a little nostalgia, I put my introductory pdf below where I emphasized on the role of culture in entrepreneurship.

eLab01 - SiliconValley

Coreweave, another IPO, more IPOs and the link to VC

Leave a reply

I should thank a colleague of mine for mentioning the surprising not to say crazy IPO of Coreweave. A new cap. table in my list which has now 948 companies. I will update my stats in the final pages of the pdf (available at the end of the post) when I reach 1’000 companies except I have been struggling with IPOs in the recent years. I will expand on this after the Coreweave case.

It’s really a strange story that I did not know and discovered on Wikipedia. 3 founders who were commodity traders launched a crypto currency mining company with obviously a data center and tons of GPUs. They pivoted in 2019 to providing cloud computing in a very aggressive manner. There is a 4th cofounder, the CTO but he apparently joined with the pivot (and mostly own stock options). A crazy growth in revenues (and losses) as the quaterly results show in the next table. Apparently there is cautiousness becuase Microsoft represents 60% of the revenues and the debt is alos about $800M. Impressibe but maybe fragile…

Now the IPOs !

The figure above represents the number of companies in my pdf by year of exit (mostly IPO filings, and much more rarely M&As). [Of course I do not have all IPOs and I found a document which shows I have 27% and 87% of the IPOs of years 2011-22.]

I had forgotten there was such a bubble and crash in 2021 so I had a look at VC investments over decades which gives the following figure. But with so few IPOs, LPs will become more and more impatient putting pressure on VCs all the more money is not as cheap as before…

Now when I hear that VC is in a crisis and I look at the levels of investments in the 7 years compared to the previous decades, I cannot help btu think there is a lot of money available…

Finally the promised pdf, but at this rate it will take me another two years before I release new statistics : Equity in start-ups – Historical data from 900+ companies.

20250717 Equity List Lebret

David Graeber, The Dawn of Everything & Bullshit Jobs

Leave a reply

This post has nothing to do with the startup world. David Graeber was an American anthropologist, author of Bullshit Jobs and co-author of The Dawn of Everything, two books as astonishing as they are stimulating.

The author is “one of the most influential intellectuals in the Anglo-Saxon world.” He can be described as an anarchist, very left-leaning, with a great distrust of centralized institutions and politics. I will simply copy the summary of these two books, which are among the most fascinating I have read in recent years.

A dramatically new understanding of human history, challenging our most fundamental assumptions about social evolution—from the development of agriculture and cities to the origins of the state, democracy, and inequality—and revealing new possibilities for human emancipation.
For generations, our remote ancestors have been cast as primitive and childlike—either free and equal innocents, or thuggish and warlike. Civilization, we are told, could be achieved only by sacrificing those original freedoms or, alternatively, by taming our baser instincts. David Graeber and David Wengrow show how such theories first emerged in the eighteenth century as a conservative reaction to powerful critiques of European society posed by Indigenous observers and intellectuals. Revisiting this encounter has startling implications for how we make sense of human history today, including the origins of farming, property, cities, democracy, slavery, and civilization itself.
Drawing on pathbreaking research in archaeology and anthropology, the authors show how history becomes a far more interesting place once we learn to throw off our conceptual shackles and perceive what’s really there. If humans did not spend 95 percent of their evolutionary past in tiny bands of hunter-gatherers, what were they doing all that time? If agriculture, and cities, did not mean a plunge into hierarchy and domination, then what kinds of social and economic organization did they lead to? The answers are often unexpected, and suggest that the course of human history may be less set in stone, and more full of playful, hopeful possibilities, than we tend to assume.
The Dawn of Everything fundamentally transforms our understanding of the human past and offers a path toward imagining new forms of freedom, new ways of organizing society. This is a monumental book of formidable intellectual range, animated by curiosity, moral vision, and a faith in the power of direct action.
From the publisher website.

Does your job make a meaningful contribution to the world? In the spring of 2013, David Graeber asked this question in a playful, provocative essay titled “On the Phenomenon of Bullshit Jobs.” It went viral. After one million online views in seventeen different languages, people all over the world are still debating the answer.
There are hordes of people—HR consultants, communication coordinators, telemarketing researchers, corporate lawyers—whose jobs are useless, and, tragically, they know it. These people are caught in bullshit jobs.
Graeber explores one of society’s most vexing and deeply felt concerns, indicting among other villains a particular strain of finance capitalism that betrays ideals shared by thinkers ranging from Keynes to Lincoln. “Clever and charismatic” (The New Yorker), Bullshit Jobs gives individuals, corporations, and societies permission to undergo a shift in values, placing creative and caring work at the center of our culture. This book is for everyone who wants to turn their vocation back into an avocation and “a thought-provoking examination of our working lives” (Financial Times).
From the publisher webiste.