Is Twitter really worth 44 billion dollars ? Nobody knows. But for some background and history, Google acquired Youtube for 1.65 billion dollars in 2005. In 2021 and for 20 billion Microsoft bought the company “Nuance”, specialized in conversational artificial intelligence. In 2016, Microsoft again spent 26 billion dollars to buy LinkedIn. In 2014 Facebook bought WhatsApp for 22 billion dollars, half the amount that Musk is investing today to acquire Twitter. And in 2012 the same Facebook bought Instagram for $1 billion. The company is now worth 100 times more. All of the companies targeted by these acquisitions have a history. And the biggest offers don’t always make the best stories.
Why Twitter ? Considering the companies that Musk runs today, why acquire Twitter, which is as far from the aerospace sector (SpaceX) as it is from the telecommunications sector (Starlink), public works (Boring Compagny builds tunnels to solve the problem of urban traffic jams), the automobile sector (Tesla) or neural interfaces (Neuralink)? Let’s formulate some hypotheses.
Maybe because Twitter is the best PR agency on the planet and no big boss would give up the chance to get exclusive use of it. Maybe because in the great contest of dominant toxic males in the “tech” sector, it was unbearable for Musk that Zuckerberg owns Facebook (and WhatsApp and Instagram), that Bezos owns the Washington Post and that he only owns a fleet of cars and rockets. Maybe the man who tweeted compulsively found it unbearable to be able to express himself so much and with so much enjoyment on a medium that was not his own. Perhaps he is making his transition to what Mc Kenzie Wark calls the “vectorialist class” :
Here, in the overdeveloped world, the bourgeoisie is dead. It has ceased to rule and govern. Power is in the hands of what I have called the vectorialist class. Whereas the old ruling class controlled the means of production, the new ruling class has a limited interest in the material conditions of production, in mines, blast furnaces and assembly lines. Its power is not based on the ownership of these things, but on the control of logistics, on the way they are managed. Vector power has two aspects, intensive and extensive. The intensive vector is computational power. It is the power to model and simulate. It is the power to monitor and calculate. And it is also the power to play with information, to transform it into narrative and poetry. Extensive vector is the power to move information from one place to another. It is the power to move and combine anything with anything else as a resource. Again, this power has not only a rational aspect, but also a poetic one.” Wark McKenzie, Degoutin Christophe, “New Strategies of the Vectorialist Class” Multitudes, 2013/3 (No. 54), pp. 191-198.
Perhaps because he simply can afford all of his whims and said whims can be read as those of the richest man on the planet who describes himself as having an autism spectrum disorder (he describes himself as Asperger’s). And that even while remaining free of any form of countertop psychologization, the story of Musk’s behaviors and decisions must be read with these two prisms in mind. Perhaps finally, because there are strategic elements in the conversational algorithmic engineering of Twitter and in the volume of data that circulates there, for the deployment and optimization of artificial intelligence technologies that Elon Musk has made one of his priorities and to which we will return. But before that, one more question. A fundamental one.
What did he buy with 44 billion dollars ? 44 billion dollars is a considerable price for a private platform, but it is a derisory price for a digital public space that today occupies the place of an information “commons”, or at least that of a “common” informational and conversational infrastructure. I say “common” keeping in mind that social networks are also defined by the number of people who are not on them, and that Twitter has “only” 436 million active monthly users. But Twitter’s influence, at least in Europe and the US, extends far beyond its user base.
For 44 billion dollars, Elon Musk offered himself (at least) three things. First, a user base. Second, an absolutely colossal amount of data (and metadata …). And finally, he bought what is probably the largest direct conversational database that has ever existed: in one place and one place only, behind one web address, www.twitter.com, there are 6,000 tweets per second, 350,000 per minute, 500 million per day, 200 billion per year.
What can he do with such a user base? Monetize it, through advertising.
What can he do with such a large volume of data and metadata? Exploit it to refine the algorithmic selection and recommendation routines that boost the platform’s interactions.
And what can he do with these hundreds of billions of tweets, which are at least tens of billions of conversations? Conversations on all subjects, in all tones, in all linguistic, dialectical, and narrative registers, between two or three people as well as between dozens or sometimes hundreds, and in connection with all the world’s news, from the most tragic to the most inessential? Before answering this question, I want to tell you a story.
The Google Books hypothesis. The scene takes place in 2005. Google, which had already become the ultimate search engine, did what seemed to everyone to be a “madness”. It announced a massive program of digitization of public domain works, which it would then make available for free on its search engine and of which it would also provide a digital copy to partner libraries. This digitization of an unprecedented scale represents a very important cost and an investment at a loss. And everyone is asking: why is Google doing this ? Why is the most powerful search engine on the planet embarking on this project ? Of course, it gains a little in terms of image and notoriety, but at the time, it had no need for it. We will discover later that the stake was to settle on the market of the online sale of books under rights and that the digitization of works in the public domain was as such only the technique of the foot in the door. But the other reason for this apparent madness, the real reason, is that Google needed to “train” its linguistic algorithms. And what better training than texts in all languages, from great authors, of different styles, eras and genres. Long before we talk about “deep learning” or “machine learning”, as early as 2005, Google will train, refine, optimize and “boost” its indexing and linguistic processing algorithm thanks to this extraordinary and unpublished digital literary database.
You now have the answer to the question of what Elon Musk could do with those hundreds of billions of tweets and tens of billions of conversations: train, refine, optimize, boost various artificial intelligence (AI) technologies that feed the companies he owns. Starting with his transhumanist project Neuralink. But AI is also at the heart of Tesla’s autonomous driving mechanisms. Both in the conversational dimension that he imagines in the “autonomous” piloting of vehicles but also in the complexity of data and decision chains to be taken into account to ensure this autonomous driving.
Elon Musk’s relationship with AI is deeply ambivalent. Ambivalent because he has often denounced its dangers, reminding anyone who would listen that a Skynet-like scenario (from the film Terminator) was not only possible but probable, and prophetizing an “AI apocalypse” (this was in 2014 during a speech at M.I.T). But at the same time, he never stopped praising the merits of AI and putting forward conversational humanoid robots, promising whole fleets of autonomous cabs for 2020 (failed), and revisiting the old fantasy of coupling man to machine through neural implants so that, he points out, mankind remains in control of the machines.
To be honest, no one knows today if Elon Musk’s plan is to use Twitter to work on artificial intelligence. But access to the entire conversational base of Twitter, the so-called “Firehose”, is a Holy Grail in many ways, and to imagine that Elon Musk has no idea what he could do with it or get from it would be, I believe, a colossal mistake.
For the rest, and beyond the case of the Twitter takeover, we have a double problem of governance, a double problem of a “cybernetic” nature (in the sense of Norbert Wiener, i.e. the science of communication control).
The first is that there is too much embodiment of the platforms in the shadow of the decision of one, whom we are reduced to hoping will behave like an enlightened despot. Facebook (now “Meta”) is Zuckerberg. Amazon is Bezos. And Twitter is Musk. I don’t know if we should always separate the man from the artist, but it’s certainly getting harder and harder to separate the man … from the platform he runs.
Our other problem is the field of ruin of the governance of these new digital spaces in the context of their unthinking by 30 years of public policies that have – at best – only reasoned in terms of “equipment”. People needed spaces for citizen dialogue and the public authorities wondered how many chairs they could provide. This is the paradox of our “technocracies”: not having seen technopower coming. Technocracies including today structurally incapable of formulating anything other than a form of magical thinking (the start-up nation and its “unicorns”) and incapable of measuring the impact of the power of technology for what it is, or preferring to project only the liberal irenicism of a potentiality of cost reduction (and of rights).
For a long time, it was believed or imagined that there could be a public digital media space open to everyone in equal participation, just as digital social networks can be (as long as we manage to intellectually disregard their economic model and their own interests). The web was partly that at its beginning, this concrete utopia. But this public digital media space is not going to happen, no matter what the founder and ex-CEO of Twitter, Jack Dorsey, says about it : he would have liked Twitter to be a “common good” free of any form of advertising. And since this is not going to happen, it is therefore necessary, at least and urgently, to propose new binding frameworks for regulating the ownership of media (traditional and social). The European DSA (Digital Service Act) goes in this direction for social media, but it will have to give itself the means of its ambitions, which is still far from being achieved. And if you read it carefully, it underlines the extent to which the absence of a “Media Service Act” applying to the regulation of private broadcasting is also felt.
Sociology is The Key. There is a determining and partly deterministic sociology of digital spaces. The fact that they are rather invested by men or by women, by young or by old people, the fact that the expression of minorities is rather favored or prevented (for example via the management of anonymity or pseudonyms), the fact that many journalists or academics are or are not present there, the fact that they are directly associated with other adjacent digital spaces, familial, professional or militant like those of messaging, etc.
From this first sociological dimension, the algorithm or rather “the algorithms” produce their effects in unison with the functionalities proposed by these spaces, but also with the “culture” specific to each platform (which can be rather centered on the text, on the image, on the music, on the video, with each time own temporalities and rhythms, etc.). Therefore, each social platform has its own “affordances”, i.e. its ability to suggest its own terms of use in a way that is as implicit as the TOS can be (sometimes) explicit.
To take Twitter as an example, there is no explicit indication that it is best to be mocking or sarcastic. Nor is there any indication that one should be angry or outraged as often as possible. Yet it is these expressive modalities, both suggested and collectively shaped, that we “buy into” in an almost physical sense, and it is these that shape our interest in the platform. If Elon Musk implements – and this is still far from certain – all of the things he has announced (less moderation, the possibility of editing tweets, reinforced user identification, a paid version without advertising, etc.), there will inevitably be a change in the platform’s “affordances” and, at the same time, a structural change in its sociology (without being able to predict today whether it will be substantial or marginal).
A November 2021 Pew Research Center study showed that :
“Most Republicans who consume news on Twitter (63%) say the site is mostly bad for American democracy, while only 26% of Democrats share this view. Democratic consumers of news on Twitter are more likely to say Twitter is mostly good for democracy (54%). Overall, Twitter news consumers are more likely than other Twitter users (and Americans in general) to be Democrats, to have a college degree, and to be relatively young. Twitter news consumers are also more engaged on the site in general – 46% say they visit Twitter every day.”
To use terminology from cybernetic theory, it is Twitter’s homeostasis, i.e., the system’s ability to maintain its equilibrium, that could be challenged and falter. And it is in any case between the possibility of this sociological change in the composition of the user base and that of its main affordances that the future of the platform and Elon Musk’s ability to keep control of it will essentially be played out. And this change has already begun with the accounts of political figures, editorialists and far-right public figures whose follower count has skyrocketed since the official announcement of the takeover. We will have to follow these organic shifts closely in the coming weeks to analyze their cyclical or structural dimension.
The poet Eugene Guillevic wrote “There are moments / When the slightest bird song / Is a precipice / That advances to swallow you.”
44 billion dollars is the price … of the precipice.
Andrei Lacatusu. “Social Decay”.
[This article is an automatic translation via DeepL of the one I published in August 2022 : “L’oiseau, le milliardaire et le précicipe“]
