Benjamin Bayart - Techno-fascime et souveraineté technologique

[00:00:06] On y va comme ça, sans intro?
[00:00:09] Oh, je, c'est vous les organisateurs, c'est vous qui savez.
[00:00:15] Ehm, c'est très perturbant de, de, d'être un des derniers conférenciers de la journée et qu'on commence à l'heure. C'est, je, je, c'est très étrange. C'est pas, c'est pas dans les traditions. Euh, pareil, si vous êtes vous êtes un public très perturbant pour moi parce que vous vous êtes pas tous tassés dans les rangs du fond. Euh, c'est très étrange. Je suis, euh,
[00:00:38] On sent que je suis pas dans mon élément habituel. Euh, du coup, je vais me présenter sommairement. Je pense que dans la salle, il y en a quelques-uns qui me connaissent déjà. Agitez un peu la mimine, les gens qui me connaissent déjà. Ouais, un petit tiers. Euh, donc je m'appelle Benjamin Bayard, contrairement à ce que dit le programme, c'est Bayard avec un T. Euh, mais c'est pas la première fois que les gens ne savent pas écrire mon nom. Euh, Je suis informaticien, développeur de formation, euh, ma spécialité quand j'étais étudiant, c'est programmation et architecture des systèmes de traitement de l'information, ça faisait ça faisait pastis, c'était très marrant. Euh, oui, ils étaient déjà bêtes les gens dans les écoles d'ingé. Euh, Je, bon, j'ai toute une carrière dans le privé qui est pas amusante. Et il se trouve qu'en parallèle, j'ai tout un lourd passé de militant. Euh, qui est probablement pour ça que vous connaissez les gens qui ont agité la main, c'est plus des gens qui ont vu ma tête dans des actions militantes que des anciens collègues, parce que j'en vois un là-bas, mais il y a pas beaucoup d'anciens collègues dans la pièce. Euh, Donc, euh, co-fondateur de la quadrature du net, s'il y en a quelques-uns parmi vous qui voient à peu près ce que c'est. Euh, ancien président de French Data Network pour les gens qui à qui ça dit quelque chose, le plus vieux fournisseur d'accès internet de France, qui se trouve être une association gérée par ses utilisateurs. Euh, j'ai créé et présidé longtemps la Fédération des fournisseurs d'accès internet associatifs. Bref, ça fait 20, 25 ans que j'ai une activité militante, beaucoup moins dense ces dernières années. Mais qui du coup m'a amené à avoir des pratiques très étranges. Donc moi je suis écriveur de code, j'ai commencé en CM2. Euh, donc ça fait, ouh là là, 43 ans que je programme, un truc comme ça. Euh, mais je suis aussi admin 6 parce qu'il a bien fallu gérer les machines, mais je suis aussi admin réseau parce que quand on est président d'un fournisseur d'accès internet, il y a intérêt qu'on sache faire du réseau. Euh, puis j'ai appris à lire du droit puisque quand on essaie de militer contre toutes les bêtises de nos politiques sur les réseaux, il faut bien qu'on apprenne à faire du droit et cetera et cetera. Donc j'ai un profil tout bizarre qui fait que je me mélange plein de, plein de compétences étranges. Et donc il a fallu que je, que j'improvise un titre pour ce que je voulais vous raconter. Euh, j'ai eu la flemme de préparer des diapos, euh, du coup, vous avez pas de bol, les gens parmi vous qui ont besoin d'un support visuel pour capter leur attention, bah vous êtes défavorisés. Il y a pas de support visuel, enfin moi j'ai un support visuel pour guider ma mémoire, mais vous vous n'en avez pas.
[00:03:22] Donc, euh, j'ai pris comme titre techno-fascisme et souveraineté numérique, parce que le mot techno-fascisme commence à émerger ces jours-ci. Euh, pour décrire ce qui se passe dans des pays euh très peu démocratiques, et alors d'habitude c'est la Russie qui est très peu démocratique ou la Chine, puis là tout d'un coup c'est les États-Unis. Euh, donc ça complique tout. Parce qu'on n'est pas, on n'est pas habitué, euh, bon, il va falloir revoir deux trois standards. Euh, j'avais pas envie de vous faire mon chapitre habituel sur le cloud souverain parce que d'abord il est déjà en vidéo sur YouTube, donc allez le regarder. Euh, et puis il colle pas à l'actualité quoi.
[00:04:04] Donc, j'ai prévu quatre parties, si je vous faisais ça comme si j'avais fait des diapos, là vous verriez une diapo avec un plan qui dit un géopolitique, deux mode de réponse.
[00:04:15] Euh, trois, euh, différence par analyse, quatre, la boétie. Donc du coup, comme si vous aviez les diapos, vous savez pas de quoi je vais parler. Euh, et c'est fait exprès.
[00:04:29] Mais c'est pour aider les gens visuels à visualiser. Bien, donc on va parler géopolitique.
[00:04:35] Euh, ça ça a l'air d'être un truc très savant la géopolitique et non, c'est de la politique et de la géographie, enfin ça va, on va y arriver. Euh,
[00:04:44] Il y a un certain nombre de d'événements géopolitiques qui parlent de numérique et qui permettent de comprendre le rôle du numérique dans le monde, euh sur ces dernières années qui moi me paraissent intéressant à évoquer. Alors je suis pas non plus remonté à Mathusalem. J'aurais pu vous faire le fichier Safari en France, histoire de 1977. Euh, mais j'ai pas eu envie de remonter aussi loin, donc je suis remonté que à des trucs dont vous devriez vous souvenir.
[00:05:13] Euh, Snowden, les révélations d'Edward Snowden sur ce que les Américains surveillent chez leurs ennemis ou à propos de leurs ennemis, mais chez leurs alliés et à propos de leurs alliés, du coup la notion d'alliés, pas clair. Vous voyez, c'est finalement c'est peut-être pas de la semaine dernière que les ricains sont devenus bizarres.
[00:05:35] Euh, Et en fait, Snowden, il nous disait des choses. Il nous disait un pouvoir qui a pas trop de scrupules couplé avec des entreprises hégémoniques, ça permet de surveiller la planète entière. Moi ça me paraît pas être une analyse de très haute volée à partir des révélations d'Edward Snowden d'arriver à ça. Et je vous jure dans la classe politique, les trois quarts des gens n'en sont pas encore là dans leur analyse. Il faut pas faut pas trop s'étonner hein que qu'ils comprennent pas, ils ont une puissance d'analyse faible.
[00:06:07] Mais pour moi, c'est important ce morceau-là, euh parce que il montre déjà tout ce dont je vais vous parler.
[00:06:14] Donc en fait, toutes les analyses qu'on voit fleurir en ce moment de gens qui se disent ah là là, les Américains sont dangereux, il faut qu'on s'en protège. Et donc on devrait bouger et c'est une bonne idée de mettre tout le mail de Polytechnique chez Microsoft.
[00:06:28] Ah bah, c'est une décision d'il y a 8 jours. Du coup, c'est sûrement pour assurer la souveraineté nationale qu'on prend une école militaire qui forme notre élite et qu'on se dit tiens, on va tout foutre chez Microsoft comme ça les données seront safe.
[00:06:42] Voilà. Donc déjà, rien qu'avec Snowden comme information, Snowden, je sais pas si vous situez dans le temps, mais c'est il y a il y a quand même quelques années, hein, il y a
[00:06:52] petite décennie. Euh, il y avait déjà assez d'informations pour dire aux politiques que c'était pas malin de faire ça. Et ça c'est de la géopolitique, c'est dire OK, les intérêts des Américains ne sont pas toujours alignés avec les nôtres. Des fois oui, hein. Je dire, quand on surveille les terroristes islamistes, euh, tout ce qu'on trouve en surveillant les nôtres, ça intéresse les Américains et tout ce que les Américains trouvent en surveillant les leurs de terroristes, ça nous intéresse. Donc des fois nos intérêts sont alignés. Curieusement, ce qu'ils trouvent en surveillant Airbus, ils nous en parlent pas.
[00:07:25] C'est ça de la géopolitique.
[00:07:28] Euh,
[00:07:31] Un autre morceau que je que je voulais évoquer, euh, ce sont les sanctions contre la Russie euh suite à l'agression de la de l'Ukraine par la Russie. Alors je sais que depuis 15 jours, les États-Unis considèrent que c'est l'Ukraine qui a agressé la Russie. Mais il y a 2 ans, c'était pas comme ça. C'est faut faire très attention, ça c'est le côté compliqué, il faut se souvenir de comment était l'histoire d'il y a 2 ans par rapport à la Pravda d'aujourd'hui. Euh, quand l'Ukraine a été agressée, il y a eu des sanctions décidées contre la Russie. Dans les sanctions, il y a eu un paquet de sanctions américaines euh y compris une forme d'embargo. Euh, ce qui s'est traduit par le fait que euh les gens comme Amazon ont coupé les services pour un paquet d'entreprise russe. Alors euh, couper les services d'Amazon, si c'est qu'on peut plus se faire livrer un grille-pain en urgence pour le lendemain, c'est très perturbant mais c'est c'est pas un drame. Euh, en revanche, quand on coupe les services de WS, parce que je pense qu'on parle un peu plus d'informatique dans la salle, euh, bah ça éteint le système d'information d'un paquet d'entreprise. Et ça c'est chaud.
[00:08:40] Et en fait, tout d'un coup, le monde entier s'est rendu compte que tiens, sur une décision assez discrétionnaire du gouvernement américain, on peut éteindre le système d'information d'une boîte étrangère.
[00:08:56] C'est chaud.
[00:08:59] Euh, Bah, ça dépend de ce qu'on vend, ça dépend de ce qu'on fabrique, ça dépend à quel endroit ça se trouve. Et si même pour le fabricant de gripin, c'est très grave. Je dire, si le fabricant de gripin, il peut plus fabriquer des gripin, euh, ben euh, déjà il va être triste. Ensuite, si ça se trouve, il va être allez, pas pauvre, mais moins riche et il y a des chances que il y a des pauvres gens qui deviennent chômeurs. Ça c'est c'est vachement embêtant pour le fabricant de gripin. Euh, et donc on s'est rendu compte que tiens, le numérique, c'est pas une commodité, c'est pas comme d'acheter des trombones, c'est en fait, il y a il y a il y a un pouvoir là-dedans. Quand je décide d'aller, de déporter mes installations numériques ailleurs, euh, je donne du pouvoir à cette personne sur moi, une capacité de me nuire. Est-ce bien raisonnable ? Suis-je bien sûr que c'est un allié ? Qu'est-ce que je fais si on me fait des faux ? Comment je réagis ? Comment je réponds et cetera ? Vous allez me dire c'est vrai de n'importe quelle forme de sous-traitance. Oui. Mais si je sous-traite euh le ménage dans les bureaux, bah au pire, le ménage est pas fait dans les bureaux. Ça, ça va mettre plusieurs semaines à être embêtant. Si je coupe le système d'information de l'entreprise, il faut combien de semaines pour que ce soit pénalisant ?
[00:10:17] Vous voyez, par exemple, je sais pas, on prend une grande banque systémique, toute l'informatique vitale est sur ses serveurs en propre qu'elle détient, c'est encore comme ça aujourd'hui.
[00:10:29] Et pourquoi ? Ben, simplement parce qu'ils savent très bien que s'ils mettent ça ailleurs, bah le ailleurs peut couper le service. Et si on coupe les services de paiement ou les services de tenue de compte, euh, ça va pas mettre 3 semaines à foutre le bazar, hein. Ça va mettre une grosse demi-heure. OK.
[00:10:49] Donc ça, c'est un élément qui est devenu plus clair lors des sanctions contre la Russie. Autre élément, euh, Quand on a euh décidé, quand on s'est dit que la fabrication de de l'armement était un sujet clé en France, euh, on a décidé qu'on ferait des Rafale et qu'on n'achèterait pas des avions américains parce qu'on voulait être indépendant. Euh, Il devient ces dernières semaines un peu plus évident que c'était pas si bête. Ça nous a coûté une petite fortune parce que il faut quand même entretenir la famille d'assaut à grand renfort de pognon, euh, mais au final, nous nos avions de chasse ils peuvent voler même si les Américains ne veulent pas.
[00:11:38] Et ça nous place à part dans ce ce jeu de tension internationale un peu crispée. C'est curieux quand ils tiennent pas le même raisonnement pour AWS. C'est très curieux.
[00:11:52] Euh,
[00:11:55] Et puis ben dans les éléments de géopolitique, évidemment, euh, il y a le le deuxième tome de l'air Trump, euh, le premier avait déjà été relativement agité, le le deuxième commence fort fort fort. Euh, tension géopolitique internationale, le mec a envie d'annexer le Canada, il veut envahir le Groenland. On rappellera le Groenland, c'est l'Union européenne, hein. Donald Trump veut s'accaparer une partie du territoire de l'Union européenne. Je,
[00:12:25] Son pote euh Poutine veut s'accaparer une autre partie du territoire de l'Union européenne, non pas l'Ukraine, ça c'est pas encore dans l'Union européenne, mais dans les trois États baltes, là, Estonie, Lituanie, Lettonie, euh ça rigole pas du tout dans ce coin-là. Euh, Donc là aussi, on a de la tension géopolitique assez assez rude. Et donc quelle est la probabilité qu'à un moment euh l'administration Trump décide de sanctionner quelqu'un euh par un embargo commercial en coupant des services ? Pour le moment, il joue au droit de douane, mais ça pourrait lui prendre.
[00:13:04] Donc voilà, ça c'était mon premier chapitre. Sur OK, c'est quoi les questions géopolitiques dont on parle ? Euh, Et en fait, les questions géopolitiques dont on parle, c'est pas tellement l'usage d'une technologie américaine parce que pour le coup, vous pouvez prendre un bout de code qui a été écrit à New York.
[00:13:21] Si il tourne sur une machine qui est sur votre bureau, modulo les éventuels backdoor et trous de sécurité qu'il y a dedans, il tourne quoi. En revanche,
[00:13:34] le fait de sous-traiter à quelqu'un l'opération de votre informatique, c'est pas la même chose que l'usage d'une techno. OK. Ça, ça crée une grosse difficulté.
[00:13:45] Et donc on a des on a des réponses, ce que je dresse un tableau assez noir, mais en fait, en vrai, on a des réponses et donc certaines sont très intelligentes. Euh, Et donc ça c'est mon mon deuxième chapitre sur c'est quoi les modalités de réponse qu'on peut utiliser et comment on fait tout ça. Euh, Alors, il y a un premier élément que je veux citer, qui est le RGPD.
[00:14:09] Vous allez me dire mais de quoi il parle ? Il nous parle de militaire d'armement de machin, qu'est-ce qu'on s'en branle des des des bandeaux de cookies sur les sites web ? Non non non non non non non non. Le RGPD est incompatible avec le droit américain, ça fait deux fois que la Cour de Justice de l'Union européenne, pour ceux d'entre vous qui ont pas fait de droit, au-dessus des cours suprêmes de tous les États membres de l'Union européenne, il y a une cour qui dit le droit européen. OK, c'est la Cour de Justice de l'Union européenne. Donc elle se prononce après les cours suprêmes.
[00:14:42] OK, c'est elle qui dit le droit européen. Elle a dit deux fois que le droit européen sur la protection des données personnelles était incompatible avec le droit américain et que donc il n'était pas légal qu'une entreprise soumise au droit américain manipule les données personnelles de quelqu'un qui habite en Europe. Ça fait deux fois qu'ils l'ont dit.
[00:15:04] On a quand même signé une troisième fois un accord international qui dit que si si on peut on peut le faire. Et l'accord en question n'a n'est pas encore passé devant la cour en question. Il se pourrait qu'elle dise une troisième fois la même chose.
[00:15:21] L'air de rien, ça veut dire que les textes européens sur la protection des données personnelles ont un effet de protectionnisme réglementaire. Puisque si on les appliquait, alors je sais, on applique pas la loi. Et si jamais on l'applique, après ça fait toute la merde que vous voyez dans les journaux télé depuis 2 jours sous prétexte que quelqu'un a envisagé d'appliquer la loi. Vous n'y pensez pas, ça ne se fait pas. Mais si on appliquait la loi, n'importe quelle entreprise qui transfère des données personnelles vers une entreprise sous juridiction américaine est hors la loi, risque une amende, l'amende maximum, c'est 4 % du chiffre d'affaires annuel consolidé du groupe. N'importe quelle entreprise qui met son fichier client sur AWS, 4 % du chiffre d'affaires.
[00:16:13] Par infraction.
[00:16:17] Et oui, par infraction.
[00:16:20] Des fois, il y a plusieurs infractions.
[00:16:23] OK. Si on appliquait le RGPD, on pourrait pousser l'industrie du numérique américain vers la sortie. Ce qui du coup inciterait les grands groupes industriels à devenir plus clients des fournisseurs européens. Ce qui pourrait créer de la croissance économique en Europe, un truc de dingue. Ce qui pourrait permettre qu'on s'approprie des technos. Et que quand on n'est pas tout à fait assez bon dans ces secteurs technologiques, on devienne meilleur parce que on est les moyens de payer des ingénieurs et cetera et cetera. Voyez, un truc de fou, hein. Donc ça, c'est une arme protectionniste. Beaucoup plus maline que ce que l'agent Orange là fait avec ses droits de douane. OK. Euh, Parce que avec ça, on est capable de pousser vers la sortie les autres et donc de favoriser un développement économique et un développement technique dans des secteurs dans lesquels on est déjà pas spécialement mauvais. Euh, tenez, on peut se tenter un quiz à la con. Qui parmi vous connaît des ingénieurs qui ont été formés en France et recrutés aux États-Unis ? Qui parmi vous connaît des ingénieurs qui ont été formés aux États-Unis et recrutés en France ? Voilà. Il y en a, c'est comme les poissons volants, il y en a, c'est pas la majorité de l'espèce.
[00:17:41] Donc en fait, on n'est pas si mauvais que ça.
[00:17:45] Euh,
[00:17:48] En fait, les les histoires de Cloud souverain, c'est pas une question euh de seulement choisir la la nationalité du fournisseur et ou donc il faut un ordinateur qui soit impérativement fabriqué à Clermont-Ferrand. Euh dont le code logiciel a été écrit en banlieue de Paris et qui tourne dans un data center qui soit ici les Moulinoux et machin, c'est pas ça le sujet. Le sujet c'est d'analyser les dépendances, de comprendre lesquelles sont acceptables et lesquelles ne le sont pas. Est-ce qu'on peut placer l'exécution de notre code sur des serveurs que n'importe qui, enfin que l'autre peut éteindre à tout moment. Et qui est l'autre qui peut éteindre ? Et est-ce qu'un jour on pourrait se fâcher avec avec lui et est-ce que c'est raisonnable ou pas de lui donner ce pouvoir de vie ou de mort ? Mais c'est et c'est vrai pour toutes les dépendances qu'il y a besoin d'analyser. Quand on décide d'acheter du gaz naturel à quelqu'un, euh, s'il nous fait faux bond. Qu'est-ce qui se passe ? On avait décidé en Europe d'acheter du gaz à la Russie, on se fâche contre la Russie, qu'est-ce que ça fait ? On a décidé à la place d'acheter du gaz aux Américains et de le faire livrer par bateau, si on se fâche avec les Américains, qu'est-ce que ça fait ? C'est ça, réfléchir aux dépendances et OK. À quel point elles sont acceptables, comment on fait pour réduire le risque ? Par exemple, on décidant qu'on va aussi acheter du gaz à l'Algérie, on réduit le risque parce que la France qui achète son gaz en Algérie pourrait revendre une partie de son gaz, se contraindre un peu sur sa consommation et revendre une partie de son gaz à l'Allemagne et donc peut-être, vous voyez, c'est ce genre d'analyse là. Euh, Et donc les histoires de de cloud souverain et de et de bêtises comme ça, c'est toujours une question de comprendre les dépendances et de les choisir. Il peut pas ne pas y en avoir, hein, je veux dire être indépendant ermite dans sa grotte, ça fait quelques siècles que ça fonctionne plus. Enfin, si vous êtes cardiaque, ça veut dire que vous devenez mort, si vous êtes diabétique, ça veut dire que vous devenez mort, euh si comme moi, vous êtes malvoyant sans lunettes très vite, vous avez les bleus partout à force de vous cogner, enfin, ça complique tout. Donc indépendant tout seul dans sa grotte en ermite, c'est pas le mood. Mais en revanche, il faut savoir de qui et de quoi on dépend et comment on accepte cette dépendance et comment on réduit le risque. Euh,
[00:19:54] Dans les modes de réponse, également, il y en a un que j'aime bien, qui est ce qui est en train de se passer autour de Twitter. Ça vous parle ? Euh, le fait qu'il y a un move vers Blue Sky, ça vous a échappé ? Non, vous l'avez vu. Euh, alors, il y a il y a il y a un truc de fond qui est le terme à la mode qui est apparu il y a peu de temps, c'est Enshittification. Est-ce que ça parle à tout le monde ou est-ce que je développe ? Ça a l'air de tous vous parler. Ah. Alors je développe. Euh, quand c'est quelqu'un qui est le premier à me le demander. independent alone in his cave as a hermit, that's not the mood. But on the other hand, you need to know who and what you depend on and how you accept that dependence and how you reduce the risk. Uh, among the response modes also, there's one I like which is what's happening around Twitter. Does that ring a bell? Uh, the fact that there's a move towards Blue Sky, did you miss that? No, you saw it. Uh, so there's a fundamental thing which is the trendy term that appeared recently, it's Ensification. Does that speak to everyone or should I elaborate? It seems not to all of you. Ah, then I'll elaborate. Uh when it's someone from the front row asking.
[00:20:24] The principle of enfication is when a platform is set up, uh, it needs to make itself known, it needs to attract people, uh people don't necessarily want to, especially when it's a product that didn't exist before, like microblogging. And so, the platform needs to be as welcoming as possible. Uh, it needs to be easy to come, it needs to be very gentle at the entry, no barriers. The service provided must be absolutely impeccable, you shouldn't bother people. Uh, you have to make sure the service is fluid, that we are happy with it. So, it's true for Le Bon Coin, it's true for Google on email, it's true for Twitter, it's all these things. Then, once you've managed to capture people and you've created a product that's addictive enough so they don't really want to leave, it needs to be a bit costly. You see, it's like leaving Facebook. I didn't fall into it, it's very easy for me to stop, I haven't started. But stopping smoking is complicated, because I started and it's been 30 years and more. Uh, so I can't stop. It's very hard because I can't do without it. And it's the same game with these platforms, they all have that addictive side that makes it hard to leave, and it's not easy to get out. But on the other hand, we make it very comfortable so you don't want to leave. But at some point, once you've become the big hyper-centralized platform, so Twitter, Facebook, you know them. Uh, well, at some point, there's the investor who says, "Sure, but uh, money." And so, we're going to start putting more ads, being a little more aggressive about you, you, you, you put an ad blocker. You don't watch the video. like YouTube for a few weeks now. You see? And then little by little, we tighten things up, uh it provides less service, but it maximizes revenue. Uh, you'll have to watch at least 3 minutes of video without pausing and without looking away because you're being monitored by the webcam to be able to continue using such and such a cool service. That's ensification. And in fact, the platform has become central enough that it no longer really has competition, so you can't get out, and you're addicted enough that we can torture you to extract money. And then, well, the platform really becomes shitty but profitable. Uh, and there can be a temptation to flee. That was the first move towards Blue Sky. And obviously, the second movement uh is Musk, the gentleman with the big arms.
[00:22:59] who throws his heart to the crowd with a very nice Nazi salute and that doesn't make people want to stay with him.
[00:23:06] For me, there's a problem there.
[00:23:09] Because there's an analysis error, but that's my, that's my chapter 3. Uh, so this movement from Twitter to Blue Sky, uh it's super interesting as a mode of response, we say because the platform belongs to a fascist, I'll take the platform that belongs to someone else. Uh, and then there's a fourth mode of response which is basically what I call the desire for a European Google, which for me is one of the most toxic elements there is. Uh, and which allows me to transition to my third chapter which is how we differentiate responses based on the analysis we perform.
[00:23:47] Uh, when we say I'm switching from Twitter to Blue Sky, we're saying I'm changing billionaires. That one looks stupid, I'm going somewhere else. I prefer watching Patrick Drahi's TV rather than Vincent Bolloré's TV, he seems less idiotic.
[00:24:04] Honestly, I'm not sure.
[00:24:07] But that's exactly what we do when we switch from Twitter to Blue Sky. Uh, when we say, "No, but Amazon are scumbags, so I'm going to leave AWS and go to GCP because at least over there they're cool guys."
[00:24:24] the analysis, well,
[00:24:27] that works if the problem is indeed Amazon. It doesn't work if we have a slightly more nuanced analysis. And so the whole question is to understand. Is the problem Mr. Elon Musk because he blew a fuse and went crazy uh and so he's the problem? Is the problem having a monopolistic, hyper-centralized platform that belongs to a private company which therefore only answers to the interests of its majority shareholder? And has no form of democracy or anything whatsoever. I mean, the worst dictatorship is a corporation. Okay? Even the People's Republic of China is less democratic than any company, even a benevolent one. Okay?
[00:25:15] Uh, so, is the problem Elon Musk? Is the problem this hyper-centralized structure, well, is Twitter itself, by structure or by essence or by divine decision, something unhealthy? Or is it the structure? It's about saying, there's a fully centralized network, entirely owned by a single company, therefore monopolistic, with no form of sovereignty, but which is used as a place for public debate, as a place where people discuss. And we took an essential cog of society, which is discussion between people, and we entrusted it monopolistically to a structure that is inherently dangerous because it can become toxic at any moment. And so, is it the person's problem? Is it the company's problem or is it a structural problem? So, I, I, I don't want to, uh, like that, it's not new as a mode of analysis.
[00:26:17] When Montesquieu explains to us that if there is no separation of powers and balance of powers, then necessarily there will be a tyrant, because the structure allows a tyrant and therefore a tyrant will eventually emerge. He does that, a structural analysis. Montesquieu, that's not recent, huh. It's about a century before the French Revolution for those who struggle, so it's been roughly 3 good centuries since he explained that.
[00:26:48] And so, for me, one of the absolutely important elements of analysis to understand, is that the toxicity in it is having this hyper-centralized platform system that allows for monopoly situations and that allows, consequently, when a toxic idiot takes the lead, any aberration whatsoever. Since it's a normal capitalist corporate structure as we know them, therefore without counter-power. Uh, so anything can happen.
[00:27:19] So, for me, in terms of analysis elements, okay, we have tools that allow us to create a form of protectionism to push geopolitically dangerous people out. We've seen that, it's the GDPR, it's a form of response. There's a second form of response which is to try to learn to detect these extraordinarily toxic structures and to avoid them.
[00:27:46] And in fact, to avoid them, we know how to do it. I note that this is not what people are doing.
[00:27:53] Because a network somewhat like Twitter, in a microblogging style, that allows for reasonably efficient discussion and does not exhibit this structural flaw, it exists.
[00:28:07] It's called Mastodon.
[00:28:10] The network is called "le fait divers" (the miscellany/news item).
[00:28:14] Uh, the network has existed for quite a few years already. Its first breakthrough was in 2017. So 8 years ago.
[00:28:21] I don't know if you know it.
[00:28:24] Yeah, there are a lot of IT people, so you have a vague idea.
[00:28:29] It's a network of interconnected systems respecting a standard, where there are thousands of interconnected servers forming the network. Don't think it's anecdotal, huh. Mastodon has 2-3 million daily users, approximately. So it's not at all negligible in front of, well, it's, there you go, it's not at all negligible in front of Blue Sky. It's obviously very small compared to Twitter, but it's not at all negligible compared to Blue Sky. So the fact that the move happens from Twitter to Blue Sky rather than from Twitter to Mastodon is absolutely not obvious. Okay?
[00:29:02] Uh, but that will truly be the subject of my fourth part.
[00:29:08] So how does it, how does it work, what is the thing actually? It's an open system, which therefore relies on standards because, in fact, for thousands of servers to be able to interconnect, there needs to be a common standard. Why aren't Twitter's servers interconnected with Facebook's servers? Because actually, they could be. They just aren't. Because they don't respect the same standard, because they don't want to create a link at all, because they don't want people to be able to navigate from one system to another and migrate. Okay?
[00:29:37] Uh, but there you go, an open system, we know what it is, it implies standards, it implies interconnections, it implies interoperability, consequently, it leads to a diversity of actors. Consequently, any of the actors can go completely crazy, we don't care, they are one among a thousand. He will be ostracized, typically. Uh, among the social network platforms built on Mastodon, there's Trump's social network, which includes Trump and his orchestra and is not interconnected with anything because a fascist network is not interesting. And so, almost all administrators said, we'll leave the idiots in their corner, we won't talk to them. And therefore, we know how to resist that kind of thing. Because there's a great diversity. Uh, the big flaw of this thing, which is looking for open systems, because it works for your company's IT, huh. Instead of creating a system entirely based on Amazon's closed and addictive tools, you decide to work only with free and open software, uh and so you don't take the slightly funny but closed protocol and you only take open protocols on which you know how to work. You can, you know. consequently, if you only do cubes, any cube host can host your system and therefore, you can change hosts. On the other hand, if you use the Amazon Poette service with the blinking thing and the little dangling gligli, well, you can't go anywhere else because there's no little dangling gligli and consequently, you can't work anymore. And your IT system cannot move unless you do 3 years of work. And since you don't have the means, well, you'll stay there. That's typically what happened to people who use VMR, if that speaks to you. Okay? Uh all their IT depends on VMWare and VMWare is not a standard and therefore the VMWare editor decides to multiply the selling price by two or three. well, it's a mess because to get out and switch to any other system, you need millions of hours of work to redo the entire company's IT system, and so, well, we'll pay. We'll grumble, we'll see if we could get out sometimes, but we'll pay.
[00:31:47] The advantage of an open system is that if you rely on an open system, well, you can get out of it. You can switch from one to the other since there are standards, there's interoperability, there are interconnections. So you can switch from one system to another.
[00:32:02] Uh, the big flaw of open systems is that they assume standards. However, a standard is something that is quite established and quite stable, which has been negotiated at length. So establishing a standard takes time. I don't know if some of you have ever followed the work of the standardization committee for anything, whether it's the C language or HTML. Uh, moving the slightest comma in documents to make version N+1 appear, it's 15 years of discussion, meetings to check that we're not doing anything stupid, because if we put something stupid in the standard, we're still going to carry it around for a long time and we'll have to deal with it. So, well, there are good reasons for it to be long, but it is long.
[00:32:41] Uh, and so obviously, all of this slows down innovation. the C language, it doesn't release a new version every 6 months, unlike Go or Python. Okay? Python can evolve its language slowly every 6 months because there's only one. There isn't a standard.
[00:32:58] So the flaw is that it significantly slows down innovation. And the known alternative to that is to create a digital common.
[00:33:07] That is to say, there are two solutions: there's the C solution and there's the Python solution, to put it simply. You are all more or less coders in the room. Uh, the C solution is to create a standard. So the C language is standardized. And it doesn't evolve every week, huh. Between Richie Kernigan's C, the C standardized in '89, the C of '95, and I imagine there are some slightly more recent C standards with two or three evolutions, which probably didn't revolutionize the thing, but evolutions. There are some from time to time. But as a result, C compilers, there are 50, 100, 1000 of them. There are C compilers that are used to compile code for Montra Quartz, there are C compilers that are used to compile code to put on your Linux or on your Windows or on your tablet or on your car's onboard computer or on anything. C compilers, there are.
[00:34:02] And they all read pretty much the same C model, they're not all perfectly up to standard, but well. 99%, it's portable this thing. Okay? That's a standard. Very, very stable, it moves very little, and hundreds of implementations exist. The electrical outlet, if you will, is a standard.
[00:34:24] There are hundreds of manufacturers of electrical outlets, whether they're male plugs or female sockets to embed in the wall or whatever, there are hundreds of them and it's quite interoperable in reality, huh.
[00:34:35] When you plug a LeGrand plug into another brand's socket, it doesn't catch fire. Crazy. Because it's a standard. The opposite is to make a digital common.
[00:34:49] This is the case for Python, for example.
[00:34:52] Or there are a few others, but I'm taking this one because I'm staying at the same level, it's a programming language, it evolves. Uh, there's only one. That is, I don't know of a second implementation of a Python interpreter other than the one distributed by the Python community. Why is it a digital common? That is, what differentiates it from Go? Well, it's maintained by a community.
[00:35:17] It is under a truly free license, so there are groups of people who contribute to it, who discuss among themselves. In fact, any of these groups, if they were a bit angry, because the community wouldn't be going in the right direction, could fork the project, cut it, say we're starting from the last known version and we're going somewhere else. And instead of calling it Python, we'll call it Cobra and it will function differently.
[00:35:43] Because we believe the community has unblocked it. And so no one can appropriate the thing. It's a common, meaning people contribute more than they take. So all the people who come to discuss, contribute, negotiate how to evolve it, they contribute more than they take. It's managed as a digital common, it truly belongs to no one, and if someone tried to monopolize it, the project could reorient itself and leave, do something else.
[00:36:09] That's a digital common. It's very complicated to produce. There are people who pretend. Google with Go pretends.
[00:36:18] Uh Go is maintained more by Google than by external contributors.
[00:36:22] When there are more external contributors than Google's marketing will, the thing will have become a digital common because at any moment, it can diverge. Uh, a good example is Terraform. You see Terraform, everyone was saying it's a digital common, we more or less agree on how all the providers, plugins, modules, widgets are written. Consequently, all hosting solution providers have provided modules for Terraform and with Terraform, you can manage pretty much all platforms. This thing really looks like a common.
[00:36:53] Except that it's Hashicorp's property and Hashicorp decided to change its license model and consequently, you can no longer use it as you wish. So it wasn't a digital common.
[00:37:05] They even decided to close the code. The license no longer allows all reuses. So, we took the last known version that we were allowed to modify freely and forked it to become Open Tofu, which looks a bit more like a digital common. We'll see in a few years. but which incorporates all functionalities, relies on the latest known code versions that we are allowed to use, and then diverges as a result, but which has a usage license that causes far fewer problems. So you see, that's a way to solve the structural problem. And in fact, by solving the structural problem, one can acquire a form of independence that allows protection against geopolitical problems.
[00:37:53] Do you see roughly how I did the splits?
[00:37:56] Okay, I started with Trump wanting to wage war on Greenland and we ended up with Open Tofu.
[00:38:03] Good. The fourth chapter I wanted to talk about, which will be quite short. Uh, that's the one I called the Boetie. So now, you might see why I'm talking about the Boetie to you. Who sees why?
[00:38:16] There's one who's following. So Étienne de la Boétie. You have read, obviously, because you are very cultivated people, you have read Étienne de la Boétie's Discourse on Voluntary Servitude. No.
[00:38:29] Read it.
[00:38:32] It's about 40 pages, written in French that's easy to read. It's not Rabelais's old French and it's not the tiring French of today's TV people. It's easy-to-read French. It's 1700s French. Uh, Étienne de la Boétie wrote that, he was 17-18 years old.
[00:38:55] Yeah, he was a kid. Étienne de la Boétie was a kid when he wrote that. Uh, it's a very short text, read it, and basically, it explains that it's not them who are powerful, it's you who are kneeling.
[00:39:07] Stop kneeling and they will stop being powerful.
[00:39:11] Uh, that's the core of his argument.
[00:39:14] But in fact,
[00:39:18] uh, there's a difficulty in what I'm telling you.
[00:39:23] Uh, how do I present this to you?
[00:39:29] Well, first of all, there's a first difficulty I've already talked about, which is the inability to perform a structural analysis. Okay? Our politicians are capable of saying "Musk bad." They can do that.
[00:39:43] They are capable of saying "Twitter is toxic," but they've only been able to say it since "Musk is bad." Before, they weren't able to say "Twitter's structure is unhealthy."
[00:39:50] So, there's a gap in their analysis, they're not able to analyze the structure. They analyze the effects, meaning once it's burning, they say, "Oh, it's hot," but why is it hot?
[00:40:01] And that's still quite annoying.
[00:40:05] Uh, that's really an important first aspect. But I, I make fun of politicians because it's easy, but I have the same criticism for all business leaders. Okay? Uh, the company that decided to put all its IT on AWS or with Google or with whatever without thinking about the structural consequences of the thing. One might say a minister rambling nonsense, well, it's the same, they don't go any further.
[00:40:32] Uh, there's a second, more complicated element: ease as a source of harm.
[00:40:41] Well, saying I'm going to give the keys to the truck to Amazon means I won't have to think anymore. sorry, I can give you a positive presentation of it. I'll be able to focus on my core business.
[00:40:54] I, I'm a baker, I knead bread. So knowing how computers work, I don't want to know, I want to make bread. It's already relatively technical and relatively complicated. Several of you tried during lockdown. Now you talk to your baker with more respect.
[00:41:11] Because he does that every day and doesn't mess up, and the bread rises every day, and he hasn't forgotten the salt. Well, that's, there you go. It's convenience as a source of nuisance. Uh, well, to say I'm going to give the truck keys to Amazon, that means I won't have to think anymore. Oh, sorry, I can give you a positive presentation. I will be able to concentrate on my core business. I mean, I'm a baker, I knead bread. So knowing how computers work, I don't want to know. I want to make bread. It's already relatively technical and relatively complicated. Many of you tried it during lockdown. Now you talk with more respect to your baker because he does it every day.
[00:41:12] And he doesn't mess up, and the bread rises every day, and he hasn't forgotten the salt, well, that's it. Uh.
[00:41:22] There's, we can say it in two parts, either it's not my business. But in that case, you have to prove it. Because I swear to you that my baker, if his IT is broken, first, he still manages to make bread, second, he manages to sell it. He might not be able to take card payments, but he will be able to sell bread.
[00:41:43] He'll take cash, or checks, or debt acknowledgements, or little pearls like at Club Med, he'll manage. Okay? BNP, if its IT no longer works, I think it can't do much anymore. Okay? And so, between the two, there are a lot of companies that depend vitally on their IT and weren't able to analyze, let's say, no, convenience isn't clever, actually. In fact, IT is my core business.
[00:42:16] You see, people like Doctolib, for example. We agree, IT is their core business. I mean, if their IT goes down, the company no longer exists. I mean, even if the Haribo distributor is still in the offices, we don't care, okay? These people run their entire system on AWS.
[00:42:36] So, they depend entirely on a provider over whom they have no control, and I think they don't have a business continuity plan in case AWS screws them over.
[00:42:47] It's annoying.
[00:42:50] Uh, and in fact, convenience is a source of nuisance. But if I tell you differently, there's a refusal of competence and mastery.
[00:43:00] Because in fact, to choose an interconnectable, interoperable solution, etc., you first have to have done the structural analysis, and then, well, you need a bit of competence, meaning you have to accept that...
[00:43:10] In fact, it's an old principle among industrialists, if you have to manage a subcontractor, you have to be more competent than your subcontractor. It's like that.
[00:43:18] Uh, if you regularly need to call on an industrial manufacturer who makes mechanical parts and stuff for you, if you don't speak his language and his business, he's going to mess with you. If you regularly have to call on an IT service provider, but you have trouble telling the difference between the keyboard and the mouse, he's going to mess with you, and not necessarily on purpose. But you just can't control him. And so he's going to do his best with what he can, with what he knows. So eventually he's going to drive prices up, but above all, you can't challenge him, you can't verify that the solution he proposes is relevant. You can't go confront him with another professional who might have another, much more effective idea to solve your problem, because you can't control it.
[00:44:09] And so, uh, between two professionals, one who says, "Well, I'll do it for you with three little scripts, it'll take two hours," let's say, 3,000 euros for the service because I want to be well paid. Then the other one who tells you, "I need to do three years of development..." with some Go, some Java, some thingamajig, and a whole CI that allows you to verify. If you don't know if the service rendered is good, you can't judge one service compared to another. In short, you need to be competent to manage a subcontractor. And so, if you want to outsource complicated IT, because when you delegate it to Amazon, you're outsourcing it, you know. That means that all the part about how we manage a data center, how we ensure redundancy, what are the climate problems, what are the energy problems, you've outsourced it.
[00:44:56] How can you have confidence if you don't understand what they're doing?
[00:45:00] When you decide that everything that is system installation, system virtualization, you're going to outsource it. How can you outsource if you can't manage your subcontractor?
[00:45:12] It's a big difficulty.
[00:45:15] And so this choice of convenience and this refusal of competence poses a problem.
[00:45:25] Yes, but I only have one hour.
[00:45:26] No, I just want to hear the question, just so you know, there was a point...
[00:45:30] The second reason is that speed is enormous. I know, but it's also, I know very well that it's not the only reason they use the service, speed is the, it's the... And well, it's a choice. But I'm not entirely certain about the notion of speed. I think that's why they made the choice, but I'm not sure it's pertinent. Uh, but we'll keep some time for questions. So for me, really, there are these three elements: there's convenience as a source of nuisance. There's the refusal of competence and mastery, and there's the inability to do a structural analysis. The refusal of competence, I've seen that in many companies. I don't think Doctolib did that out of refusal of competence, I think it was out of convenience and speed. And that they told themselves, structural analysis, we don't care. Uh, but suddenly, that means that France entrusted its COVID vaccination campaign to a company that was potentially fragile, uh, and therefore that today, there are thousands of doctors who no longer make appointments otherwise than through this company.
[00:46:34] which finds itself in a kind of vaguely monopolistic nerve center, a little bit of an essential resource, not far from being a vital operator, since we can paralyze French medicine by paralyzing Doctolib, and they haven't bothered to have a business continuity plan. There's a structural analysis problem. I understand the convenience, but there's a structural analysis problem there. Uh, and so, that's what I wanted to bring you to, is that in fact...
[00:47:05] What prevents companies from doing this analysis are these elements. The refusal of competence is something I know very well. Because you would need to have an IT team that is quite capable, that knows the infrastructure well, that knows the company well, and therefore doesn't leave, so you can't rely on providers that you're going to fire every six months. Uh, it creates a lot of difficulties, it means that you become responsible for what you do. That is, if it doesn't work, it's our fault. Whereas if we've outsourced, and additionally, we've outsourced to the number one in the market, it's no longer our fault. Alright, I'll conclude on this. You may have already heard me tell the difference between a manager and a director. I learned that when I worked at France Telecom. There's an enormous difference between a manager and a director. It's that the manager is not a director, and that the director is not responsible.
[00:48:00] That's the difference.
[00:48:03] It's easy.
[00:48:05] Well, yes, the manager is the one who has to get up at night if it doesn't work. So, it will work.
[00:48:12] He's going to put an unapproved piece of tape with a piece of string and something that pulls on the bell so that, but it's going to work, because he wants to sleep at night. So, whatever happens, he'll find a solution for it to work. The director, what he wants is for it not to be his fault.
[00:48:29] Anyway, whether it works or not, he won't get up at night. He has a manager for that.
[00:48:34] But on the other hand, if it doesn't work and he's made a somewhat committing choice. We could blame him for it.
[00:48:45] However, he doesn't want us to blame him. So, he chose the most expensive, the number one on the market, the one everyone chooses.
[00:48:55] It's absolutely not adapted. He went and bought Ferraris to deliver, to make deliveries. So that's stupid. It's absolutely not adapted, it's totally overpriced, but we won't be able to blame him for it. Because he took the sector leader.
[00:49:12] And that's the whole difference between a director and a manager. A director has a refusal of complexity and a refusal of competence and a refusal of responsibility. He doesn't want it to be his fault. As long as there's someone whose fault it is in his place and to whom he can transmit the blame, generally, he's a little more at peace. We have too many directors.
[00:49:31] We have about ten minutes left if you have any questions.
[00:49:43] Thanks already for this moment. You talk a lot about reducing dependencies, and isn't there another solution to rather increase interdependencies? So I need you more and you need me more, and so that means you're not going to hit me because you'll need me.
[00:50:00] Well, that can work, it's a bit like the principle of nuclear deterrence. I need to be able to kill you.
[00:50:08] But it can be a response, to replace a dependency with an interdependence. But that implies a certain form of education.
[00:50:19] That implies that the other understands that there is an interdependence and that therefore they understand. Because in fact, in reality, between the European Union and the United States of America, there is an interdependence. I mean, the stupid things he's doing right now on customs tariffs, the orange agent there, uh, that's going to do at least as much harm to the American economy as to the European economy. I'm just not sure he understood it.
[00:50:41] No, I'm not sure.
[00:50:42] No, I think he thinks little, this gentleman.
[00:50:47] And in the series, he refuses complexity, uh, he doesn't, I don't think he's stupid, to really speak. It's not like that. But I think he refuses complexity and that therefore the first analysis that comes to him, he considers it to be exact, and he acts very brutally according to that. And actually, that's why when you respond quite brutally, he backs down a bit. Uh, typically, when he had extremely brutal behaviors towards Mexico or Canada, and there were very brutal responses in return, he backed down. Uh, but you see, this idea of interdependence, that implies a kind of common intelligence where, okay, we agree, we understood each other. A bit like what happened between Russia and, well, between the USSR and the United States, during the tough periods of the Cold War. That is, in short, all the officers of the Russian army and all the officers of the American army know very well that if it ever goes to World War III, it's the end of the world. Everyone is going to atomize everyone, there will perhaps be survivors who will be tardigrades under the polar ice caps, but that's all, uh, there will be nothing left. Uh, so, we're going to try not to do it. But you see, that implies a form of common intelligence. However, what we regularly see in the world we live in is that sometimes intelligence disappears.
[00:52:12] Uh, whether it's under, it can be in the form of a state that is tilting, like what we're seeing right now in the United States. Uh, it can be in the form of, uh, other, other crises, well, you see, uh, Afghanistan is a civilized country, in reality.
[00:52:31] Right now, no. But it's not because the Afghans are idiots, they just happen to be in power in Afghanistan right now. Uh, and so, we could have found intelligent solutions with around Afghanistan 50 or 100 years ago, because they were educated, instructed people, etc., and created interdependencies. But once foolishness takes over and takes control, interdependence is no longer a guarantee of anything. Once foolishness, uh, takes over and takes control, interdependence is no longer a guarantee of anything. Similarly, I'm very wary because there's a bias.
[00:53:08] Uh, oh, but for once, that comes back to what I'm saying about reducing dependency, uh, understanding it and choosing it. Possibly choosing to have several. That is, saying, "Okay, I don't want to manage my IT at all, so I have to go through three, four, five cloud providers." But then, I make the effort to have people competent enough to manage my subcontractors and to manage the way I can at any time switch from one to the other, how do I create it's extremely complicated, huh? It's extremely complicated to do. I know very few people who are able to say, "Okay, my system is portable enough that I can, in such and such a time, switch from one provider to another," and that implies insane development constraints, etcetera, huh?
[00:53:49] But interdependence works between actors of roughly equal size in a multipolar world where there are many of us, uh, typically, it doesn't work at all between my company, whatever the size of the company, and GCP.
[00:54:04] Because there's a fundamental asymmetry.
[00:54:08] Another question.
[00:54:15] Thanks already, it's super interesting. I'm very much in agreement with the analysis of manager/director, but in fact it underlies the very question of the company and the capitalist model behind it. It's short-term interest, it's uh, what does the organization promote, what's the plan behind it?
[00:54:36] That's a more complicated question.
[00:54:40] Uh, because in fact it's less related to what I told you for an hour. Uh, if you haven't seen it, I recommend the conference I gave in Luxembourg a few months ago, something that must have been called Voxed. Uh, the video is floating around on YouTube.
[00:55:01] You search Voxed, Bayard, Luxembourg, you'll find it. In particular, I explain that it's wrong, companies do not have the objective of making money. If their objective was to make money, they wouldn't be organized like that.
[00:55:14] Uh, so no. The goal of the company is to satisfy the libido of the boss. That's how it is.
[00:55:21] And so a company, there is certainly a side where it has to bring in money for the shareholder, it's one element among others for the shareholders whose libido is focused on money, but it's also stories of power relations and domination relations and how power and domination relations are organized in the social group that is the company. Capitalism being one of the ways to organize these power and domination relations. And in fact, if you try to analyze the behavior of companies just by money, it doesn't work. They should be organized in a very different way and they would be infinitely more profitable by being organized in a very different way. However, the boss might not be the boss. And so, because he wants to have the big red car because he needs it to screw around.
[00:56:09] It's a bit more detailed in the Voxed presentation.
[00:56:15] We want the title of the talk.
[00:56:18] Oh, it was called Geopolitics of Data because I gave it a generic title. But really, if you search Voxed, Luxembourg, Bayard...
[00:56:25] YouTube, you'll find it in a few seconds.
[00:56:28] Yes, question. I'm not going to ask questions about the cloud.
[00:56:33] You have to put the microphone close to the mask.
[00:56:36] Like this. Um, my question was, to limit dependencies, or not dependencies but for example on the example of cloud providers or even on Twitter or another. That means increasing, well, we have less of a big dependency and several small ones.
[00:56:49] But having multiple dependencies increases the cost of complexity that we have to synchronize what needs to be synchronized or align what needs to be aligned, entre les coups. What is the right way to gauge this complexity and how can we accept it?
[00:57:07] Well, in fact, you don't necessarily increase complexity, you can end up reducing it. For example, uh, you decide you don't want to be dependent on your cloud provider, so you're not going to go to Amazon, and you decide you want to be able to change cloud providers. So you're going to be interested in hosts who provide you with reasonably standard solutions. Well, I don't know, are you going to say that you're going to take people who provide you with Proxmox, or are you going to say that you're going to take people who provide you with OpenStack? Then, it's not at all full feature, you lose a lot of features. But you say, "Okay, I'm content with that as a feature." Uh, because then, I know how to move. And in fact, I have four providers who provide me with this well-standardized thing, but I limit myself to the standard.
[00:57:54] So there are a lot of functionalities I don't have access to, but I'm mobile. Typically, you say, "Okay, I'll keep the hyperscalers, but I'll only use their VM functionality." I'm just going to rent VM from them, and on top of that, I'll deploy systems that I'll operate myself, etc. Well, in that case, guaranteeing portability, you know, writing the Terraform boot script that deploys an infra on Amazon or on Scaleway or on OVH or on Google.
[00:58:26] So I've already made it much more portable by saying I've restricted myself to a very small subset of functionalities that are reasonably standardized, and where the small compatibility layer between my different providers isn't very hard to write. So in fact, it's rather by removing perhaps unnecessary complexity, or by moving it, and saying, uh, "Okay, I agree to outsource this to them because I know I can find five subcontractors who produce it, and so I can switch from one to another." And so, there's the rest of the complexity, I'm going to take it on because if I delegate it, I become dependent on a single provider.
[00:59:04] So it's more in that place that you're going to find things. But otherwise, you also have the solution that consists of saying, uh, "Okay, if you have the critical size for it, I choose a provider to whom I'm going to outsource all of this, but I'm a shareholder." But I have a form of control over it, but I've put my hand in, but I've built with some colleagues a consortium capable of, and so I... And when I say a consortium, everyone thinks, "Oh yes, BNP, Société Générale, Crédit Mutuel, BPCE, etc. will ally to create a consortium for banking data hosting." It can be a charitable work like that.
[00:59:46] Or it can be models that I have practiced, that is to say, uh, three associative internet access providers, so things that are not even, uh, the annual budget of these associations is not even the small amount of our former common employer.
[01:00:05] And yet they will be able to say, "Okay, we're joining forces, all three of us, to build such a common brick." Uh, I'm thinking of how we created Gitoyen. You see, five, six associations have teamed up to create a network operator. Because for each of them, being a network operator was too complicated, too costly, too time-consuming, but that combined for the five of us, it was not a silly thing, and so we could bring on board five, six, seven others, and so we created a consortium which is another association, very small. But to which we outsourced this thing, rather than being dependent on a provider who would be unique and who could kick us out at any time. So we bought our independence, saying, "Okay, there, there's a node of know-how that needs to be held." And so we're going to create a small group of volunteers who will carry that know-how, who will learn, who will grow in competence, and so on.
[01:01:01] So it's really a question of understanding at what point someone can kill me, do I accept it, do I not accept it? And how do I manage it?
[01:01:11] And it can be to reduce dependency, to reduce complexity. To say, "Okay, this is too complicated. If I do that, I become dependent and vulnerable, and I don't want to be vulnerable, so I'll do without it."
[01:01:26] Okay, one last very quick one, very quick.
[01:01:32] Well, it's really to conclude because it's the end and it's not a question. The Quadrature du Net has made its call for donations.
[01:01:39] to cover its entire 2025 budget and they're only at 49%. So I invite you to participate.
[01:01:51] I won't contradict him.
[01:01:52] Thanks.

Transcript (Translated)