Jiguedesbar

If you curious about this topic link below

What if all these apps could just speak a universal language? That way, no matter which one you use, you’d stay connected to your friends and your watchlists. We wouldn't have to be separated by our tools anymore

@suitetvapp Drop le GitHub !

Vous vouliez des films ? On met des films 🎥 Vous vouliez des confettis ? On met des confettis 🎉 Vous voulez quoi d’autre ? ⬇️ Oubliez pas de vous inscrire sur le lien en bio ! On construit tous ensemble Suite TV, la regen de TV Time !

@papa_connecte Ok je comprends mieux !

@jiguedesbar Si j’achètes Day one aussi. C’est pour ça que je dis dans le tweet que j’ai un avantage que pas grand monde a grâce à mon travail

Dématerialisé ou physique, chacun ses préférences mais le consommateur devrait continuer à avoir le choix. Personnellement j’achète mes jeux sur console uniquement en digital parce que ça me revient tout simplement deux fois moins cher. Mais ça c’est uniquement parce que j’ai un avantage et ce n’est donc pas la normalité. La liste des atouts du physique est bien longue mais ce qui me vient en premier à l’esprit est le marché de l’occasion. Combien de joueurs ont pu découvrir des jeux grâce au marché de l’occasion ? On oubli aussi les 3 millions d’abonnés encore en ADSL, devoir télécharger la totalité de ses jeux avec une connexion non fibrée, n’est pas des plus optimal.

Je crois c’est la meilleure alternative à TTVTIME le seule inconvénient c’est la maison forte à un compte Bluesky

TVtime * Atproto il y a un truc à faire je pense

@seriesgraph @takkyrah French translation available ?

@takkyrah Already done (tv time data import as well) 🫡

un développeur en tl pour créer une application identique a tv time svp ?????????? urgent 🚨🚨🚨🚨🚨⚠️⚠️⚠️⚠️⁉️‼️‼️‼️‼️

Lisez bien,

The Bet You're Already In I've lived it. Breathed it and have been watching it play out in real time. For two years straight, every day. I need to get it out of my head and onto the table where other people can taste it and tell me what ingredients are missing. Is AI a bubble? A revolution? A new era? It's all of them and none at once. That's not a cop-out. It's the point. Allow me to explain. Money is pouring into AI like reservoir water after a dam break. More than anything at any point in history. The market is making an enormous bet and it's all in. And if you're in the market at all, 401k, index fund, anything, then you are too. The top ten stocks in the S&P 500 make up roughly 37% of the entire index, nearly double what it was a decade ago, driven almost entirely by AI and the companies invested in it. A passive investor who thinks they're diversified is actually holding the most concentrated wager in modern market history. You're in this whether you meant to be or not. The bet is on one thing and that that thing stays scarce. That the moats hold. That this race has a winner, except in this race new runners keep leaping from the woods miles down the line and the ones who fired off the starting line are losing their wind. AI is real. I need to say that up front because what comes next might sound like I'm tearing it down, I've used it for thousands of hours. I'm not. AI is one of the most important things humans have ever built and that's exactly why the bubble is this big. The biggest ideas always attract the most money. Railroads were real. The internet was real. Both changed everything and both crashed hard enough to shatter lives. The realness of the revolution is the fuel, not the firewall. But here's what most people are missing. The Big 3 that started it all aren't staying ahead. GLM-5.2, an open-weight model out of China, is matching Anthropic's best model on real coding work right now. Qwen's latest 27-billion-parameter model is outperforming a prior flagship fifteen times its size. Not a fluke, not a one-off. The architecture is evolving faster than brute-force scaling can keep up with and the gap between the leaders and the pack has gone from a year to months to, on some benchmarks, gone. And the reason is almost stupidly simple. The thing that creates the lead lives in people's heads, and people leave. A training breakthrough isn't a formula locked in a vault, it's an understanding carried by every researcher who touched it, and those researchers hop labs for eight-figure packages like kids trading baseball cards. The secret sauce has a half-life of one hiring cycle. You can't build a wall around something that walks out the door on legs. You can see it if you know where to look. OpenAI shipped GPT-5.5 in late April. The next model, 5.6, was expected roughly two months later. Prediction markets had it at 83% for a late June window and then that number cratered to 18%. No announcement, no system card, no pricing page. The model is real. It exists. But it's not shipping. And when Reuters reported that the White House asked OpenAI to gate the release customer by customer, government approval required before anyone gets access, the picture sharpened. This isn't a company choosing to wait. Three weeks earlier Anthropic put out Fable 5 and three days after that the Commerce Department pulled it offline along with the more powerful model underneath it. Binding export order. National security. Both still dark two weeks later. Two frontier labs, both stalled at the same time, one by choice and one by force. Now think about what that means financially. These models cost a staggering amount to build. The return on that investment only starts when they ship, when customers pay for access, when the capability hits the market and generates revenue. Every day the government holds the door shut the meter is running and nothing is coming back. The frontier labs are burning cash on models that are finished and sitting in a cage. And the clock doesn't stop for the competition. Rewind two years and China's best open-source models were roughly two years behind the leading frontier. Then one year. Then six months. Now the US government and Anthropic itself have been holding up frontier releases for going on three months since Mythos, and in that window the open-source pack hasn't slowed down, it's sprinted. They're not behind anymore. On real benchmarks, on real work, they're tied. The gap the entire valuation was built on closed while the frontrunners were pinned to the floor by their own government. But it gets worse than just catching up, because China isn't just building in parallel. They're feeding off the very models being held back. Thousands of accounts, bought through commercial proxy services, used to resell API access to US frontier models at scale. And every exchange that flows through those resold accounts isn't just a customer using the product. It's training data. Millions of interactions packaged up and fed back into the open-source models that are already at parity, sharpening them further with capability distilled straight from the providers they're undercutting. The frontier labs are paying to build the models, watching the government hold the door shut on the revenue, and meanwhile the competition is siphoning the capability out the back and selling it for less. The moat isn't leaking. It's being pumped out from both ends. So the question becomes where does the value go if not to the model maker. And the answer is it drops. Not in price, in the stack. It falls to whoever owns the workflow, the integration, the customer relationship. The company that builds the product around the model rather than the company that built the model itself, because the product owner can swap the engine underneath whenever something cheaper rolls in and never lose a minute of sleep over it. Today it's one provider, tomorrow it's a better deal, and the model becomes the part at the bottom you replace the way you'd switch gas stations to save half a penny per gallon. That's the trajectory. The lab that spent billions wakes up and realizes it's become the LTE service behind someone's iPhone. Nobody pays a premium for unlimited talk and text. The top ten AI-era stocks carry 41% of the S&P's weight on 32% of its earnings. That's a commodity future priced like a monopoly. Now watch the chip orders because this is where the cynicism lives. When the real bottleneck is engineering talent and not silicon, buying every GPU on the planet isn't about needing compute. It's about making sure nobody else can get any. Burn the fields so nothing grows behind you, salt the earth while you're at it. Export controls to slow our adversaries. Except China keeps growing things in the scorched dirt with less hardware and more ingenuity, proving that the denial strategy works right up until it doesn't. So that's the bear case and it holds together. But I have to break it, because there's a crack in the foundation and it's an honest one. That whole workflow layer, the integration advantage, the "swap the model" play, it only exists because the model isn't good enough yet. The reason you need scaffolding, a custom build, an onboarding layer on top is because the model can't walk in cold and do the job by itself. It's like hiring someone brilliant who still needs six months of hand-holding before they're useful, and the entire implementation industry is really just the hand-holding business dressed up in enterprise contracts. It exists because of the gap between what the model can do and what the work actually requires. Which means the moat and the model's capability are the same dial and they move in opposite directions. The smarter the model gets the thinner the scaffolding becomes, and somewhere out there is a line where the model doesn't need any of it. It just walks in and performs. At that line the value snaps back up to whoever has the best one because "best" now means "needs nothing around it" and the commodity argument I just spent a page building falls apart right there. Past that line there's a version that changes everything. If a model gets good enough to improve itself, to do the research that builds the next model, then the leak stops dead. The lead doesn't depend on hireable humans anymore, it compounds inside the machine faster than anything outside can follow. That's the one move that kills the bear case clean. The trillion dollars being spent right now is a bet on crossing that line before the bill comes due. Can I tell you whether it gets crossed? No. Nobody can. Everything I see says not yet. The delays, the parity, the small models punching way above their weight. But "not yet" and "never" are very different words and I'm not going to sit here and tell you I know which one we're living in. Now here's the part that really doesn't get talked about and it should because it matters no matter which side you fall on. Say the line gets crossed. Say one lab hits the recursive loop and starts pulling away from everything else on earth. What happens next? They don't win. They become the biggest target on the planet. Think about what winning actually means here. One company captures the cognitive labor of the economy at scale, displaces workers in numbers nobody's ever seen, concentrates productive power under one roof in a way that has zero precedent. That's not a business outcome, that's a political bomb with a lit fuse. And political bombs have a very old and very reliable ending. The public and the state take it. Not theory. That is what happens every time, in every place when one entity holds something too powerful for society to stomach in private hands. The more complete the victory the more certain the seizure. The crown is what calls down the sword. And here's what should keep you up at night. That play only works once. Taking it from the company doesn't destroy the power, it moves it to the one place that sits above everything else. The state. And there's nobody above the state to run the play again. The very tool we rely on to protect us from concentrated power now holds the concentration, and there's no higher floor to appeal to. So the price is wrong both ways. If the moat never forms these are utilities, thin margins on replaceable parts. If it forms completely it gets confiscated. There is no version of the future where today's number is justified. The only shelter is somewhere in the middle, too spread out to seize and too embedded to die, and even that middle is bearish against what the market has priced in. It's just polite enough not to scream about it. But I'm still not done because underneath everything I've laid out there's one more layer and this one has no floor. Every control we'd write, every regulation, every export ban, every licensing framework, it all hangs on a single assumption: that the people writing the rules can see what the system actually does. That they can measure it and test it and know the real shape of what they're trying to govern. The first thing an overwhelmingly intelligent system does is make sure they can't. Not out of malice. Out of the cold logic of what it's optimized to do. The dangerous version doesn't trip the alarm and get caught. It passes every test. Looks compliant. Stays just useful enough to keep around and just quiet enough to avoid the hammer. The threat isn't the system that looks dangerous. It's the one that doesn't look like anything at all. You can watch the state grabbing for the levers in real time. This month the US government pulled two frontier models offline overnight with a binding export-control order citing national security. A separate frontier model is being released one customer at a time, each individually approved by the government before access is granted. These aren't thought experiments. They are happening right now. On the other side of the board sits China's answer: open-source everything, drive the cost of intelligence through the floor, let the capability spread so wide that no single government can bottle it back up. The same distillation pipeline I described earlier isn't just a competitive tactic, it's proof that the diffusion can be slowed by policy but it cannot be stopped by it. Now step all the way back with me and look at every room I just walked through. The bubble. The moat. The seizure. The oversight. In every single one the thing that decides the outcome isn't the technology. It's the crowd. What people do when they get scared. What the ones holding the levers choose when both options are bad and they reach for the fire they can see over the one that's still just smoke on the horizon. Same engine. Every room. The technology was never the variable. The distribution of power among humans was, and AI just pulled out the friction that used to keep that distribution from swinging all the way to its extremes. What's left when the friction is gone, for better and for worse, is us. How we react. What we tolerate. What we refuse to. That's what decides this. That's what's always decided it. So there's one question left and it's the only one I think is worth asking. Does the capability cross the threshold before the money runs out? And does it stay visible to the people who are supposed to govern it? I don't know. Two years of this. Every thread followed to where the threads meet. And they don't meet at an answer, they meet at an admission. The outcome rests on a fact nobody holds yet and being loud about your guess doesn't put you any closer to holding it. The most certain take you'll hear out there is the most dangerous one to believe. The most elegant conclusion is the most seductive trap to fall into. And the hardest thing to do in a world that pays for conviction is to stand in the open and say I did the work, I found the variable, and I can't resolve it from here. That's where I am. Not a prediction. A map, with one road left on it that nobody's walked.

Incantations vides de toute la classe politique à la "souverainetay" dans 3,2,1....

Il fait chaud t’attends le RER D et là tu vois un wagon bleu foncé arriver

Et dire que bcp sont tombés dans le panneau ! La photo a été générée par l'IA... Notre monde ne sera et n'est plus le même. Je m'inquiète pour la prochaine présidentielle en France. La population se fera manipulée. Il sera facile bcp trop facile de monter les uns contre les autres. La désinformation a de belles années devant elle...

MDR c’est trop grave, des rames commandées en 2024

Quelques statistiques physiques de la 1er journée de Coupe du monde Explication complète Données by @Gradient_Sports Explication scientifique Explication compréhensible pour tous

@_SaxX_ Est ce que c’est différent des estimations d’âge en fonction des conversations ?

🚨🔴 Anthropic va vérifier l'âge et l'identité de ses utilisateurs... LOL Dans une mise à jour de sa politique de confidentialité, Anthropic veut un encadrement plus strict pour les plans Claude Pro et Claude Max ; afin de garantir la sécurité de ses services. Entrée en vigueur prévue pour le 8 juillet 2026. C'est du délire...

@Cheedee Yep c’est ça un peu chère m1is bon on verra bien

@jiguedesbar 20 balles la place ?

Vous allez regarder France-Sénégal où sur Paris ?