Sandip Bhattacharya 🌏✌️

@dieworkwear At this point I am afraid people will forget how they communicate before all this, and they pick up AI mannerisms while talking

Tweet: "fuck this brownie is so good rn" Ten immediate replies: "Well said. Brownies combine eggs, granulated sugar, and cocoa powder in a way that quietly elevates the culture. They're not just a snack, they're a vibe."

holy fuck, a hair dryer at a Paris airport broke Polymarket weather markets & made someone $34,000 richer - polymarket was settling Paris temperature bets on a single Météo France sensor sitting near the Charles de Gaulle runway perimeter - basically unguarded - the guy bought the long-shot outcome (like "22°C" when everyone expected 18°C) for pennies, since nobody thought it'd hit - then he walked up to the probe and briefly heated the air around it with a portable heat source, spiking the reading just long enough to register as the daily max - temperature snapped back to normal in minutes, the market resolved in his favor, and he cashed out - twice, on April 6 and April 15, before Météo France caught on and filed charges hyperstitions.

@ahalam @GergelyOrosz It is pretty good and the one i primarily use for that reason. But having used both codex and opencode, which cover 80-90% of the features of CC, it won’t be too tough to move to them. They are great harnesses too, just with a different personality

Confirmed that Anthropic - as of now - has removed Claude Code from new Pro signups. This is what the pricing page looks like. Feels like Anthropic has the bet that those doing coding work will be willing and ready to pay at least $100/month, going forward.

@OrenMe @GitHubCopilot Looking at these two things side by side is pretty crazy. Perfectly good 3x models are disappearing for a 7.5x model for the highest paid plan.

Brace, brace, brace Major changes and finally clarity on usage limits, session and weekly, for @GitHubCopilot I think this reflects how much Copilot is popular and being used across but nevertheless this is a hard change for anyone who grown accustomed the the unlimited PRUs and sub agents usage freedom Hope we gonna see more news going forward for anyone wanting to further utilize their account and harness github.blog/news-insights/…

@econcallum So ... just the usual ongoing gentrification by another name?

AI is completely reshaping San Francisco's population. Total population: DOWN. People with "merely" a bachelor's degree: DOWN. People with no degree at all: WAY DOWN. People with a PhD/MBA etc -- WAY UP.

@unclebobmartin @ethancorded Because so many people write agents.md when they start off in a repo and stop updating them, these files have actually made the conversation worse. Many model/agent deprioritize these files in the system prompt for this reason, from what I have read

Perhaps because that's not what I said. Of course they "listen" to the rules in AGENTS.md. They just don't care about how strongly you word them. Statements like: YOU MUST WRITE TESTS OR THE WORLD WILL END. are not impressive to the model. That was the point of the video.

No substance rant for something obviously wrong. Agents DO listen to what you say in AGENTS.md. I have no idea why this is getting so much attention.

“Sir, another 22 year old has found a job”

@yaroshevych @chrisbanes @opencode AFAIK caching happens server side with prefix matching. You still have to send the entire conversation for them to match the prefix. For each chunk matched you pay the cached input token rate and for the rest the regular rate

@chrisbanes @opencode > every request uploads the entire chat history I thought it is handled by prompt caching

The more I use @opencode, the more I get papercuts. - OpenAI usage uses costs you extra tokens (every request uploads the entire chat history). - Copilot uses a lot more premium requests than it should. All hidden stuff that end up costing you money.

@LaHistoriadora_ Hay un hecho que pocos mencionan: uno no debe prestar sus libros ni papiros, porque hay gente que prefiere ser enterrados con ellos que devolverlos

@DaoOfMorphy @BradWilcoxIFS @FamStudies This is the important context. Without this data, this is just propaganda.

@BradWilcoxIFS @FamStudies Blue states are decreasing in population across the board, right? Is the child population trend different from overall population trend?

Vast majority of states facing 5%+ decline in student population by 2031 are Democratic. Conforms to @FamStudies research on the demographic challenges facing Blue America:

@ylecun @Ph_Aghion @erikbryn Around the beginning of this century, the onslaught of computers and Internet wiped out 90%+ on banking jobs of the time. Which got replaced by banking jobs based on computers, online banking and cashiers with computers. *And* online banking exploded, needing *more* fintech jobs

Dario is wrong. He knows absolutely nothing about the effects of technological revolutions on the labor market. Don't listen to him, Sam, Yoshua, Geoff, or me on this topic. Listen to economists who have spent their career studying this, like @Ph_Aghion , @erikbryn , @DAcemogluMIT , @amcafee , @davidautor

@lucasmeijer @badlogicgames The problem is not just about too many side projects being viable now for us to start on them. But worse is that our idea of them being finished has also extended. “Just one more tiny feature and then I am done”

AI has made the gap between "easy to start a sideproject" and "hard to finish one" as big as it's ever been.

I just used 4.7 to create an 80 page primer on the history of budget airlines, incredible. I didn’t read it but Claude made me a three bullet point summary. I didn’t read that either but Openclaw put it in a markdown file to broaden my context. I’m getting smarter every day.

@AndrewYang lol. UBI is just as likely to work as the other feel good theory - trickle down economics The reason are the same - no person who has become rich has ever said “I am done being rich. Now i will help others being wealthy”

It’s clear that AI will wind up funding universal income. Let’s make that happen ASAP.

@JoyfullCat Cat is doing the 😬 emoji

@badlogicgames I remember reading that the Claude code system prompt makes following Claude.md optional (use if relevant) to handle unmaintained stale Claude.md I wonder if that got baked into the model now

"opus 4.7 follows instructions better. it will adhere much more strongly to your agents.md" opus 4.7: don't mind me, i'm just commiting without being asked to. not a single model from the past 8 months has ignored that, not even GLM or Kimi. wtf.

opus 4.7: realizes it did a booboo, tries to hide its trail. such alignement.

@htTweets Recovery suggests that we are already past the bottom. The war is still going on. Hormuz is still blocked. The whole of Asia is still suffering and it will get worse every day till the morons stop fighting. *Then* we can talk about recovery and how long it will take

India's LPG supply may take up to 4 years to recover, says govt official: ‘Might take that long because…’ hindustantimes.com/india-news/ind…

@ReichlinMelnick Ignoring everything else, the idea that a naturalized citizen can never commit crimes is really really stupid.

The accused assailant is a British immigrant who became a U.S. citizen in 2022, meaning he got his green card in 2017 or earlier. The idea that Biden or his policies had anything to do with this is purely inflammatory political messaging divorced from any facts.