Datarama retweeted
is Opus 5.5 nerfed?
I just had Gemini 4 Argon analyze 1,400 Claude Code sessions and it found that on November 3, something definitely happened.
the model name didn't change, but the way it does reasoning took a sharp turn.
Datarama retweeted
This is f*cking insane. Andrej Karpathy, the man Elon called the #2 computer vision mind on Earth:
"Prompting is going away. Delete everything, keep the graph"
In 2 hours, he shows how to build graphs and why they're the only layer left standing
Prompts → Agents → Loops → Graphs
A loop keeps one agent working until the job is done. A graph decides which agents exist and what each one hands to the next
Worth more than any $2,000 AI engineering course
Bookmark and watch it today
Then read how an AI employee runs QC for motion engineering with Opus 5.5 ↓
Datarama retweeted
Politiker, die glauben, mit einem Schulsystem welches in PISA abstürzt und ohne Änderung der verkorksten Energiepolitik und Entlastung vom überbordenden (Sozial)Staat im „Wettbewerb“ bestehen zu können, sollten keine Verantwortung für das Land bekommen.
Deutschland wird den Wettbewerb um die niedrigsten Energiepreise oder Arbeitskosten nicht gewinnen. Unsere Stärke liegt woanders: in hervorragend ausgebildeten Menschen. Heute habe ich in Bielefeld die Landesbesten von #NRW gefeiert. Sie sind die Leistungsträger von morgen.
Introducing Jev @typesafeai powered social-api
Try it at treg.to/social
- One API to access any social data
- Track every mention of you & your competitors
- Find trending posts in your niche, filtered by Jev
- Under $0.05 to pull 1,000 competitor ads
Open source, 0% markup
👇 Repo link below
Datarama retweeted
If you want to really understand AI agents, spend some time on this new site.
You won't regret it.
secondbrainos.dev turns the whole subject into a map you can explore, and it's honestly brilliant.
Every page, term and tool is a bubble.
The lines show what depends on what, so you can actually see how the pieces fit together.
here's what's inside:
> a 7-module course, from a single prompt to a production agent
> 5 handbooks, each exploring one layer and ending with a build you can finish in an evening
> a 65-page hands-on project where an agent builds and maintains a second brain for you
> a glossary of 57 terms and 100+ tools
130 pages. 1,236 connections. 9-10 links per page.
It's FREE, written in plain Markdown, and the repo (in 🧵↓) is public too.
Datarama retweeted
Interesting bits from Brian’s Airbnb earnings call about AI agents and consumer behavior:
— Airbnb gets ~500k bookings a day
— ”very non-material” traffic from Muse/Instinct
— says people *want* to browse: average user looks at dozens of listings and 25+ photos
— 90% of users message the host
— most traffic comes between 9 to 5 when people are at work (using Airbnb as escapism)
— 80% of travels say they like doing their own travel research
Datarama retweeted
Third-worldism starts with not thinking of the day after tomorrow
When we still built Grand Central, much less Stratford Hall, the goal was to build for forever and lift the spirit. To make tomorrow grander and greater
Look around. That civilization no longer exists. Instead, we have a mixture of schemes, cardboard houses, ugly assortments of concrete and glass, and hordes of migrants who have been imported to fill them so that GDP might notch another mark a bit higher for one quarter more, damn what consequences may result
Such are the first steps to becoming a Third World country
Soon those wretched and ugly structures will fall into disrepair. The vagrants and squatters will creep in a bit more. The migrants will pile up a bit more. The mob will take that which it could never earn.
Welcome to the favela. Welcome to the Third World. Welcome to the world created by those who brought it here
Datarama retweeted
Fun fact: one of the top 10 Ashkenazi maternal lines, J1c7a, was found in Bronze Age Gotland and the Wielbark Culture, making a Gothic origin likely.
Not a simple E. European story (their mark got overwritten by Slavs)—but probably via the Goths’ migrations to Italy or Iberia.
Datarama retweeted
As an overseas Chinese who has worked on China’s energy transition and energy security for more than two decades, I’m tired of German commentators selectively using China and mis-represent China to fit a green narrative.
Yes, China is building vast amounts of wind and solar. But unlike the permanent subsidy model some Germans still defend, China ended national feed-in tariffs for new solar and wind projects from June 2025. After more than twenty years of government support, these technologies are expected to compete in the market.
At the same time, China is adding more coal capacity than the rest of the world combined, and building more nuclear reactors than the rest of the world combined.
My generation and older generation of Chinese still remembers food and energy insecurity. We do not take modern industrial prosperity for granted. China’s approach has always centered on securing energy and food supply in a changing climate — not on treating intermittent renewables as an ideological end in themselves.
That pragmatic multi-source approach is the actual story about China, not the misleading version currently circulating in Berlin by wind and solar lobbyists.
China verabschiedt sich nicht nur für den Klimaschutz von Öl. Der Staat hat verstanden, dass Erneuerbare Energien Länder deutlich unabhängiger und damit sicherer machen
Diese Groschen ist in Berlin leider noch nicht gefallen.
zeit.de/wissenschaft/umwelt/…
Datarama retweeted
Wer ist jetzt der Narr?
Whitehall, März 1638. William Laud, Erzbischof von Canterbury, geht in den Ratssaal, und der Narr des Königs Charles ruft ihm eine Frage zu. Aus Schottland ist die Nachricht gekommen, dass sich das Land gegen Lauds neue Liturgie erhebt, und Archy Armstrong will wissen: „Who's fool now?" Wer ist jetzt der Narr?
Dazu kommen gröbere Worte: Betrunken hat er den Erzbischof einen Mönch, einen Schurken und einen Verräter genannt. Laud beschwert sich beim Rat, und der verurteilt den Narren noch am selben Tag: Ihm wird der Narrenrock über den Kopf gezogen und wird vom Hof verbannt.
Was er verspottet hatte, nahm seinen Lauf. Lauds Liturgie trieb den König in den Krieg mit Schottland, der Krieg zwang ihn, das Parlament einzuberufen, und das Parlament brachte Laud in den Tower. 1645 verlor der Erzbischof den Kopf, 1649 der König. Armstrong starb 1672. Er hat beide überlebt.
Der Hofnarr hatte ein politisches Amt. Er durfte dem Herrscher unverblümt und brutal sagen, was der gesunde Menschenverstand nahelegte und sonst niemand auszusprechen wagte: dass der Plan nicht aufgeht, dass das Volk murrt, dass der Minister ein Narr ist. Der Narr des Habsburgers Leopold von Österreich soll ihn 1315 vor dem Zug gegen die Eidgenossen gefragt haben, wie man aus dem Land wieder herauskomme, wenn alle nur berieten, wie man hineingelange. Der Herzog zog trotzdem und verlor bei Morgarten. Ein kluger Fürst hielt sich den Narren, weil die Warnung billiger war als der Fehler.
Gestürzt hat Charles I. weder sein Narr noch dessen Verbannung. Zu Fall gebracht hat ihn die Politik, über die der Narr gelacht hatte. Die Verbannung zeigte nur an, dass dieser Hof keine Warnung mehr hören wollte. Wer den Narren bestraft, beseitigt nicht den Fehler. Er beseitigt die Warnung.
Berlin, Oktober 2026. Der Arzt und Journalist Paul Brandenburg hat sich die Rolle genommen, die an den Höfen der Narr hatte: Er sagt, was andere nicht sagen wollen oder nicht sagen dürfen, in einer Sprache, die viele für zu brutal halten.
Am 28. Oktober nahm ihn die Bundespolizei am Flughafen BER fest, als er sein zehnjähriges Kind nach Berlin zurückbrachte; seither sitzt er in Moabit in Untersuchungshaft. Vorgeworfen werden ihm sechs Beiträge im Netz, in denen er den Regierenden Bürgermeister Kai Wegner beleidigt und verleumdet haben soll. Haftgrund: Flucht. Ein Mann, der nach Angaben seines Anwalts mehrmals im Monat nach Berlin kommt und mit seinem Kind am Flughafen festgenommen wird, soll flüchtig sein.
Das Bundesverfassungsgericht hat 2022 einen Mann, der nach einem Freispruch erneut wegen Mordes vor Gericht gestellt wurde, gegen Meldeauflagen auf freien Fuß gesetzt.
Bei sechs X-Beiträgen über den Regierenden Bürgermeister hingegen muss es für die Berliner Justiz Untersuchungshaft sein, in einer Stadt mit fast zehntausend offenen Haftbefehlen.
Über den Regierenden, dessen Ehre so geschützt wird in Kürze: Am 3. Januar 2026 saßen nach einem Brandanschlag Zehntausende Berliner ohne Strom und Heizung. Herr Wegner erklärte am Tag darauf, er habe eingeschlossen im Arbeitszimmer koordiniert, gar mit dem Kanzleramt telefoniert. Gelogen. Später räumte er eine angebliche Stunde Tennis am Mittag ein, mit seiner Lebensgefährtin, der Bildungssenatorin. Was er an diesem Tag wirklich tat und wo er war, weiß die Stadt bis heute nicht.
Ein Gericht musste die Senatskanzlei zur Auskunft über seine Telefonate verpflichten; er selbst sagte: „Dazu ist alles gesagt." Im Juli gab er die Spitzenkandidatur ab, im September fiel seine CDU von 28,2 auf 18,8 Prozent.
Tacitus schrieb über den Prozess gegen den Historiker Cremutius Cordus, man könne über die Dummheit derer nur lachen, die glauben, mit der Macht der Gegenwart das Gedächtnis der Nachwelt auszulöschen; die Bestrafung der Köpfe steigere ihr Ansehen, den Verfolgern bleibe die Schande.
Das stand schon 1638 in jeder Bibliothek, und es steht dort noch. Armstrongs Frage geht deshalb an den Berliner Senat. Wer ist jetzt der Narr?
Datarama retweeted
my spidey senses are telling me the labs are, as we speak, using unguardrailed versions of their unreleased models to systematically reverse-engineer ALL of the internet’s software, apps, and games, then feeding the reconstructed source code straight into their training pipelines while simultaneously building guardrails to ladder-pull the rest of us
Datarama retweeted
Wir fassen zusammen:
1. Grüne Fundamentalisten schalten die deutschen Atomkraftwerke 2023 ab.
2. Die Energiepreise explodieren.
3. Die Grünen werden mehr und mehr verhasst.
4. Die Grünen stellen 2026 fest, dass Atomstrom doch billiger gewesen wäre.
Uuupsi, dumm gelaufen...🙄
Datarama retweeted
America is a Latin American country now, so let me explain politics:
We elect communists to kill the rich. But communism brings poverty & crime. So we elect fascists to kill the communists. But the fascists hand the economy to oligarchs. So we elect communists to kill the rich.
Helen Andrews was right.
15 years after taking student loans, men have paid them down substantially compared to women who have barely made a dent.
Lower pay, longer career exits, and especially the 2007 Public Service Loan Forgiveness racket which erases the debt for employees of nonprofits or a 501(c)(3).
So effectively, taxpayers are literally funding the college degrees of an army of women who then staff the NGOs and advocacy shops actively wrecking the country.
If these women had to pay for their own education with their own careers instead of offloading it onto everyone else, they surely wouldn’t outnumber men on campus.
Datarama retweeted
Texas results by group (NBC exit poll):
🔴 White men: +23 Republican
🔴 White women: +13 Republican
🔵 Latino men: +19 Democrat
🔵 Latino women: +42 Democrat
🔵 Black men: +77 Democrat
🔵 Black women: +84 Democrat
🔵 All others: +10 Democrat
Like it or not, demographics are destiny
Datarama retweeted
We decoupled memory from the model.
Living Brain is one persistent brain for ChatGPT, Claude, Cursor, Codex and your own agents. Write in one, retrieve in any.
92.2% on LoCoMo. 464ms median retrieval. Up to 22.6x fewer tokens per question.
LLMs have parameters. Living Brain has plasticity.
Never getting in a Lear 60 again
And with commercial being 10x cheaper and 10x safer, I’m only flying the big iron from the majors
What’s the point of saving two hours on average when you have starlink on 50% of @united flight already
See you in the lounge besties
Long post worth the read. I have a close friend who flies private jets. He flew Lear 60 for over 15 years and now flies a Gulfstream. This is a copy and paste of his response to me asking him if I thought Ice was the issue.
“We always fly in ridiculous cold temperatures.”
“Usually you get the bad ice when the temperatures are like 0 to negative 20. Believe it or not, there's a point where it gets too cold for ice. At negative 40, you pretty much can't get ice. Besides that, the airplane is equipped with hot wings and hot tail.”
“I flew the Lear 60 for like 15 years. It's a no joke airplane and probably the most dangerous airplane I've ever flown. It has a super critical wing. The airflow has to be perfect. When it gets into rough air and turbulence, it will easily stall. It has stalled on me at 41,000 feet just because we got into some turbulence. (On the way to Palm Beach as a matter of fact. ) we recovered and got her flying again around 30,000 ft. Scary
But here's the problem. This airplane has a terrible coffin corner. (At high altitudes) what that means is, you fly in a really narrow window where if you lose 20 knots, it stalls. And if you gain 20 knots, it goes into overspeed. The air goes supersonic over the wing and it goes into a high speed stall, or worse, a Mach tuck.
So if you stall, the normal recovery is to push the yoke forward to lower the nose and gain air speed again. At low altitude, you can push really hard and recover quickly. But in the 60... if you push hard, it will gain air speed and go into a high speed stall, it doesn't take much. It's very sensitive. The 60 has super short wings that don't make much lift. So then you high speed stall, and the computer is yelling at you... OVERSPEED! So then you pull back and enter a low speed stall. And you go back and forth trying to keep it in the air speed window.
This whole thing is pretty scary. And you should recover basically by using your finger tips using super small corrections. You can't over do it. If you try to muscle the recovery, she gets mad and will become unrecoverable. If she enters a mach tuck. You're done.”
“I taught the Lear 60 to other pilots and mentored them. I told them... this airplane has a soul and it's trying to kill you. Do not ever try to top the weather. Don't go over a thunderstorm. Go around it. If it means a half hour longer, then do it. Going over the thunderstorm will go into turbulence that you might not recover from.
These pilots went right over a thunderstorm. I think it stalled which caused them to drop down into the thunderstorm and ended up in mach tuck. They picked up so much speed that it ripped off the horizontal stabilizer. At that point... chuck Yeager can't fix it.
These same pilots did something similar a couple months ago. ATC got mad at them. They poked the bear one too many times.
This is not a beginner airplane. The Gulfstream would never do this. It has HUGE margins of error and won't stall like that. But then again... it's got HUGE wings. It's a pussy cat.
The Gulfstream is like driving a 4 wheeler. The Lear 60 is like riding a uni cycle.”
Thoughts from my Shanghai field trip
Sentiment among Greater China investors was far worse than I expected. Unlike the US market, A shares had not seen a single rebound since July. Still holding Chinese and Hong Kong AI stocks that had crashed and Korean memory stocks that had collapsed, investors kept pressing me on when memory would bounce back.
Many investors said the AI trade was over, and a few said they doubted whether even 2027 capex, not just 2028, would go ahead as planned.
Overall, the bearish sentiment was so strong that even I, a bull before coming to China, started to waver.
It got to the point where, faced with the bear case that memory prices could start falling in the third quarter of next year, I took it seriously and couldn't refute it.
As the conversations went on, I gradually began to understand where their pessimism was coming from. In the end, they were all asking the same question.
So what comes after AI coding? What will be the next engine of ARR for Anthropic and OpenAI?
Several candidates came up. AI4S, AI chip design, and so on... But none of them clearly convinced investors.
I couldn't offer a clear answer either. Sure, there are rumors that Anthropic has made major breakthroughs in both AI4S and AI chip design, but what we actually want to know is when any of that turns into money.
With AI4S, there could be beneficiaries right away, such as companies that sell lab monkeys or diagnostic kits. In fact, there already are. But while we can trade those names in the short term, what we ultimately want to know is whether Anthropic can make money by developing drugs. That is still several years away.
As for AI chip design, the TAM itself is hard to calculate, and even if you do calculate it, it is hard to monetize. That makes it difficult for the market to believe in.
Honestly, this feels like the hardest market in history. If a company isn't making money, selling makes sense. But right now the market sells even when companies are making money, even when they beat, and even when they say demand will stay solid all the way through 2028 and 2029.
So my recommendation to investors was to buy memory, and within memory, DRAM. Not because I have conviction in the DRAM cycle, but because I see it as a relatively defensive choice within the AI trade.
Memory stocks are now trading close to 1x P/B on 2027 BPS. Historically, memory stocks have staged technical rebounds when they fell below 1x P/B.
And even if the broader market falls further from here, the three DRAM makers will keep buying back their own shares. I thought that would guarantee a certain level of buying and concluded the stocks would have downside support.
Of course, I don't know whether memory stocks can break above their previous highs. That depends on whether sentiment across the broader market recovers. But I think there is a fairly good chance that memory outperforms the rest of the AI trade.
Datarama retweeted
Half of Europe's tourist nights happen in just 102 of its 1,247 regions
#30DayMapChallenge Day 3: Polygons
Datarama retweeted
running a company nowadays feels kinda dystopian
my team is 50+ people in 12 countries, but i talk to my agents more than my human teammates
- voice note to grokbot
- mattpocock/skills /grill-me so the agent interrogates the idea before anything gets built
- spec drops into linear through the linear mcp
- exa + firecrawl agent reads every competitor site, pricing page and changelog
- ideabrowser.com to find ideas and trends
- apify pulls g2, reddit and app store reviews
- prompt: "find the 10 complaints that show up most and quote the users"
- context7 so the agent reads the latest docs before it touches any api
the design
- claude code + paper mcp
- 4 versions of every screen before a single line of code
- pick pieces from 2, it merges them. diverge then converge
- frontend-design skill from anthropic/skills + our own design system skill
- cleanshot screenshots of any ui i like go straight in with "make it feel like this"
the build
- pingdotgg/t3code as the control surface for every agent, from my phone too
- stablyai/orca runs a fleet of agents in parallel, each in its own git worktree
- graphify maps the whole codebase into a knowledge graph so agents navigate instead of guess
- headroom compresses tool outputs before they hit the model, so tokens go 20-95% further
- lsp plugins so the agent sees types and errors like an ide does
- security guidance hook flags risky edits before they land
- supabase mcp for the database, stripe for billing
- github mcp opens the pr, vercel spins up a preview
- loops run overnight, i check what shipped over coffee
the user testing
- lightpanda as the headless browser, way lighter than chromium so you can run hundreds at once
- agents click every flow on the preview like a first-time user
- one persona is impatient, one is confused, one is on a phone
- playwright + chrome devtools mcp for regression tests
the bugs
- sentry mcp catches the error
- agent reads the stack trace and opens a fix pr in under 10 minutes
- human just approves
the data
- posthog mcp cron at 7am on yesterday's funnel
- any step that drops more than 10%, the agent watches 20 session replays and tells me why
- stripe mcp pulls every cancellation reason into one doc every monday
the agents' tools
- orgo so every agent gets its own computer
- agentmail so every agent gets its own inbox
- agentphone so i can call them from the car
- composio connect-apps for every other connector
- akitaonrails/ai-memory, a git-backed markdown wiki so they remember what we decided last week
the skills
- wshobson/agents so skills work across claude code, codex and opencode
- my writing voice
- our design system
- how we write specs
- skill creator so the agents write new skills themselves
some helpful tricks
- one agent per platform. the x agent never touches email
- cheap model for triage, opus for anything that ships
- every output gets scored 1-10 by an llm judge. under 8 goes back
- every agent writes a 5-line log at night so i can read the whole company in 2 minutes
- every agent mistake becomes a new skill, so it only happens once
the team
- humans get version 5, not my voice note which they are happy about lol
- their job is creative feedback, relationships and telling the agents what they got wrong
a human pm reviews the funnel once a week. this reviews it every morning
running a company is now voice notes and inference






























