We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
Top pick: Elizabeth Dias's NYT piece on Anthropic, recommended by Bill Gurley
Gurley said he was "grateful" that Elizabeth Dias of The New York Times wrote this article. It covers how Anthropic thinks about its model, software consciousness, and the model's role relative to religion. His reason is practical. If you are a customer or partner of the company, press evaluating it, or a government regulating it, "it is important to know how 'they' think about their models." He says the company's approach is "very different from any software company that ever existed." He doesn't say whether he agrees with it: "Judge for yourself... but be aware and informed" .
Tim Ferriss: Small Giants by Bo Burlingham
In a thread arguing that "everything big starts small and focused," Ferriss asks: "Do you really want to build and manage a big company?" For most people, he says, it's "an all-consuming taskmaster." He then recommends Small Giants "for some fantastic examples of companies that choose to be the best rather than the biggest" .
Guillermo Rauch: Jeff Dean's latency numbers and a testing-to-proofs essay
In a video on his agentic engineering workflow, the Vercel CEO recommends Jeff Dean's "Latency numbers every programmer should know." He says he has shared the table "so many times with engineering teams" because it gives "rough senses of scale and orders of magnitude." He adds: "you don't have to memorize everything" .
He also calls an essay on "the spectrum of testing to proofs" one of his "all-time favorite essays on engineering." The essay compares tests, type systems and proof systems on a simple algorithm. It shows that code covering every edge case needs a lot of boilerplate and compiler help . He doesn't name the author or title.
Chamath Palihapitiya: Andrej Karpathy on reading LLM output
Chamath shared Karpathy's post with the comment "The future is taste and judgement" . Karpathy suggests ways to make model output easier to understand:
- asking for writing in the ASD-STE100 controlled language
- diagrams
- interactive HTML pages
- bespoke explainer videos, the format he's "most bullish on"
His argument is that as LLMs do more of the legwork, more of our work moves "into oversight and understanding" .
Jeff Bezos: José Andrés on housing
Bezos called Andrés's post a "100% correct diagnosis and prescription," adding: "it's not just in Spain" . Andrés, writing in Spanish, says Spain doesn't lack decrees; it lacks homes. His prescription is to build much more: more available land, permits in months rather than years, and more density .
Fred Wilson's old AVC posts, as used by USV's Nikhil Raman
On a video with Wilson, Raman said he searched his texts and found that he regularly sends Wilson's blog posts to people. One was a roughly 13-year-old post on Don Valentine's 2x2 of founder quality against how easy founders are to work with. Raman sent it during a debate about whether today's founders have "softened." Wilson's point from the post: hard-to-work-with great founders are where "we make all our money" . The second was a decade-old post on career investors versus founder investors. Raman sent it while thinking about how a non-founder VC earns credibility with founders .
Shorter mentions
- No Priors with Fractile founder @goodwin_ml, from Sarah Guo (her own show). She called it "really fun." Topics include how to win the AI chip market structurally, memory bandwidth and new hardware scaling laws . Full interview.
- KQED's report on California legalizing balcony solar, from Matt Mullenweg. He called it "Another win for solar!" and said he hopes Tesla offers products for it . Link
- A MilHistNow post on the 1942 V-2 test, from Paul Graham. He said it explained why he'd never considered what the first man-made object in space was: "We don't talk about that milestone" .
- Guillermo Rauch recommends Jeff Dean’s Google resource “Latency numbers that every programmer should know.” Its table gives rough orders of magnitude for engineering-system latency, and Rauch says understanding these system concepts is useful even if programmers need not memorize every number.
- Rauch calls an essay about “the spectrum of testing to proofs” one of his all-time favorite engineering essays; its author is not named in the exchange. The essay compares tests, type systems, and proof systems for correctness, and discusses the boilerplate and compiler support needed to cover edge cases.
- Rauch praises 3Js eval.com, whose creator is not named in the exchange, as a site for voting between model-output alternatives (for example, which grass looks better). He uses it to illustrate the value of judging AI-generated 3D against a high bar, not just for novelty.
- Nikhil Raman said he had sent Matt Mandel an older Fred Wilson blog post—about 13 years old—quoting Don Valentine’s 2×2 of founder quality and how easy or difficult founders are to work with. It came up in their debate over whether tech founders today seem uniformly “nice.” Wilson recalled Valentine’s view that the great founders who are hard to work with are where “we make all our money,” connecting the idea to founders following their own instincts.
- Raman also said he had sent someone a roughly decade-old Wilson post on career investors versus founder/operator investors. He brought it up while considering how, as an engineer but not a founder entering venture capital, to build founder empathy and the ability to help; the discussion’s takeaway was that either investor path can work, while firsthand founder experience can lend advice credibility.
Paul Graham shared a MilHistNow post with an embedded video describing Nazi Germany’s 1942 V-2 test fire as the first man-made projectile to leave Earth’s atmosphere and reach space . He said the post explained why he had never considered what the first man-made thing to reach space was, adding that “We don’t talk about that milestone” .
@saranormous called a No Priors episode with Goodwin, founder of Fractile, “Really fun,” highlighting discussions of AI-chip-market strategy, AI for chip design, memory bandwidth, and new hardware scaling laws . The show’s stated takeaway is that accelerating experiments is becoming as important as accelerating AI itself, from chip design to scientific discovery . Episode post.
Patrick O’Shaughnessy promoted his conversation with Ben Thompson, calling Thompson one of his favorite business thinkers and inviting readers to “Enjoy!” (conversation) . In an excerpt, Thompson argues that AI may currently generate more incremental monetization for Google and Meta through improved advertising than AI labs get from selling AI; their ad marketplaces let them test generated content, and even a few percentage points of improvement could be worth billions .
Tobi Lutke recommends attending or tuning in to endurance racing at Road Atlanta, starting noon EST the following day, saying it was fun to introduce old friends to the sport . The linked context identifies the event as Motul Petit Le Mans, a 10-hour endurance race in Atlanta .
Brian Armstrong shared an AiFi video clip, captioning it “AiFi gives everyone a financial advisor.” The linked Base post features Armstrong and @everythingempty of Virtuals.io discussing Butler’s thesis, AiFi’s future, and adoption of tokenized stocks.
Brian Armstrong urged followers to try Coinbase’s AI-agent workflow by pasting Connect to https://coinbase.com/skill.md into a preferred agent; he said it can access portfolios, trade crypto, stocks and derivatives, and pay for premium data with x402. He linked Coinbase’s “AiFi: Start Trading With Your First Five Prompts” guide, a five-prompt walkthrough for moving from a market hunch to a structured trade.
- Packy McCormick recommends Not Boring’s Deep Dive on Airbound, adding, “We are going to fly.”
- In his Starship coverage, he points readers to SpaceX’s post about the orbital flight, saying, “just take a look at these beauties.”
Tim Ferriss invited listeners to enjoy his new podcast episode, “How to Simplify Your Life,” featuring Chip Conley, Gregory McKeown, @gtdguy, Elizabeth Gilbert, and Oliver Burkeman. Listen on Spotify or Apple Podcasts.
Sarah Guo calls a No Priors episode with @goodwin_ml, founder of @fractile_ai, “really fun.” It covers AI chip-market strategy, AI chip design, frontier-lab requirements, memory bandwidth, and hardware scaling laws. The full interview is available on YouTube or podcast platforms: YouTube.
Bill Gurley recommends Elizabeth Dias’s New York Times article about Anthropic’s thinking on its model, software consciousness, and the model’s relationship to religion. He says customers and partners, press evaluating the company, and government regulators should understand its approach because it differs from that of any software company before, while urging readers to judge for themselves and stay informed. Article
Aravind Srinivas praised AskPerplexity’s Computer demo: a searchable 3D map of nearly 26,000 restaurants and cafes across New York City, with searches by dish or neighborhood and the ability to step inside venues. He said projects like this can be left to an agent for hours to produce something great, with creativity as the limit.
Chamath Palihapitiya shared Andrej Karpathy’s post, adding that “the future is taste and judgement.” Karpathy’s post suggests using LLMs to create controlled-language writing, diagrams, interactive webpages, and bespoke explainer videos; he says explainer videos are the format he is most bullish on.
Chamath Palihapitiya enthusiastically recommended The All-In Podcast’s “Super Intelligence Pod”. The episode covers the Trump Super Intelligence Summit and how the AI Safety Accord was created and signed, as well as data-center national security and the cybersecurity arms race .
Scott Belsky shared a clip posted by @yzycrave of Ye performing “Runaway” with Pusha T live on SNL, saying he remembered being mesmerized by it; the post says the performance was 16 years earlier .
Tim Ferriss recommends Small Giants by Bo Burlingham for its examples of companies that choose to be the best rather than the biggest, in a discussion of the trade-offs of building and managing a large company .
Matt Mullenweg endorsed companies making “more art,” replying to @eglyman’s account of a company putting on a Broadway musical about accounts payable. He urged combining efforts to fix things with making them “beautiful and sing”; his link was https://x.com/eglyman/status/2106095498704851419.
Elon Musk endorsed SpaceX’s post showing Starlink V3 deployment, calling it a “Beautiful view”; the post says Starship’s first orbital flight delivered 26 Starlink V3 satellites and includes images.
Jeff Bezos endorsed José Andrés (@chefjoseandres)’s X post on Spain’s housing crisis, calling its diagnosis and prescription “100% correct” and saying the point applies beyond Spain. Andrés argues that the problem is a shortage of homes, not decrees, and calls for substantially more construction, including more available land and faster permits.
Weekly Dose of Optimism #213

Cover image generated by Runway (opens in new tab).
Hi friends 👋,
Happy Friday and welcome to our 213th Weekly Dose of Optimism.
You might notice that the title card looks a little different than it normally does, and you might also notice that our fifth story comes with its own short film. Good eye. Over the coming quarter, we’re partnering with our friends at Runway (opens in new tab) to bring the stories we tell here in the Dose and in not boring to life. I am… not a filmmaker, but I am a writer, and it’s wild to be able to make something that feels like a pro video in just a few prompts, and I’ll get better at it as we go.
If you haven’t yet, check out our Deep Dive on Airbound (opens in new tab). We are going to fly.
Let’s get to it.
This week’s Weekly Dose is brought to you by… Reducto (opens in new tab)
The Only Document Parsing Product You’ll Need
Reducto (opens in new tab) is the document ingestion layer trusted by AI teams at Scale AI, Harvey, Toast, and Vanta to reliably turn complex files into structured, grounded data for production AI systems at scale. Processing more than a billion pages per month, Reducto (opens in new tab) has built its reputation on the long tail of difficult documents that break conventional OCR: dense tables, unusual layouts, handwriting, low-quality scans, strikethroughs, and formatting-dependent information.
Its new frontier parsing model, r-1 (opens in new tab), combines layout detection, reading order, tables, formatting, grounding, and granular citations in a single request. It reduces parsing errors by up to 20%, improves latency at high volumes, and costs 1¢ per page all-in.
Already using another parser? Apply for up to \$5,000 in migration credits (opens in new tab) and hands-on support to benchmark r-1 against your current stack on the documents that matter most.
Fourteenth time’s the charm!
On Monday, Starship reached orbit for the first time and once there, spat 26 Starlink V3 satellites out of a Starlink-deploying slot.
Starship’s previous flights reached space but followed suborbital trajectories that brought them back down. This time, Starship fired an engine to enter orbit, released its cargo, and later fired an engine to leave orbit before splashing down.
Were there some things that need fixing? Of course there were. An engine shut down early during ascent, and SpaceX shortened the mission as a precaution, bringing the ship down at a prearranged location in the northern Pacific.
But they keep getting better. They’ll fix the engine and catch thousands of these Statue of Liberty-sized rocketships with giant chopsticks, make them rapidly reusable, and keep sending them up to blanket the world with internet, and then maybe data centers and power and telescopes and drug factories (opens in new tab) and whatever else might make sense to move to space when launch is cheap and constant.
In the meantime, just take a look at these beauties.
Joe Gebbia, RISD grad and co-founder of Airbnb, runs the US Government’s National Design Studio. It shows. The government made a good website, a great one even, in “the first step toward replacing thousands of outdated government websites with one product.”
For now, you can ask questions about government services and get answers grounded in government sources. GSA says (opens in new tab) it brings together information from more than 29,000 government websites, which is an absolutely ridiculous number of government websites.
Adam Bhaloo, who’s on the NDS team, tweeted (opens in new tab) that “Next, we’re building features that let you renew your passport, enroll in Medicare, and book a campsite at a national park, and so much more, all in one place.”
I can’t wait. I have a friend who’s planning a trip to Paris with a bunch of family members and is stressing out about being able to get her passport renewed in time. The process is ridiculously complex; I once had to sleep in my car in Washington, DC to get a new passport for a trip. I hope they bring this miracle technology (modern websites) to the state-level next so that I never have to interact directly with the DMV again, either.
The government works for us, and we should be able to tell the government what we need and have its software figure out how to get it done.
God Bless America.gov.
A couple of years ago, Matic (opens in new tab) co-founder & President Mehul Nariyawala (opens in new tab) came to my house in Brooklyn on a Saturday to help set up an early version of the Matic in my house. He brought his laptop, too, and he walked me through the company’s master plan.
Build an incredibly useful robot in a big market, use that to learn and generate revenue, and then build on that learning to build more in-home robots that can do more and more useful stuff to give people their time back.
Over the past couple of years, they’ve executed like crazy on the first part, putting robots in 16k homes covering 500M square feet. This week, he unveiled that plan to the public.
The basic argument is that starting with robots that can do everything it the wrong order, and that instead, you need to start with what it can do really well, collect data, learn how to navigate a home and all the edge cases in it, earn trust, and then grow.
The next phase for Matic will be to add manipulation, to give robots the ability to act on the things they’ve learned to see. The video shows both a non-humanoid looking thing with arms picking things off the floor, and a humanoid folding laundry. Presumably, they’ll build the former first, and then the latter.
As someone who would rather spend time with my family than doing chores, this vision can’t come to life soon enough.
Speaking of family-friendly machines…
I’ve been in California this week, away from my kids, and whenever I’m away from them, I get the urge to call them. Of course, I can call Puja and FaceTime with the kids, but there are random times throughout the day when I think, “I wish I could just call Dev or Maya right now.”
But what are we going to do? Give our kids cell phones and get them hooked early? No shot.
Sam Terris is trying to change that with Freckle, a phone for kids that will cost \$199 plus a \$25/month cell subscription (opens in new tab).
Freckle runs its own ClayOS, with parent-approved calls and texts, maps, and tools for exploring. Kids can photograph something with it and then ask AI to explain what they’re looking at. Parents can set up real-world “Quests” for their kids too. Freckle doesn’t have open web, social media, or an app store, so there isn’t a risk of kids getting sucked in (unless they figure out how to hack it, in which case, good on them).
In the launch video, it shows two kids bumping their Freckles to turn them into walkie-talkies, which I would have gone bananas for as a kid.
The philosophy seems to be to help kids get outside and explore - they have a map, a way to ask questions about the world, and a way to keep in touch with parents so parents don’t have to keep their eyes on them at all times. It’s just a pre-order phone, and there’s a lot of societal cruft to be figured out, but it seems like a good step back towards letting kids be kids (opens in new tab).
Nature; video created with Runway (opens in new tab)
Saving the wildest story for last…
Researchers at Harvard and Brigham and Women’s Hospital found that transplanted hearts shifted toward the biological age of their recipients. Older hearts showed signs of rejuvenation in younger bodies, while younger hearts showed accelerated aging in older ones.
They first studied mice, using patterns of chemical tags on DNA to estimate the hearts’ biological age and found the effect. Then, they examined biopsies from 11 human heart-transplant recipients with substantial age differences between themselves and their donors and found a similar pattern. In a larger analysis of patient records, several measures of function, including exercise capacity, were associated with the recipient’s age rather than the donor’s. Which is nuts!
Of course, this is a preprint on a small human biopsy sample, and changes in an aging clock don’t mean all of the heart’s accumulated damage has disappeared. The mouse experiments also didn’t show that receiving a young heart rejuvenated the rest of the body, just the heart itself, which makes sense.
But come on! This is not a result I would have expected, and it has both practical and more profound implications.
Practically, it might mean that doctors can use older donor hearts in more patients, which would expand the donor pool and save more lives.
Profoundly, it implies that some aspects of an organ’s age seem responsive to the body around it. Researchers may be able to use what they learn here to understand what it is about younger bodies that rejuvenate older organs and potentially reproduce whatever it is to reverse aging more broadly.
May we all be young at heart.
EXTRA DOSES: Science Breakthroughs, AMD Buys World Labs, General Intuition and Varda Raise
- Packy McCormick recommends Not Boring’s Deep Dive on Airbound, adding, “We are going to fly.”
- In his Starship coverage, he points readers to SpaceX’s post about the orbital flight, saying, “just take a look at these beauties.”