We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
The direct technical recommendation
Latent Space podcast with Alex Krentsel on RSI for agents
- Content type: Podcast episode.
- Author/creator:Latent Space; Alex Krentsel is the guest, in conversation with swyx.
- Link:Episode on YouTube
- Recommended by: Martin Casado.
- Key takeaway: Casado says he “really recommend[s] a listen,” calls Krentsel “one of the clearer thinkers” on RSI for agents, and praises his work on the topic.
- Why it matters: This is a focused lead for readers studying the specific question of RSI for agents, with a clear expert rationale for prioritizing it rather than a bare episode mention.
An empirical counterweight
An interactive tool to explore historical data on food production, and updated forecasts from the USDA
- Content type: Article with an interactive data tool.
- Author/creator: Hannah Ritchie’s Substack. The tool covers more than 50 years of production, yield, and harvested-area data for crops and animal products.
- Link:Substack article
- Recommended by: Garry Tan, who labels it a “Whitepill timeline cleanser” and points to record crop production as evidence of progress and abundance.
- Key takeaway: Ritchie tracks the USDA’s monthly World Agricultural Production releases. In the August snapshot, corn and wheat are slightly below last year but still the second-highest on record; rice is broadly similar and may set a small yield record; soybean production and yields are on track for records.
- Why it matters: It converts an optimistic macro claim into inspectable data while keeping the important qualification: a “fine” global total can conceal a bad year in particular countries.
Historical perspective
The World of Yesterday
- Content type: Memoir.
- Author/creator: Stefan Zweig, translated by Anthea Bell.
- Link:Recommendation thread
- Recommended by: Patrick Collison.
- Key takeaway: Collison’s endorsement is direct—“The World of Yesterday is amazing.” The linked recommendation highlights Zweig’s observations about Vienna, Germany, and the world between the two world wars as the book’s particular strength.
- Why it matters: It offers a literary, firsthand lens on a society undergoing historical dislocation—useful for readers seeking how large-scale change was experienced from inside ordinary life.
A formative research source
The 2012 “Alex Net” paper
- Content type: Research paper.
- Author/creator: Not named in the interview; Sam Altman says it was done in partnership with his co-founder “Ilia.”
- Link: No URL supplied in the interview.
- Recommended by: Sam Altman, as a formative endorsement rather than a conventional reading-list instruction.
- Key takeaway: Altman says this was the first time an AI approach seemed to him like it might work. He then watched it scale and improve, concluding that—even with a low probability of success—AI was the most important thing he could work on.
- Why it matters: It is a rare, concrete account of a paper changing an AI leader’s estimate of what was possible: from an exciting but seemingly distant idea to a plausible technical path.
A visual recommendation
A short film on another way to live and the built environment
- Content type: Short film.
- Author/creator: @culturaltutor.
- Link:Original short-film post
- Recommended by: David Perell, who calls its premise “the most important thing to accept.”
- Key takeaway: The film asks viewers to examine everyday objects and argues that modern design has become boring even though buildings and ordinary infrastructure can still be interesting and meaningful. It also argues that the aesthetics of the built environment affect how people think, feel, and behave.
- Why it matters: It is a compact prompt to treat the physical environment as an active input into daily life rather than as neutral background.
Answer: Yes — if "resource" is interpreted broadly enough to include software tools, the interview contains one explicit recommendation beyond the AlexNet retrospective mention: Altman calls using the tools his "number one piece of tactical advice," specifically telling people to get fluent with ChatGPT's capabilities rather than use it only like "a better version of a Google search."
AlexNet mention (context, not a recommendation): Altman recalls that "in 2012, the Alex Net paper came out done... in partnership with my co-founder, Ilia," and that "for the first time, it seemed to me like there was an approach that might work."
Other resource details stated in the transcript (mentions, not recommendations):
- GPOSS — an open-source model released "yesterday," "as smart as 04 mini," "runs locally on a laptop."
- Orion / GPT-4.5 — "too big of a model" and "unwieldy to use."
- ChatGPT enhanced memory — launched "earlier this year," a favorite feature that makes the assistant feel like it "gets to know me."
- GPT-5 healthcare — "significantly better at healthcare related queries," "more accurate," and "hallucinates less."
- AlphaFold and GLP-1s — referenced by the interviewer in a medication/protein-folding scenario, not recommended by Altman.
Gaps: No other named paper, book, or external resource is explicitly recommended in the supplied transcript; the only explicit recommendation located is the ChatGPT/tools advice at L251-L252.
Direct answer: the supplied bundle is the Substack post titled "Tracking global food production through a record El Niño" (title is from bundle metadata; the body lines do not state it — its first line is the deck “An interactive tool to explore historical data on food production, and updated forecasts from the USDA” ()). The core data and main argument are verifiable from the body: it tracks how global food production and yields are faring through a strong, potentially record-breaking El Niño, with a one-word global summary so far of “fine”? based on USDA's August release ().
- Data source and cadence: The post says it follows the USDA's World Agricultural Production releases, which are updated monthly with the latest forecasts of production and yields for food products worldwide, and it introduces a small interactive tool showing historical USDA data plus the latest current-year forecast; users can explore more than 50 years of production, yield, and harvested-area data for any crop or animal product ().
- Current global snapshot (USDA August release): Corn and wheat yields and production are down slightly from last year but are the second-highest on record; rice production looks similar, with yields possibly on track for a very small record; soybean production and yields are looking at a record year ().
- Main argument and caveat: The global summary is “fine” — the original text puts a question mark after “fine” — but the footnote explicitly warns this does not mean good or fine everywhere: some countries are too small to register in the global total, and a bad year in one place can be offset by a good year elsewhere ().
- Attribution and date gaps: Bundle metadata lists no author, and no byline or publication date appears in the body. The
hannahritchieSubstack URL in a link () and thehannahritchie.github.iotool URL () are the only line-level attribution evidence; the post's internal timing is “this summer” with the USDA August release ().
Patrick Collison (Stripe CEO, Arc Institute cofounder) recommended Stefan Zweig's memoir The World of Yesterday (trans. Anthea Bell) as "amazing" . His post shares a thread from @EconTalker that recommends the book alongside Joseph Roth's novel The Radetzky March (trans. Michael Hofmann) ; EconTalker calls the memoir fascinating, with "superb" observations about Vienna and Germany between the world wars, and praises The Radetzky March's "exceptional" writing and atmosphere, set during the decline of the Austro-Hungarian empire . Thread link: https://x.com/econtalker/status/2088706230621515974
Sam Altman credits the 2012 AlexNet paper — which he says was done in partnership with his co-founder, Ilya — as the first time he saw an AI approach that might actually work; watching it scale convinced him AI was the most important thing to work on despite low odds . (Research paper; no link provided.)
Garry Tan (@garrytan) recommended a Substack article at https://hannahritchie.substack.com/p/tracking-global-food-usda as a “Whitepill timeline cleanser,” highlighting record crop production year after year as evidence of progress and abundance produced by markets and technology working together for millennia .
Martin Casado (a16z) recommended the Latent Space podcast episode with Alex Krentsel on RSI for agents, calling Krentsel "one of the clearer thinkers on the topic in this space" and saying he's "doing some amazing work on it"; the conversation is with swyx . YouTube: https://www.youtube.com/watch?v=5lFD-34dhqE.