In his landmark multi-volume biography of Stalin, Stephen Kotkin shows how totalitarian power worked not just through terror from above, but through millions of everyday decisions from below. Currently a senior fellow at the Hoover Institution after 33 years at Princeton, Kotkin brings both deep archival work and personal experience to his understanding of Soviet life, having lived in Magnitogorsk during the 1980s and seen firsthand how power operates in closed societies. Tyler sat down with Stephen to discuss the state of Russian Buddhism today, how shamanism persists in modern Siberia, whether Siberia might ever break away from Russia, what happened to the science city Akademgorodok, why Soviet obsession with cybernetics wasn't just a mistake, what life was really like in 1980s Magnitogorsk, how modernist urban planning failed there, why Prokofiev returned to the USSR in 1936, what Stalin actually understood about artistic genius, how Stalin's Georgian background influenced him (or not), what Michel Foucault taught him about power, why he risked his tenure case to study Japanese, how his wife's work as a curator opened his eyes to Korean folk art, how he's progressing on the next Stalin volume, and much more.

Beset by colonial controversy, difficult finances and the discovery of a thief on the inside, Britain’s No 1 museum is in deep trouble. Can it restore its reputation? By Charlotte Higgins

Artificial general intelligence — an A.I. system that can beat humans at almost any cognitive task — is arriving in just a couple of years. That’s what people tell me — people who work in A.I. labs, researchers who follow their work, former White House officials. A lot of these people have been calling me over the last couple of months trying to convey the urgency. This is coming during President Trump’s term, they tell me. We’re not ready.One of the people who reached out to me was Ben Buchanan, the top adviser on A.I. in the Biden White House. And I thought it would be interesting to have him on the show for a couple reasons: He’s not connected to an A.I. lab, and he was at the nerve center of policymaking on A.I. for years. So what does he see coming? What keeps him up at night? And what does he think the Trump administration needs to do to get ready for the AGI — or something like AGI — he believes is right on the horizon?

In the first episode of the Solid Podcast, we talked with Dennis Wingo, founder of Skycorp, in the former NASA McDonald’s where he’s been restoring the first images of the moon taken from space.

After an hour of recounting his techno-archaeology exploits — reverse-engineering the arcane analog image-transmission systems that NASA’s engineers developed in the 1960s — Dennis paused and said, “and that’s just one of our history projects.”

That teaser is where we begin today’s episode. Ready to apply modern computing to another analog challenge, Dennis turned his attention to the reboot of the International Sun/Earth Explorer-3, a research satellite launched in 1978 and commended to the heavens in 1997.

NASA decommissioned the equipment for communicating with the satellite in 1999, so Dennis set about reverse-engineering the ISEE-3’s control system and devising a way to communicate with it. In the 1970s, he would have needed custom analog hardware, but now, general-purpose hardware is powerful enough that he could do it all with software.

Wingo’s team built a system on top of an Ettus software-defined radio that could analyze the satellite’s 2.2-gigahertz S band signal; all that remained was to transmit a wake-up call and then listen through a receiver powerful enough to hear the 36-year-old satellite’s signal. The Arecibo Observatory in Puerto Rico did just that.

The “steely-eyed missile men” of the 1970s had done some magnificent engineering work. When Wingo re-established contact with the satellite, it was off course by 3,000 kilometers — after traveling 24 billion kilometers.

The story is an excellent illustration of how software can replace physical complexity — one of the key themes we follow at the Solid conference, and Dennis will deliver a keynote at Solid San Francisco in June, where he’ll talk about his extraordinary combination of techno-archaeology and modern computing.

We’re kicking off our newest series, the O’Reilly Solid Podcast, with an episode recorded in the manager’s office of a McDonald’s at NASA’s Ames Research Center. David Cranor and I (Jon Bruner) visited McMoon’s, as it’s known, to talk with Dennis Wingo, founder of two audacious “techno archaeology” efforts.

In the first episode, we discuss the Lunar Orbiter Image Recovery Project, which has rescued NASA’s first high-resolution images from satellites orbiting the moon. Dennis’ team reverse-engineered the extraordinary analog image transmission system that the satellites used in 1966 and 1967, digitized 14 tons of magnetic tape, and interpreted them to compose imagery at vastly higher resolution than NASA was originally able to recover from the satellites.

Before the invention of the charge-coupled device (CCD), collecting and transmitting images was an electro-mechanical enterprise. The process required to get images from the moon to the earth highlights the ingenuity of NASA’s early engineers — and the relative ease of working with electronics today, when crossing between physical and virtual is straightforward.

Narrated by David Duchovny, Searching for Hobey Baker re-contextualizes and brings to life the story of one of the greatest college athletes who has largely been lost to history.

After Hobey Baker makes his triumphant Princeton hockey debut against Williams, we travel back to his early days attending St. Paul’s School in New Hampshire. He develops a reputation as both an athletic phenom and a kind, generous sportsman. After the economic crash of 1907, his father struggles to send him to college at Princeton where he becomes a two-sport star in football and hockey. After graduation, Hobey embarks on a summer motorcycle trip around Europe – a grand experience interrupted by the ominous clouds of conflict circling the continent.

Dharmesh Shah is the co-founder and CTO of HubSpot (currently valued at $30 billion) and one of the most fascinating founders I’ve ever met. Dharmesh is the keeper of HubSpot’s Culture Code, built ChatSpot (an AI chatbot built on top of HubSpot CRM) and a game called WordPlay (which grew to 16 million users), and also founded and writes for OnStartups, a top-ranking startup blog and community with more than 1M members. He’s also invested in 100+ startups including OpenAI, AngelList, Coinbase, and Dropbox. In our conversation, we discuss

• The biggest lessons he has learned from building HubSpot

• The importance of leaning into your strengths

• Dharmesh’s data-oriented approach to public speaking

• How he developed HubSpot’s culture code

• The decision-making process at HubSpot

• His contrarian approach to building products

• Why founders and product teams are all fighting the second law of thermodynamics

• How “flash tags” can save your teams time

• How to decide what ideas are worth investing in

From copyright violations to environmental concerns to the looming threat of the singularity, AI is a hot-button topic these days. Paul and Rich talk through many facets of this conversation, and discuss how they think about the AI components of Aboard. Plus: A little roleplay in which we learn that Paul thinks Aboard is an earnest mid-century cartoon character.

To better understand how neural networks learn to simulate writing, researchers trained simpler versions on synthetic children’s stories.

Demis Hassabis - Scaling, Superhuman AIs, AlphaZero atop LLMs, Rogue Nations Threat Wednesday 28 February 2024 https://www.dwarkeshpatel.com/podcast Open in Pocket Casts Share Here is my episode with Demis Hassabis, CEO of Google DeepMind

We discuss:

Why scaling is an artform

Adding search, planning, AlphaZero type training atop LLMs

Making sure rogue nations can't steal weights

The right way to align superhuman AIs and do an intelligence explosion