“Is San Francisco Back?”
The opening debate.
Yes: Mike Solana, who edits one of the most-read tech dispatches in the city and arrives armed with the statistics of recovery.
No: Sam Kriss, the British critic who flew to San Francisco to tell a clubful of San Franciscans their city has been replaced by a corporate simulacrum. Numbers vs, character, in the city that purports to invent the future.
The Clash
Solana arrives with data: violent crime down 20%, murder down 13%, tax revenue set to rise by $1.4 billion. These numbers are significant.
Kriss doesn’t dispute them. He disputes what they prove. A city’s recovery isn’t visible in a spreadsheet, Kriss holds; it’s proved by whether its streets produce culture, argument, strangeness. San Francisco’s current iteration has produced driverless cars and opaque B2B advertising. The Hippies and the Beats would disown it.
The House must decide: is a city its statistics, or its soul?
Solana has the numbers and he’s justified in deploying them; the city has measurably improved on every metric he cites. But Kriss has something the numbers can’t capture: the texture of a place, the feeling of its streets, the palpable strangeness that made San Francisco San Francisco. A room full of San Franciscans listens to a British critic call their city a “Playmobil toy town” and applauds loudly. Solana calls the applause absurd masochism. He may be right. It doesn’t change the vote. Solana proves the city is alive again. Kriss convinces us that it isn’t yet itself.
“Should Robots Take Our Jobs?”
Yes: Noah Smith, the economist who sees in automation the same liberation that lifted humanity up from backbreaking farm work and into something better.
No: Brian Merchant, who wrote the book (literally, or at least a book) on the Luddites and argues the original sin isn’t the machine, it’s who controls the machine. History vs. the present tense, with disruption on both sides of the ledger.
The Clash
Smith points to agriculture: once 90% of all humans worked on farms; now almost none do, and the freed labour built the modern world. The pattern always resolves in abundance.
Merchant’s reply: the Luddites weren’t anti-technology; they were protesting factory bosses who weaponized machines to destroy skilled workers’ livelihoods. The difference between automation-as-liberation and automation-as-extraction isn’t the machine; it’s who captures the gain.
The House must decide whether this time looks like the first story or the second.
Smith has history behind him: every previous wave of automation produced net abundance for most people.
Merchant has the present: 42% of AI pilots abandoned last year, creative workers already displaced, gains flowing upward to a very small number of people. Smith has the long view. Merchant has the current view. The House couldn’t choose.
“Should We Have Superbabies?”
Yes: Josie Zayner, the biohacker/entrepreneur who has already edited her own genome and cannot fathom why everyone else is still waiting.
No: Kristen V. Brown, the science journalist who argues that the genome is not software, that the risks are not theoretical, and that what we might lose in the editing is harder to name than what we’d gain. The Nietzschean against the humanist.
The Clash
Zayner’s argument is visceral: there are roughly 10,000 single-gene diseases, millions suffer from preventable genetic conditions, and the technology to address them exists now. Every year we don’t use it, people die. Now that we know this and don't act, those deaths are on our hands.
Brown’s reply: the motion asks “Should we design our children,” not merely cure them, and those are not the same thing. Most traits are polygenic, shaped by thousands of genes and their environmental interactions. Off-target effects are real. Mosaicism is real. And heritable edits don’t just affect the person born with them; they change the genome of every descendant they will ever have.
The House must decide whether the urgency of suffering outweighs the risk of irreversibility.
Zayner’s case is the most emotionally compelling in the room: people suffer, the technology exists, and philosophical hesitation has a body count. She is right about that. But Brown finds the gap that sinks the motion. It asks us to “design our children,” not merely cure them, and those are different things. The complexity of polygenic traits, the irreversibility of heritable edits, the compounding inequality of a genetic trust fund across generations: these are not hypotheticals. They are what happens when the promise meets the biology. Zayner brings the urgency. Brown brings the consequences. The House chooses consequences.
“Can You Teach an AI Taste?”
The last word of the debate. The empiricist against the romantic.
Yes: Jasmine Sun, who spearheaded the event and argues that taste is just training by another name.
No: Robin Sloan, the novelist who argues that taste requires risk, courage and consequences.
The Clash
Sun argues that taste is not a mystical gift but an acquired capacity (Pierre Bourdieu said as much in 1979) and if humans develop discernment through exposure, education and love, then a model trained on all of human culture is learning taste by the same mechanism.
Sloan’s reply: taste comprises more than just pattern recognition. Its most essential component is commitment. A model that generates a billion songs has not truly chosen any of them. The decision is the thing. An AI’s choices are never final, and never costly. Something that costs nothing to decide means nothing once decided.
At the opening of the second round, each Disputant pulls ace from his and her sleeve.
Sloan reaches for his rhetorical leg sheath and produces: Move 37.
Now it's Sun's turn to pull something from her ankle holster of logic, and it's... also Move 37!
The House must determine whether taste resides in its documented patterns or in the price of choosing.
Sun makes the best empiricist case available: taste is learned, AI learns, therefore AI can have taste. Her logic is clean and hard to dismiss. But Sloan pinpoints the element Sun's framing leaves out: finality. We don’t just choose; we give something up in making the choice, and we live with what we pick. An AI that can output a million variations, depending on the prompt, has not chosen between them in any sense that matters. The choice was still always going to be made by a human. Sun shows how AI can learn the grammar of taste. Sloan wins with the argument that taste is a function of human limitations.