James Cham

Sometimes, all it takes is to find kindred souls to feel like the world is an amazing place.

@joshelman @heeney_luke Roy goes to New York City more often than he goes south of SFO!

@heeney_luke People in SF often think LA and NYC are closer and easier to visit than Palo Alto

On the CalTrain and there’s a guy breaking up with his girlfriend because he can’t handle long distance (he’s SF, she’s Palo Alto)

@ashleevance First 1500 words unlocked?

to paywall or not paywall. that is the question . . . . . it's a 5,000-word story

So much talk on how to become more productive as an individual, esp with AI. Much less about how to work better as a team, using those tools. When well-oiled teams (with superb individuals) >> superb individuals Eg pair programming when using AI agents? Bet it would work well

Some of my favorite people quote Douglas Adams so often it is my first guess that an unusual turn of phrase or observation is actually an Adams reference.

Where does the human economy reside on a cosmic scale?

@ejames_c Congraulations!

One of the best unintended side effects of becoming a new dad is having the wife-approved right to get a proper camera — my first in more than a decade. Yay Ricoh GRIV:

Shout out to @replit engineers and support team keeping everything together as users run armies of agents building everything they ever dreamed of 😅

🚨 New Experiment: Everyone thinks AI firms will look like little companies. A manager model decomposes the task and worker models do subtasks. The manager red-teams, revises, and recombines. A seemingly simple org chart. But when I ran the experiment, the current in-vogue org setup, manager-subagent, cost 4x more and performed worse than letting a rather simple market do the trick. I tested 3 ways to organize multiple AI models: 1. Solo: Onefrontier model does everything itself 2. Hub-Spoke: A "manager" model splits tasks, delegates, red-teams, revises 3. Market: Models bid on tasks, winner gets the job, reputation updates I also tested were 3 types of tasks - Coding, Reasoning and Synthesis. - Coding required most "global state" management, which the solo model did best at. In future @a1zhang's RLM will probably do even better here - Reasoning is the hardest to cleanly decompose, and the market worked the best here - Synthesis too, the market beat hub-spoke as the framing could be ambiguous The reason is, a hub isn't a "manager" as we know it. It's a model that must somehow know: - What the subtasks are - What good recomposition looks like And if either fails, as it does for complex or not-easily-decomposable tasks, competent workers still produce garbage. As we move from coding to letting multi-agent systems do work across the entire economy we'll end up with more not-easily-verifiable tasks with ambiguous settings and uncertain payoffs. In those, we won't be able to use the factory approach to get work done. The Coasean argument is that firms will get smaller, and the smaller firms will transact more, since the organisational premium reduces with AI. But how? Through central hubs, or markets? The fact is, Coase here needs Hayek. Setting up markets is not trivial, as @AndreyFradkin and I looked in our recent paper. Essay: strangeloopcanon.com/p/why-smart-pl…

Metacognitive intelligence in human-AI teams youtu.be/04hgEIJ-uvc?si… by Aaron Benjamin

Theory of constraints and The Goal by Goldratt en.wikipedia.org/wiki/The_Goal_…

A running thread on Artificial Collective Intelligence.

Would understanding the political economy of oil helped us make better decisions after World War II?

@emollick Everything is a diffusion of knowledge process!

Generally, I would say X is not real life, but I am surprised about how often I get asked by executives about which AI lab is winning or what is up with a particular model in ways that indicate that they clearly come from X discussions & rumors (often filtered through LinkedIn)

Replit, turned 10 🎂 To celebrate we’re making it totally free for 24 hours starting at 5am PT. But our work—to make coding accessible for all—goes back to 2011. Watch the highlights from the journey: It’s been an honor to help millions learn & ship. Here is to the next 10!

Tom Malone on collective intelligence in Superminds en.wikipedia.org/wiki/Thomas_W.…

Toyota Production System as the right approach towards automation (and organizational effectiveness in general), best explained by Spear and Bowen in this 1999 piece static1.squarespace.com/static/5356f7d…

At @PWVentures, we've been quietly building AgentCribs, a private community of founders and hackers building custom agentic coding/business/marketing/etc setups and sharing wins and losses. Now, we're opening up the community a bit with our first event on May 6 in San Francisco! I'll be sitting down for a fireside chat with Peter Levine (legendary @a16z investor who sat on our board at @GitHub for many years) to discuss the bleeding edge of AI ideas and realtime insights on where the market is headed. The evening will also include an update from the core AgentCribs team, demos and discussion from people building with agents, and an invitation to participate in ongoing collaboration. If you've been hacking on your own real-world agentic workflows, built your own harness, 10x-d Claude, or just want to know what the future looks like, you belong!

@timhwang The theologian Andy Crouch makes a similar point!

Religion is actually extremely futuristic cultural technology since it represents the accumulated praxis of living with disembodied, inscrutable intelligences

Thank you @semil, @aashaysanghvi_, and @DivyaDhulipala for a wonderful day at LP-GP Alignment Summit. Beautiful location, greater roster of speakers, and the guest list. 🙌🏽

Organizations are already superhuman intelligences. The University of Pennsylvania or Walmart or whatever is far more capable than any human. That is why the focus on AIs as individual productivity tools hits a natural limit, many benefits of AI depend on integration with firms.