“The age of the idea guys has begun.”

Justin & Gigi vibe about Zig.

Articles mentioned:

• LLMs as a tool for thought by Amelia Wattenberger
• Micropayments and Mental Transaction Costs by Nick Szabo
• How our interfaces have lost their senses by Amelia Wattenberger

Talks mentioned:

• The Art of Bitcoin Rhetoric by Bitstein

Books mentioned:

• Human Action by Ludwig von Mises
• Working in Public by Nadia Eghbal

In this dialogue:

• nak
• Files
• SyncThing (and how it BitTorrent Sync became Resilio Sync)
• Convention over configuration
• Changes & speciation
• File systems as sources of truth
• Vibe-coding shower thoughts
• Inspiration and The Muse
• Justin's LLM setup
• Tony's setup (o1-pro as the architect)
• Being okay with paying for LLMs
• Anthropomorphising LLMs
• Dialog, rubber-duck debugging, and the process of thinking
• Being nice and mean to LLMs
• Battlebots & Gladiators
• Hedging your bets by being nice to Skynet
• Pascal's Wager for AI
• Thinking models vs non-thinking faster models
• Sandwich-style LLM prompting, again (waterfall stuff, HLDD / LLDD)
• Cursor rules & Paul's Prompt Buddy
• Giving lots of context vs giving specific context
• The benefit of LLMs figuring out obscure bugs in minutes (instead of days)
• The phase change of fast iteration and vibe coding
• Idea level vs coding level
• High-level vs low-level languages
• Gigi's "vibeline"
• Peterson's Logos vs Vervaeke's Dia-Logos
• Entering into a conversation with technology
• Introducing MCPs into your workflow
• How does Claude think?
• How does it create a rhyme?
• How does thinking work?
• And how does it relate to dialogue?
• Gzuuus' DVMCP & using nostr as an AI substrate
• Language Server Protocols (LSPs)
• VAAS: Vibe-coding as a service
• Open models vs proprietary models
• What Cursor got right
• What ChatGPT got right
• What Google got right
• Tight integration of tools & remaining in a flow state
• LLMs as conversational partners
• The cost of context switching
• Conversational flow & how to stay in it
• Prompts VS diary entries
• Solving technical vs philosophical models
• Buying GPUs & training your own models
• Training LLMs to understand Zig
• Preventing entryism by writing no documentation
• Thin layers & alignment layers
• Working in public & thinking in public
• Building a therapist / diary / notes / idea / task system
• "The age of the idea guys has begun."
• Daemons and spirits
• Monological VS dialogical thinking
• Yes-men and disagreeable LLMs
• Energy cost vs human cost
• Paying by the meter vs paying a subscription
• The equivalence of storage and compute
• Thinking needs memory, and memory is about the future
• Nostr+ecash as the perfect AI+human substrate
• Real cost, real consequence, and Human Action
• The cost of words & speaking
• Costly signals and free markets
• From shitcoin tokens to LLM tokens to ecash tokens
• Being too close to the metal & not seeing the forest for the trees
• Power users vs engineers
• Participatory knowing and actually using the tools
• Nostr as the germination ground for ecash
• What is Sovereign Engineering?
• LLVM and the other side of the bell-curve
• How nostr gives you users, discovery, mircopayments, a backend, and many other things for free
• Echo chambers & virality
• Authenticity & Realness
• Growing on the edges, catering to the fringe
• You don't own your iPhone
• GrapheneOS
• WebRTC and other monolithic "open" standards
• Optimizing for the wrong thing
• Building a nostr phone & Gigi's dream flow
• Using nostr to sync dotfile setups and other things
• "There are no solutions, only trade-offs"
• Cross-platform development
• Native vs non-native implementations
• Vitor's point on what we mean by native
• Does your custom UI framework work for blind people?
• Ladybird browser & how to build a browser from scratch
• TempleOS
• Form follows function & 90's interfaces
• Lamentations on the state of modern browsers
• Complexity & the downfall of the Legacy Web
• Nostr as the "new internet"
• Talks by Ladybird developer Andreas Kling
• Will's attempt of building it from scratch with Notedeck & nostr-db
• Justin's attempt with rust-multiplatform
• "If it doesn't have a rust implementation, you shouldn't use it."
• Native in terms of speed vs native in terms of UI/UX
• Engineer the logic, vibe-code the UI
• From Excalidraw to app in minutes
• What can you one-shot?
• What do you need to care about?
• Pablo's NDK snippets
• 7GUIs and GUI benchmarks for LLMs
• "Now we're purpose-building tools to make it easier for LLMs"
• "Certain tools really make your problems go away."
• Macros and meta-programming
• Zig's comptime
• UNIX tools and pipes
• Simple tools & composability
• Nostr tools for iOS & sharing developer signing keys
• Building 10 apps as one guy
• Simplicity in a community context
• Most people are on phones
• Most people don't install PWAs
• Zapstore & building our own distribution channels
• Web-of-trust and pushing builds quickly
• Improving homebrew by 10x
• (Micro)payments for package managers
• Guix and bitcoin-core
• Nix vs Guix
• Reproducible builds & web-of-trust
• Keet vs "calling an npub"
• Getting into someone's notifications
• Removing the character limit was a mistake