Problems

• Public thinking and writing culture
  • How can we find the thinkers using AI? ie. people who write blogs
    • I’d like a search engine which indexes only on blogs.
    • In many cases it’s more accurate to model the world as 500 people than 8 billion. 500 Inc would be a funny name for this.
  • How can we rebuild blogging culture?
    • Blogs and public writing seems to be on a downtrend?
  • How can we build the .pdf of a website? There isn’t a good portable web container format yet. Websites are too dynamic.
• Software enshittification.
  • How can we revert to software that doesn’t enshittify with constant marketing popups and slowness?
  • In the past 15 years, it’s clear software has regressed. Open-source is better, faster, and free. For everything but multi-device sync.
• Software cost rises.
  • Meetup.com and Couchsurfing.com has died because people want free meetup software. The rise of Luma.
  • How can we build alternatives where they live? After all - it’s very simple websites, cost-wise. You could run it on a single box.
• Recommendation systems
  • Recommending live music gigs based on your Spotify profile using music embeddings
  • Can we build open-source recommendation systems?
    • Imagine WordPress blogs and pingbacks, instead of closed-source twitter funnels.
• Email.
  • Can we fix email so it’s not controlled by only Google, Yahoo, Microsoft and a handful of “approved” email domains.
  • Right now you can’t run your own network and get through spam filters.
• Governments:
  • One of the things which baffles me is that there’s no written website that describes all of the australian government and its functions, branches, etc.
  • How can we optimize a system if we don’t even have it written down?

Trends.

• From public forums back to private group chats and big centralized feeds.
• Compression of information. Many books nowadays - too long for attention spans. This is maybe good. Back then there was no competitive pressure on length.
  • What if newsfeeds were built around educative material?

Demands.

• https://www.financialdatasets.ai/ for crypto

Opportunities.

What are people not really getting?

• AI-native software. Most software, AI is just bolted-on. Rarely do I try something (aside from ChatGPT), where fundamentally - the interface feels like extended intelligence, rather than “a faster horse”.

Ideas.

• Machine learning models which look like Docker containers / The Matrix-style hotloaded knowledge (reference)
  • alpine image is just English language
  • image on top is “human reasoning”
  • image on top is hotloaded domain-specific knowledge
• HTML/Markdown which includes:
  • Tags for reading/writing from blockchains (open databases).
  • Tags for reading/writing from prediction markets (open prediction data sources).
• The big question of AI is Douglas Adam’s. We have an LLM that knows every book ever written. But what do we even ask it? Where do we start to learn?
  • need a generative wiki with a personalised recsys to explore topics
• IDE for computational futures modelling:
  • AI can interpolate over statements ie. “tokens per second will increase 100x over the next 10yrs” -> “year 5, tokens/s is 4x what it is today”
  • people complain about the future being super uncertain, but I disagree. it’s actually quite clear what might change.
  • it’s just many things will change. and we don’t have tools for modelling this.
  • it looks like some sort of connected proposition graph system.
  • unlike ChatGPT, it’s not a Q&A interface. Because we don’t know what the questions are.
  • it’s more an exploratory - these trends will continue, connect them to these dots over here.
  • for example, if AI really does get 100x better, then we can rewrite a web browser. and build a 2x cheaper browser -> good for web browser rendering / scraping business.
• How are we going to preserve our consciousness?
  • Where will AI agents store their memories? It won’t be S3 or Ethereum.

Questions.

• “If trees are bio-architects, how can we borrow their electro-chemical growth logic to print self-healing buildings in situ?”