The Magic of Small Databases — Tom Critchlow

Summary

Critchlow imagines a "Substack for databases" — an easy tool for creating, maintaining, and publishing small datasets, with optional paywalling, email updates to subscribers, and built-in data cleanup via ML and web scraping. The core insight: publishing small, curated datasets can be market-making, just as Substack made newsletter publishing market-making for writers.

Relevance

This is the founding vision for [[01-projects/data-marketplace/index]]. The value isn't in big data — it's in small, opinionated, well-maintained datasets that individuals and small teams publish and monetize. The "Substack for databases" framing is the cleanest pitch we've seen for the concept.

It also connects to [[06-reference/concepts/compounding-knowledge]] — a curated dataset that gets updated over time compounds in value. Each addition makes the whole more useful. And the email-updates-to-subscribers model maps to [[01-projects/newsletter/index]] audience-building patterns.

Open Questions

• What's the minimum viable version of this for Ray Data Co? A single published dataset with a subscription layer?
• Which domains have the highest value small datasets? (Real estate, recruiting, niche industry data?)
• How does AI-assisted data cleanup change the economics? Can one person maintain datasets that previously required a team?