Mat Ryer joins Jerod Santo and Adam Stacoviak to work through three concrete problems: Rob Pike receiving AI-generated 'acts of kindness' spam, Microsoft's GitHub consolidation choking open source infrastructure, and Tom Tunguz's 12 predictions for 2026 including agent-first design, vector database growth, and AI costs that may exceed human labor costs.
The Rob Pike incident is documented by Simon Willison and is worth reading on its own. The GitHub monopoly piece, sourced from ploum.net, is the sharpest argument here: it traces how a single platform owning the default social graph of open source is a structural risk, not a preference problem. Tunguz's prediction list adds the forward-looking pressure, naming specific bets on where the infrastructure money goes next year.
The episode runs with a bonus 3 minutes behind Changelog++ membership. Read the three linked sources first: Willison on Pike, the ploum.net GitHub piece, and Tunguz's 2026 predictions. Then listen for the argument about whether AI tooling costs are on a collision course with headcount budgets.
[READ ORIGINAL →]