The Human Scale · Part 2
Chapter 22: The Possibility Space
Structural change is not something that happens to the system while individuals wait. Around the edges of the system described in Part I, experiments are being tested that attack specific failure modes: content provenance standards, deliberate friction, bridging-based ranking, rebuilt local information economics, and deliberative citizen formats. None are proven at scale, and platform incentives push back hard. But each demonstrates the ranking signal, the distribution speed, and the verification cost are design choices, and different choices produce different public squares.
Key insights
- Content provenance standards move the cost curve on verification even when adoption is uneven
- Small frictions (share prompts, forwarding limits, velocity delays) measurably reduce low-quality spread without censoring anyone
- Bridging-based ranking surfaces content when people who usually disagree both find it useful, early trust results cross political identities
- Local information economics is where structural repair and proximity-based accountability meet
- Deliberative citizen formats consistently reduce contempt and improve understanding, but remain small and rarely visible, and their measured effects fade without reinforcement
- Every structural experiment depends on adoption at human scale, oases are where possibility takes root first
The system was built through design choices responding to incentives. Different choices exist. They are being tested, and several of them work at the scales where they have been tried.
Sections in this chapter
- Lowering the cost of proving what's real
- Adding friction back on purpose
- Ranking for bridges instead of reactions
- Rebuilding local information economics
- Formats built for disagreement
- Finding your spaces
- What this chapter does not claim
Reflection
Which of your spaces, a community, a workplace, a local institution, could run on different rules than the broader system, and what structural tool from this chapter would make it work better?