ZAOfractal
Search cmd K

01 - Fractal Democracy: Theory

Origin

Fractal democracy began as a critique of voting. Daniel Larimer - creator of BitShares (2014), Steem/Hive (2016), and EOS (2018) - turned to governance theory after three platform launches. He published “More Equal Animals: The Subtle Art of True Democracy” on February 20, 2021 (verified on Amazon, Goodreads, Apple Books, Google Books, Kobo).

The critique

Larimer argues mainstream governance fails in four ways:

  1. Large-group voting breaks mathematically. When millions vote, any individual vote has near-zero influence. Rational ignorance and voter apathy follow.
  2. Capital corrupts governance. Token-weighted voting (one token = one vote) recreates plutocracy. Whoever accumulates the most tokens wins. This contradicts democracy.
  3. Representatives lose accountability. Fixed-term officials get captured between elections.
  4. Direct democracy does not scale. Asking a billion people to vote on every law is impractical.

The solution: sortition + nested small groups

The algorithm:

  1. Randomly sort all citizens into small groups (3-6 people; 6 is ideal because it produces 15 pairwise comparisons).
  2. Each group spends 1-2 hours in deliberation and selects a representative by consensus.
  3. Representatives form new groups (also of 6) and repeat.
  4. Continue through 3-4 rounds until a final decision-making body of ~6-12 people emerges.
  5. The final group executes.

Properties:

  • Every person is in a human-scale conversation where their voice genuinely matters.
  • Deliberation beats voting - groups discuss pros and cons, not just tally.
  • Scales to any population: 100 people = 2-3 rounds, 10,000 = 3-4 rounds, 1 billion = 5-6 rounds.
  • Representation is granular: each person evaluates only ~5 peers, not 10,000 candidates.

This is the fractal structure - a self-similar pattern that repeats at every scale.

Why Fibonacci scoring

The Fibonacci sequence (1, 1, 2, 3, 5, 8, 13, 21, 34, 55, …) is the reward curve for ranked output. Larimer chose it for four reasons:

  1. ~62% growth per rank - each level is the golden ratio (phi ~1.618) times the previous level. Meaningful incentives for top performers without winner-take-all.
  2. Softer than Pareto. The 80/20 distribution is mathematically harsh. A 6-rank Fibonacci (55/34/21/13/8/5) gives the top 2 ranks ~57% of Respect and the bottom 4 ~43%. Fairer than 80/20, still rewarding excellence.
  3. Measurement theory. Larimer treats participants as imprecise instruments measuring peer contribution. Human ranking has wide error bars. Fibonacci is forgiving of misranking by 1-2 places, while a steep exponential (32/16/8/4/2) makes mis-rankings catastrophic.
  4. Level 8 cap. Past level 8 (Fib numbers 89, 144, 233, …) the curve amplifies measurement error more than signal.

Why soulbound Respect

The reputation token must be non-transferable, by contract-level enforcement, not convention:

  • If you could buy Respect, the system becomes plutocratic (the very thing fractal democracy rejects).
  • If you could sell Respect, members game the system for short-term profit instead of contribution.
  • Soulbound means governance power literally equals your track record of community contribution.

Larimer’s framing: “Fractal governance is more likely to distribute inflation to those producing public goods which grow the value of the currency instead of being siphoned off by insiders and graft.”

Sortition has history

Sortition (random selection of decision-makers) is not new. Ancient Athens used it. Modern citizens’ assemblies (Ireland’s abortion referendum, France’s Climate Convention) use it today. Fractal democracy is sortition plus nested representation plus a soulbound reputation primitive that lets the process compound over time.

Sources

View source on GitHub →