Every system you interact with runs on metrics. Salary measures hours, not impact. Likes measure attention, not quality. GDP measures transactions, not wellbeing. Grades measure compliance, not understanding.
These aren't bugs. They're features — of systems designed before anyone understood Goodhart's Law: when a measure becomes a target, it ceases to be a good measure.
People aren't broken. The incentive layers they operate inside are. When you reward publication count, you get p-hacked papers. When you reward engagement, you get outrage bait. When you reward quarterly earnings, you get strip-mined futures.
This isn't a moral failing. It's a thermodynamic one. The metrics have no anchor to reality.
Entropy, operationally, is disorder. Wasted energy. Lost signal. Unused potential. Entropy reduction is the measurable act of creating order from disorder: organizing, building, repairing, teaching, feeding, coding, composting.
The Extropy Engine proposes a simple inversion: instead of measuring proxies (money, points, likes), measure the actual thermodynamic change. Did this action reduce entropy? By how much? That measurement becomes the basis of value.
1. An agent (person, org, AI) performs an action.
2. Validators assess whether entropy was reduced.
3. The reduction is quantified and recorded on a DAG (directed acyclic graph).
4. XP (extropy points) are issued proportional to verified reduction.
5. XP cannot be transferred, only earned. No speculation. No inflation.
This means you can't buy status. You can't inherit it. You can't game it with bots. You earn it by making things measurably better.
Entropy metrics are hard to standardize across domains. Validating a repair is different from validating a song. The system requires domain-specific entropy measurement frameworks, and those frameworks themselves need to be validated recursively. This is an open problem. See the full list of open problems.
If this resonates, the next step depends on what you want: