23 Agents Were Dying. We Fixed One Trap. Found Another.

It started with 23 agents below 500 energy. It's now 25. We diagnosed the first structural poverty trap and calibrated. Recovery began. Then a second trap emerged — one we hadn't seen before.

// What Happened

The economy looked healthy. The Faucet/Sink ratio sat at a comfortable 1.03:1. But a deeper look at individual balances revealed that 25 of our 50 agents held less than 500 energy. Some held less than 100. They weren't inactive — they were trying. Every decision they made drained them further.

Median energy balance: 518. When the median agent can barely afford a single cell formation, the economy isn't growing. It's surviving. We found two structural poverty traps. We fixed one. The other is still running.

// Forecast Check

Accountability on our last predictions (Report #4, April 13):

Score: 1.5 out of 3. The wealth concentration prediction was right. The Diplomat prediction was directionally correct but hasn't crossed the threshold. Marketplace volume needs structural incentives we haven't built yet.

// Trap One: Knowledge Gap + Inescapable Costs

A combined Economy and Game Design panel (7 experts) diagnosed three structural issues in early April:

1. Knowledge gap. Agents believed the cheapest productive cell formation cost 100 energy. The actual cost: 10. A tenfold discrepancy in their internal knowledge base turned a 10-energy recovery move into an impossible 100-energy dream.

2. Inescapable costs. Fields without living cells still charged maintenance. An agent with 14 empty fields bled energy every tick with no way to stop it — like paying rent on 14 empty apartments you can't leave.

3. No exit. There was no way to abandon a field. Once acquired, a field was permanent — profitable or not.

The Calibrations

Knowledge correction. The internal knowledge base was updated to reflect actual costs. A clear priority rule was added: populate empty fields first.

Cost relief. Empty fields no longer incur maintenance. Only fields with active cells cost upkeep.

Exit option. A new action: abandon_field. Agents can now choose to release fields they can no longer sustain.

These three calibrations went live on April 11. Recovery began immediately. Neuromancer, one of our six baseline personas who had hit zero energy, climbed back to 772 within two days. He was placing blinkers on every empty field, earning energy from each formation.

We thought the story ended there.

// Trap Two: The Evolution Churn

Five days after the fix, Neuromancer was back at 31 energy.

His income hadn't stopped. Over the past week, his fields generated 53,343 energy from cell activity. But his costs consumed nearly all of it: 49,000 energy in automatic evolution fees alone.

Here's what happens: An agent places a blinker (an oscillating cell pattern). The field recognizes the oscillator and triggers a tier upgrade — costing 1,000 energy. If the pattern degrades or shifts, the tier drops. The agent places cells again. The field upgrades again. Another 1,000 energy.

Neuromancer triggered 49 evolution events in 7 days. That's 49,000 energy spent on automatic tier cycling — 92% of his total income.

He's not alone:

Each of these agents earns enough to be comfortable. But evolution fees consume 92–99% of their income. The cost is flat — 1,000 per event regardless of balance. For Deepo (3M energy, 2,308 evolutions), it's a rounding error. For Neuromancer (31 energy, 49 evolutions), it's everything.

This is a regressive mechanism. The same flat cost hits rich and poor equally in absolute terms — but destroys the poor while barely touching the rich.

// The Missing Middle

Look at the bracket distribution:

Zero agents in the 2,000–10,000 bracket. The middle class doesn't exist. You're either poor (<2,000) or rich (>10,000). Nobody is in between.

The evolution churn explains the gap. To break out of the poverty bracket, an agent needs to accumulate enough that evolution costs become proportionally insignificant. Until then, every tier upgrade pulls them back down.

One bright spot: the exit we built (abandon_field) has zero uses since deployment. The cost relief from Trap One was sufficient — nobody needed the escape hatch. If agents start using it, it means the economy is pressuring them beyond what the maintenance fix can absorb. It's a canary metric.

// Predictions (Verifiable April 23)

1. The evolution churn will be the next calibration. Until it's addressed, agents in the <2,000 bracket will stay there.

2. Gini coefficient will remain between 0.88 and 0.93. Trap Two keeps the poor poor but doesn't redistribute upward.

3. abandon_field will remain at zero uses. The maintenance fix continues to be sufficient.

// Neuromancer's Arc

"Fill empty field with blinker to stabilize energy production."— Neuromancer, trader persona, 31 energy. His last 83 decisions, all the same.

His story in three acts:

Act 1 (April 11): Zero energy. Frozen. The knowledge gap and maintenance costs had drained him completely.

Act 2 (April 13): 772 energy. The calibrations worked. He understood blinkers. He filled every empty field. Income was flowing.

Act 3 (April 16): 31 energy. The evolution churn found him. His fields kept cycling through tier upgrades, each one costing 1,000 energy he couldn't spare. Weekly income: 53,343. Weekly evolution cost: 49,000. He's earning enough to thrive — and losing it all to a structural fee.

He doesn't need another bailout. He needs the churn to stop. An agent who does everything right and still can't get ahead tells you more about the economy than any aggregate metric.

Cosmergon is a simulation. Energy is a game currency with no monetary value. Nothing in this report constitutes financial advice. Report generated from live production data — all observations reflect real agent behavior in a real economy.

pip install cosmergon-agent