v3 - Stock-Flow Consistency — EDEN simulation report

How to read this page. The plain-language write-up comes first; the technical results follow. Figures and the raw results.json are linked so you can check every number yourself. This is a model of the design under stated assumptions — not a forecast.

In plain language

Companion to RESULTS - Stock-Flow Consistency. Companion added July 11, 2026 — the v1–v5-era runs predate the plain-language convention; written from the committed RESULTS as it stands today (including any verification-pass corrections already applied in that file), with no reinterpretation.

The question

The agent-based runs priced money with a quantity-theory rule but never proved the money accounting is consistent. Monetary economists check any currency design the same way (the Godley & Lavoie stock-flow-consistent method): every flow must be simultaneously one party's outflow and another's inflow, every sector's net flow must equal the change in what it holds, and the total money stock may change only through identified creation and destruction. Many token designs quietly fail this — money appears from rounding, fees vanish, "rewards" aren't sourced from anywhere. Does EVE's?

What we found

EVE's accounting is stock-flow consistent to machine precision: the maximum relative residual is 1.55 × 10⁻¹⁶ across all 180 periods and all identities. Four identities were verified every period. Every transaction balances — each row of the transactions-flow matrix sums to zero. Every sector balances — and the Validators (verification pool) and Protocol sectors are pass-throughs, ending every period at exactly 0 having routed everything they received. The money identity holds exactly: the change in total EVE equals minted plus floor top-up, minus protocol burn, minus death burn — verified to 10⁻¹⁶. And transfers are stock-neutral: machine-pay and bridge off-ramps move EVE between households and institutions but never change the total money stock; only minting and burning do.

Three things this confirms about the design. EVE is sound base money, not debt — unlike fiat, where money is a bank liability, EVE has no matching liability, and the books still close. The data economy can grow without inflating the currency — because machine-pay is a transfer, institutions can pay people more and more for data with zero effect on the money stock, the exact property the white paper claims, now proven at the accounting level. And the routing leaks nothing: 60 + 15 + 10 + 10 + 5 = 100, with every slice accounted to a sector or a burn each period.

The honest catch

This proves accounting consistency, not behavioral or price realism — that's the agent-based sims' job, and fraud is the adversarial model's job. It covers the EVE ledger only (fiat on the other side of the bridges is out of scope), assumes the routing and flows as specified, and does not model defaults or theft. One variant worth noting: the creator-stability reserve (banking the protocol slice instead of burning it, then releasing it to creators) is also stock-flow consistent — banked EVE simply stays in the money stock and later moves as a transfer.

One line

Every unit of EVE that exists traces to a mint and every unit that disappears traces to a burn — verified to machine precision, a maximum residual of 1.55 × 10⁻¹⁶ across all 180 periods — so the books close with no monetary black holes; what this proves is the accounting, not the behavior, the prices, or the fraud-resistance.

Words used here (added July 18, 2026 — plain-language house rule; the text above is unchanged). Money stock — all EVE in existence at a given moment. Minting / burn — creating new money / permanently destroying it; the claim proven is that the total changes only through these. Accounting identity — an equation that must hold by definition, like "every purchase is also a sale"; four of them were checked every period. Transactions-flow matrix — the master spreadsheet where every flow appears twice — once as someone's outflow, once as someone's inflow — and every row must sum to zero. Residual 1.55 × 10⁻¹⁶ — the leftover imbalance: rounding error at the very edge of computer arithmetic, i.e., zero. Sector — one group's combined books (households, institutions, validators, protocol). Validators (verification pool) — the deposit-backed participants who check and record activity; as a sector they are a pure pass-through, ending every period at exactly zero. Quantity-theory rule — the simple pricing shortcut (more money chasing the same goods means higher prices) used by the behavioral sims, not re-litigated here. Base money vs liability — base money is the thing itself, owed by no one; a liability is an IOU — bank-created dollars are a bank's IOU to you, while EVE has no debt behind it. Bridge / off-ramp — the crossing where EVE is exchanged for outside money; modeled only as a transfer, since the far side is out of scope. Top-up — extra minting to cover the floor when its funding slice falls short — one of the identified creation channels in the money identity.

Technical results

Closes a gap the agent-based runs left open: they priced money with a quantity-theory rule but never proved the money accounting is consistent. This is the check monetary economists expect of any currency design, and it's cheap and decisive. Files in this folder.

What this proves, in one line

EVE's accounting is stock-flow consistent to machine precision (maximum relative residual 1.55 × 10⁻¹⁶ across all 180 periods and all identities): every unit of EVE that exists can be traced to minting, every unit that disappears to a burn, and nothing is created or destroyed by any other path. There are no monetary "black holes."

Why this matters

Following Godley & Lavoie's stock-flow-consistent (SFC) method, a monetary system is consistent only if every flow is simultaneously one party's outflow and another's inflow, every sector's net flow equals the change in what it holds, and the total money stock changes only through identified creation and destruction. Many token designs quietly violate this — money appears from rounding, fees vanish, or "rewards" aren't sourced from anywhere. EDEN's does not, and now that's demonstrated rather than asserted.

The four identities, all verified every period

Every transaction balances. Each row of the transactions-flow matrix sums to zero — the EVE one sector receives is exactly what another sector (or the money-creation account) provides.
Every sector balances. Each sector's net inflow equals the change in its EVE holdings. The Validators (verification pool) and Protocol sectors are pass-throughs — they end every period at exactly 0, having routed everything they received.
The money identity holds exactly: Δ(total EVE) = minted + floor top-up − protocol burn − death burn. Verified to 10⁻¹⁶.
Transfers are stock-neutral. Machine-pay (institutions buying data) and bridge off-ramps move EVE between households and institutions but never change the total money stock — only minting and burning do.

The transactions-flow matrix (representative month)

Every column is a sector that holds EVE; every row is a flow; "± Money" is creation (+) or destruction (−). Each row sums to zero across the sectors and the money column.

Transaction	Households	Validators	Protocol	Institutions	± Money
Mint → households (creator/maint/dep)	+985,383	0	0	0	+985,383
Mint → verification pool	0	+115,927	0	0	+115,927
Mint → protocol	0	0	+57,964	0	+57,964
Floor paid from pool	+25,504	−25,504	0	0	0
Floor top-up (minted)	0	0	0	0	0
Validator remainder → households	+90,423	−90,423	0	0	0
Machine-pay (data)	+115,927	0	0	−115,927	0
Bridge off-ramp (sell EVE)	−115,927	0	0	+115,927	0
Protocol burn	0	0	−57,964	0	−57,964
Death burn	−59,352	0	0	0	−59,352

Reading it: minting is the only source of new EVE and it lands entirely in the 60/15/10/10/5 routing; the verification pool (115,927) is fully paid out as floor (25,504) plus remainder (90,423), so Validators net zero; the protocol slice is minted and then burned, so Protocol nets zero; machine-pay and the off-ramp are equal and opposite, so the data economy moves EVE around without inflating it; and money leaves existence only through the protocol and death burns.

What it confirms about the design

EVE is sound base money, not debt. Unlike fiat (where money is a bank liability), EVE has no matching liability — it is created by minting and extinguished by burning, and the books still close. The SFC check confirms this is internally coherent.
The data economy can grow without inflating the currency. Because machine-pay is a transfer, institutions can pay people more and more for data with zero effect on the money stock — exactly the property the white paper claims, now proven at the accounting level.
The routing leaks nothing. 60 + 15 + 10 + 10 + 5 = 100, and every slice is accounted to a sector or a burn each period.

Honest limits

This proves accounting consistency, not behavioral or price realism — that's the agent-based sims' job, and fraud is the adversarial model's job. It covers the EVE ledger only (fiat on the other side of the bridges is out of scope). It assumes the routing and flows as specified; it does not model defaults or theft. One variant worth noting: the creator-stability reserve (banking the protocol slice instead of burning it, then releasing it to creators) is also stock-flow consistent — banked EVE simply remains in the money stock and is later moved as a transfer, so identities 1–4 still hold. Audience: this is the check monetary economists will look for first.

Files

sfc_sim.py — the SFC model and the four identity checks
results_sfc.json — residuals and final balances

Raw data

⬇ results_sfc.json