06Known limitations · 2026-10-12

Phase 3 review findings

What the 2026-10-12 internal review found when the frozen RC4 code was tested with every GENE bought from the pool, and what is in development for RC5. Published beside the frozen Known limitations, not inside them.

#The finding

GENE's whole supply is minted once, to the hook, and placed in the pool at launch (GeneToken.sol:15-18; GeneHook.sol:233, 420-435). Every trade pays the 1% LP fee on its input token, and the fee is rounded up. GENE paid as fees is placed as asks above the price (GeneHook.sol:541-552), so it leaves the launch curve.

After the first trade of any kind, holders therefore own less GENE than the pool gave out, and the price can never return to the launch price. The all-time low stays at the launch tick, 178,080 (Genome.sol:18; GeneHook.sol:266, 774). Every protocol ETH placement is anchored below it, in the band from tick 178,200 to 179,400 (GeneHook.sol:484, 498-501, 516-518, 782-786). No seller can reach that band.

#What each gene does in a genuine market

As RC4 implements them:

• GENESIS: acts in RC4. Places all 1,000,000,000 GENE in one position at launch. This is the curve every trade uses.
• METABOLISM: acts in RC4; its ETH half is a permanent lock. It collects fees and places GENE asks and ETH bands within budgets and caps (GeneHook.sol:479-552). GENE fees become asks above the price, which buyers trade into. ETH fees land in the out-of-reach band or stay in reserve: a permanent ETH lock, not a bid and not support.
• REGENERATION: expresses in RC4, cannot act. Fix in development (RC5). It takes 20% of ETH fees into its reserve, and regenerate() buys only up to 100 ticks (about 1%) above the all-time low (GeneHook.sol:561-591, 707-718). The price cannot return to that range, so it never buys and its reserve stays idle.
• IMMUNITY: expresses in RC4, cannot act. Fix in development (RC5). After confirmed stress it places up to 1 ETH per call and 1.5 ETH per epoch, funded by net selling (GeneHook.sol:493-507). That ETH lands in the same out-of-reach band, so it has no market effect.

#Measured

The review ran its tests on a scratch copy byte-identical to aca5fcd. Traders were given only ETH; every GENE in circulation was bought from the pool.

• Full exit. After 26 epochs of trading, every holder sold everything. The price stopped at tick 177,578, 622 ticks short of the band. All 21.79 ETH the protocol had placed sat in that one band.
• REGENERATION. It expressed at epoch 20, collected 1.094 ETH and spent nothing across 8 calls.
• Unplaced fees. In a launch-heavy run, 5.5 of 10 ETH of fees (55%) was never placed, because a budget does not carry over to a later epoch.
• Thirteen routes to the band. Buying every ask then selling everything, holder burns, donations, side pools, plain transfers, multi-hop routes and flash accounting were tried. None moved the all-time low, and none brought a settled price into the band after the first trade. Flash accounting can enter the band inside one transaction, but no anchor, budget or public call observes it, and the attempt lost 9.38 ETH.

#What this changes in the published figures

• Security conclusions stand. No function removes liquidity, no path extracts value, and the attack matrix has 0 TRANSFER and 0 SELFFUNDED cells. Tests that add GENE with a test cheatcode explore more price paths than genuine trading allows, so their safety results still hold.
• Economic figures are stress tests. Known limitations §0 to §3 (the all-time-low ratchet costs, falling-market depth, IMMUNITY in sell-offs, REGENERATION on the curve) describe prices below launch. Reaching them needs GENE that genuine trading cannot produce.
• Depth readings. GeneLens.report() counts every hook position within each move, including ETH in the out-of-reach band (GeneLens.sol:209-259). This site labels that reading on Analytics.
• The property that does hold. The full-exit price, the price if every holder sold everything, rises with cumulative sell volume. A closed-form calculation matches the measured value within one tick.

#In development: RC5

RC5 is a proposed change, in development. It is not frozen, not audited and not deployed, and its economic limits are not yet approved.

• The candidate design makes REGENERATION and IMMUNITY act through a capped buyback-and-burn priced from the block-start price, instead of an anchor that never moves.
• METABOLISM's ETH placements stay a disclosed permanent lock.
• Any RC5 is a new candidate: a new genomeHash, a new hook address and a new audit package.

Until a frozen RC5 candidate exists, this site describes the genes as RC4 implements them. RC4 stays frozen as the security baseline.