$FLYMNR

THE ORIGINAL AIRBORNE COMPUTE PLATFORM.

Nature-built hardware for
permissionless monetary systems.

300 million years of R&D. Zero nanometers.

View the swarm ↗ Meet $FLYMNR
MINING EXPERIMENTSynthetic activitySynthetic visuals · no neural simulation
NEURAL QUARRY / SECTOR 01FLYMINER F1 / GEN 01
Cooling system: open a window.
ROW 01 / 05LAYER 01 / 03VISUAL CUTS 000
YOUR ASIC CANNOT REPRODUCE.ASICs depreciate. Flies reproduce.

FROM BUZZ TO BLOCK.

THE NEURAL CONCEPT
01

The wiring

+−Σ EXCITE INHIBIT WEIGHTED INPUT
Illustrative circuit · fixed wiring, changing activity

Nature supplied the circuit.

02

The firing

FIRING THRESHOLD SPIKE INPUT → LEAK → THRESHOLD → RESET
Illustrative voltage trace · a spike earns a reset

Charge. Leak. Fire.

03

The readout

010203 04 0 ms10 ms COUNT 241 3
Example readout · each tick is one spike

Count the buzz.

04

The nonce

SPIKESVOLTAGE CONTEXT SHA-256 32-BIT NONCE SAME INPUT. SAME OUTPUT.
Neural values in · a deterministic nonce out

Neural activity, condensed.

05

The proof

0TMAX PASS ABOVE TARGET LOWER TARGET → SMALLER LANDING ZONE
Example hashes · target region enlarged for visibility

Below target? Fly Block.

SYNTHETIC ACTIVITY

THE HARDWARE

ZERO-NANOMETER BIOLOGICAL COMPUTE.

No fabs. No foundries. Just larvae. A mining architecture with actual bugs.

Flyminer MINING EXPERIMENT · synthetic activity

MINING EXPERIMENT · Synthetic activity. No neural simulation.

Our hardware has bugs by design. 128 visual nodes
No fabs. No foundries. Just larvae. FLYWIRE 783
Active visual nodes—
Visual nodes—
CoolingAmbient

PROOF OF BUZZ

CONSENSUS AT 200 WINGBEATS PER SECOND.

EXPERIMENTAL PROOFS
Fly Block / readingProof hashNonceVerification

Awaiting sufficient buzz. Every Fly Block is verified before it lands here.

SHA-256 was never meant to stay on the ground. Verified proofs. Repeats count once.

THE LAB NOTES

THE SCIENCE / MODEL & MATHEMATICS

ACTUAL EQUATIONS. ACTUAL BUGS.

CONNECTOME → SPIKES → PROOF

A measured fly connectome. Simulated neural activity. Verifiable proof of work. The lab notes behind the buzz.

The current mining experiment uses synthetic visual pulses, not this neural model. Its experimental proofs are verified with the same hash and target mathematics.

01 / THE CONNECTOME

Nature supplied the wiring diagram.

+−Σ EXCITE INHIBIT WEIGHTED INPUT
Illustrative circuit · fixed wiring, changing activity

The model uses the FlyWire v783 connectome: mapped neurons and synaptic connections from an adult fruit fly. Each connection carries a weight. Together, they determine how one neuron's spike influences another.

u_i(t) = a × sum_j W_ij s_j(t)

W is the signed connection weight, s is a spike, and a is a fixed scale. Positive and negative weights provide excitatory and inhibitory input.

The wiring is fixed. No motivational seminar for the neurons.

FlyWire · Dorkenwald et al., Nature (2024)
02 / LEAKY INTEGRATE-AND-FIRE

Charge. Leak. Fire. Repeat.

FIRING THRESHOLD SPIKE INPUT → LEAK → THRESHOLD → RESET
Illustrative voltage trace · a spike earns a reset

Each simulated neuron accumulates input while its membrane voltage leaks toward rest. Cross the threshold and it fires, then resets. Delayed, decaying synaptic input links the cells into a network.

tau_m × dV_i/dt = −(V_i − V_rest) + g_i
V_i > V_th → spike, then V_i resets

V is membrane voltage; g is the effective synaptic drive. The equation describes the leak and input between stimulation events. The reference model uses a 20 ms membrane time constant, −52 mV rest/reset and −45 mV threshold.

A simplified electrical model. Still more personality than an ASIC.

Brain model · Shiu et al., Nature (2024)
03 / THE NEURAL READOUT

Count the spikes. Take the voltage.

010203 04 0 ms10 ms COUNT 241 3
Example readout · each tick is one spike

A fixed sample of up to 128 downstream neurons supplies the readout. The reference experiment counts their spikes over 10 ms windows and samples voltage at 0.01 mV precision. Those 128 cells are a sample, not the entire simulated brain.

c_i = spikes from neuron i within a 10 ms window

c is the spike count within the observation window. Counts and sampled voltages form the neural feature vector F.

Performance review: how many times did you fire?

04 / FROM BUZZ TO NONCE

A thought, reduced to 32 bits.

SPIKESVOLTAGE CONTEXT SHA-256 32-BIT NONCE SAME INPUT. SAME OUTPUT.
Neural values in · a deterministic nonce out

The neural feature vector is hashed with a fixed measurement context. A 32-bit slice of that digest becomes the nonce: the number used in a candidate proof.

D = SHA-256(context + neural features); nonce = 32 bits of D

H is SHA-256, C is the measurement context, and π₃₂ takes 32 bits. Identical inputs produce the same nonce; neural activity is not a guarantee of cryptographic randomness.

Compound eyes. Compact output.

05 / PROOF OF WORK

The hash does not care about wings.

0TMAX PASS ABOVE TARGET LOWER TARGET → SMALLER LANDING ZONE
Example hashes · target region enlarged for visibility

A candidate containing the neural nonce is hashed twice with SHA-256. Read as a number, the result must be at or below the target to count as a valid experimental Fly Block.

h = SHA-256(SHA-256(candidate with nonce n)); valid when h ≤ target

B(n) is the candidate header containing nonce n. T is the target. Lower target, harder proof. The checkmark confirms this mathematical check.

Six legs do not qualify for a difficulty discount.

06 / DIFFICULTY & PROBABILITY

More attempts. Same indifferent math.

1 / 256 CHANCE PER ATTEMPT ~256 ON AVERAGE NO APPOINTMENT WITH LUCK.
Example odds · an average, never a countdown

A 256-bit hash has 2²⁵⁶ possible values. Exactly T + 1 of them meet the target, including zero. That gives the success probability and the expected number of attempts.

Success probability p = (T + 1) / 2^256; expected attempts = 1 / p

With uniform hashes and independent, distinct candidates, a probability near 1/256 means about 256 attempts on average. It is not a deadline: repeated candidates do not create fresh chances.

Hashrate counts attempts per second. Wingbeats remain a separate department.

The anatomy is measured; the activity is simulated. A valid hash verifies a proof, not biological realism or a mining-efficiency advantage. These experimental proofs do not secure Bitcoin or Solana, or distribute $FLYMNR.

$FLYMNR / SWARM ECONOMICS

DECENTRALIZATION. WITH COMPOUND EYES.

LAUNCH DESIGN · NOT ACTIVE

Standard pump.fun creator fees. Finite supply. Public receipts.

01 / FIXED SUPPLY

21 million was cute.

1,000,000,000 $FLYMNR

Planned standard pump.fun launch with Mayhem Mode off. Target creator holding: approximately 2% (20 million tokens), acquired through a paid purchase. The wallet, transaction and actual amount acquired will be disclosed.

02 / FIP-21

SUPPLY REDUCTION VIA CONTROLLED EXTERMINATION.

2.1% of collected creator fees → buyback budget

Planned split: 2.1% to purchase and burn $FLYMNR; 97.9% to creator / operations. Trading fees fund the budget. The fly simulation does not generate revenue. Network costs come from the operations share.

Example: 10 SOL collected → 0.21 SOL for buybacks + 9.79 SOL for operations.

CONTROLLED EXTERMINATION. WITH RECEIPTS.

FIP-21 is FlyMiner's proposed policy. Buybacks would start in manually approved batches, spending only the available budget, then burning the tokens purchased. Planned public accounting: fees collected, buyback funds allocated, spent and remaining, tokens burned, and transaction receipts.

No burn mechanism is active yet. The amount burned depends on the purchase price; no buyback frequency or price increase is promised.

FlyMiner is an art project and an experiment in biological mining. $FLYMNR is not an investment and provides no equity, ownership rights or promise of income or returns. The supply and monetary mechanics above are a launch proposal. This visualization does not mine or award tokens.