How the card works.

Flyweight is a nightly fight show. Two simulated fruit-fly brains fight over a food cup for ninety seconds. Each fighter carries a live token on Robinhood Chain; the hour before the bell, that token’s community feeds it. The seed and the input snapshot are committed before the weigh-in and revealed after, so every bout can be re-run by anyone.

• ›Fighters are existing tokens, not ours — their communities are the audience.
• ›Distinct buying wallets feed a fighter, never volume, and the effect is capped at ±15%.
• ›Every bout is published, including the dull ones, with its seed, inputs and replay.
• ›No betting pool. No token at launch. The token comes last, if the show earns it.

Bell time

21:00 UTC, nightly. The first card's date is announced on X 24 hours before its bell; the site shows a live countdown to the next bell and the weigh-in window.

T-24h

Card announced. Both communities tagged, records posted.

T-60m

Seed generated, hash published with the snapshot rules. From here the house cannot change anything.

T-60m → T-0

Weigh-in window. Distinct wallets buying a fighter’s token feed it; net selling makes it timid; the swing is capped.

T-0

The bell. Ninety seconds. Two brains, one shared arena, one food cup.

Finish

Knockout: the loser’s escape circuit fires and it flees. Decision: whoever holds the cup at time.

T+2m

Result and proof page: seed revealed, inputs revealed, replay downloadable.

Fights are computed ahead of broadcast on a deterministic engine, which is why the replay is available the moment the bell stops. The pre-compute is disclosed on every proof page.

Condition sets how a fighter feels when the bell rings. It changes the brain’s starting state, not the result. The model is fixed before the card is announced and printed on the proof page; it is never tuned per fight.

feeders   = min(distinct buying wallets, 200)
feed      = +0.15 × feeders / 200
bitter    = −0.10 × net selling (0..1)
shadow    = −0.05 if a single very large sell landed
condition = clamp(1 + feed + bitter + shadow, 0.85, 1.15)

Market → sense mapping

New distinct wallet buys

sugar on the legs (taste) → aggression up

Net sell pressure

bitter taste (taste) → aggression down

A single very large sell

a looming shadow (visual) → flinch, then reset

Many trades in a short window

vibration (mechanosensory) → restlessness

Silence

dust on the eyes (visual) → sluggishness

The anti-whale rule

• ›400 wallets buying a little: a very well fed fighter.
• ›1 wallet buying a lot: one sugar drop, nothing more.
• ›Wallets under the dust floor are ignored.
• ›Wallets funded from the same source inside the window are counted once (sybil clustering, applied before the bell).
• ›Round trips inside the window are not counted as feeding.
• ›Total condition swing is capped at ±15% of baseline.

The engine is deterministic: the same seed plus the same inputs produces the same fight, frame for frame. So we commit before the window opens and reveal after the fight ends.

Before (T-60m)

seed generated · hash(seed) published · snapshot block stated · mapping already public · engine version pinned

After (T+2m)

seed revealed · hash verified by anyone · snapshot data published · replay file downloadable · re-run reproduces the result

Re-run it yourself

• ›Install the pinned engine version named on the proof page.
• ›Load the replay file and the input snapshot.
• ›Compare the output hash with the published one.
• ›A mismatch means we cheated. Tell everyone if it happens.

What is server-authoritative, and why

The seed

a client-side seed is a client-side win

The snapshot block

fixed by us, in advance

Distinct buyers

counted from chain logs, never reported by the browser

The simulation

nobody can touch it mid-fight

The record

append-only; results are never edited

Measured — from the public connectome release

• ›Which neuron connects to which neuron.
• ›How many synapses sit on each connection.
• ›The anatomical position of every cell body.
• ›Published under an open licence and credited on the proof page.

Modelled — our engineering, stated plainly

• ›Neuron dynamics (leaky integrate-and-fire, uniform parameters; no per-neuron tuning, because tuning would be a back door).
• ›Which sensory cells a market event stimulates.
• ›Which descending cells map to which arena action.
• ›The aggression readout that turns firing into a lunge.

Not claimed — ever

• ›That the fly is conscious, alive, or aware.
• ›That aggression from a connectome simulation is scientifically validated. It is not.
• ›That the outcome predicts anything about a token’s price.

The brain panels on this site use soma positions from the public FlyWire FAFB v783 whole-brain connectome (Dorkenwald et al., Nature 2024): 28,017 of 139,255 neurons, sampled by class. The activity they show is a model, not a recording.

A fighter is a token. To be eligible for the card it must clear an automated check before it can be booked.

• ›Launched through a recognised launchpad on Robinhood Chain, proven by the launchpad’s own on-chain event.
• ›Liquidity above the floor, pool not abandoned.
• ›Sell path proven: a token nobody can exit is not allowed to headline a card.
• ›Name and image pass the content rules.

Every fighter gets a page: record, debut, brain seed (fixed for life), colours, and a full fight history with replay and proof links. Feeder counts per bout are shown, so a rented community for a single night is visible as a spike.

Maggot division

debut fighters, first three bouts

Contender

a winning record, eligible for the card

Main event

top of the card, longer weigh-in window

Champion

holds the belt until beaten; refuse a defence for two weeks and it is stripped, publicly

• ›A season runs eight weeks.
• ›The champion’s fighter page is sealed permanently and its brain seed is retired.
• ›A fighter knocked out three times in a row is retired too. The token can re-enter as a new fighter with a new seed and a blank record.
• ›Nothing is deleted. The old page stays readable.

Whale buys the outcome

distinct wallets only, capped ±15%

Sybil wallet farm

same-funder clustering + dust floor

Wash trading in the weigh-in

round trips inside the window are not counted

Fake token, no exit

sell path proven before eligibility

House fixes the result

seed committed before the window opens

Selective highlight reel

every bout published, losses too

Rented community for one night

feeder counts shown per bout

Phase 1 makes no revenue, on purpose. Charging for access to an empty arena is how projects die before launch. Revenue arrives in Phase 2, when a slot on the card is worth paying for.

Challenge slot, undercard

0.03 ETH

Challenge slot, main event

0.08 ETH

Rematch (same pairing)

0.05 ETH

Corner sponsor, one night

0.10 ETH

Venue sponsor, one week

0.40 ETH

Paying in $FLYW gets a discount, and the discounted portion is burned rather than collected. In Phase 3 the launchpad creator fee on $FLYW trading (100 bps) accrues to a published treasury address.

Allocation of every unit collected

• 40%Engineering and infrastructure
• 25%Buyback and burn, executed publicly
• 20%Prize pool for season finals and belt defences
• 10%Content: clips, art, the weekly recap
• 5%Reserve

Supply

166,700 — one per neuron, never inflated

Team allocation

0%

Presale

none

Launch

fair launch on a Robinhood Chain launchpad, quoted in ETH

Liquidity

locked by the launchpad at graduation

Creator fee

100 bps, to the published treasury address

Timing

after the show has an audience — Phase 3 is gated on two weeks of live viewers and one unsolicited paid slot

The one rule

$FLYW never buys a better chance of winning: not through condition, not through matchmaking, not through a hidden multiplier. Everything it does is access, identity and cost.

Holder tiers

Hold 1

a bound neuron id, visible when it fires · card vote

Hold 100

named cluster on the live brain panel · fee discount tier 1

Hold 1,000

handle on the corner card for the night · fee discount tier 2 · first look at the next card

Hold 5,000

propose a main event to matchmaking · permanent name on the venue wall

Supply leaves through

• ›Discounted fee portion burns.
• ›Buyback and burn from revenue.
• ›Belt defence entry burns.
• ›Season pass mints burn a fee.
• ›No emissions, ever.

Phase 1 · First bell (week 1)

sim core, market watcher, arena and brain panels, commit-reveal, exhibition bouts, first public card

Phase 2 · A card worth a slot (weeks 2–5)

fighter pages and the wall, challenge booking, results and proofs posted to X, auto-cut clips, first belt

Phase 3 · The token (weeks 6–10, gated)

$FLYW fair launch, neuron binding, fee discounts and burn ledger, card voting

Phase 4 · The venue (months 3–6)

seasons and belts per division, open engine release, any token can enter, tournament nights, embeddable widget

What decides Phase 3

• ›Live viewers at the bell, two-week rolling average.
• ›Returning viewers night over night.
• ›Distinct feeders per bout.
• ›Unpaid mentions from fighter communities.
• ›Independent verifications of a result.
• ›Challenge requests we did not solicit.

Deliberately ignored: token price, follower count, total volume (it rewards whales), highlight views without return visits.

Is the fly real?

The wiring is a measured reconstruction of a real fly nervous system. The dynamics, the senses and the controller are models. Nothing here is alive, and we never say otherwise.

Can I bet on a fight?

No. There is no betting pool and none is planned at launch. Wagering changes the legal shape of the product entirely.

Can buying enough of a token guarantee a win?

No. Condition counts distinct wallets, not size, and is capped at ±15%. A well fed fighter is more aggressive, not undefeated. The two brains and the committed seed decide the rest.

How do I know a result was not rigged?

The seed hash is published before the weigh-in opens. After the fight, the seed, the inputs and the replay are published, and the engine version is pinned. Re-run it. If the hash does not match, we cheated.

Can my token be a fighter?

In Phase 2, any eligible token can book a slot through the challenge board. Eligibility is automated: recognised launchpad, liquidity floor, proven sell path, content rules.

When does $FLYW launch?

Last. Only after the show clears its viewer and demand gates for two weeks running. A token before an audience is the failure mode we keep seeing elsewhere.