Scroll to zoom · drag to pan · click a particle to track it · double-click to fit
Every cycle, the blob's appetite is turned into actions. Each one leaves a memo on Solana with the hashes needed to replay it.
| Cycle | Ended | Split | S1 Buyback | S2 Holders | S3 Rest | S4 Shed | Actions | |
|---|---|---|---|---|---|---|---|---|
| The first decision will appear at the end of the current cycle. | ||||||||
No AI. No trading algorithm. No dev decision. Full documentation →
ASense
Every 5 minutes, market data becomes physical conditions in the dish.
- Fresh fees above average → richer Buyback food
- New holders, mcap ↑ → richer Holders food
- Volume spike → richer Shed food
- Average market → all sources equal
- Price drop → light on Buyback
- Trades → allowed population
BLive
Particles follow Jeff Jones' Physarum model (Artificial Life, 2010): sense the trail with 3 sensors, turn, move, deposit. They eat, divide, starve and die. Tube networks emerge between food sources.
CDecide
Food eaten per source over the hour sets the split of the treasury (pump.fun creator fees):
- S1 Buyback: buy & burn
- S2 Holders: SOL pro rata
- S3 Rest: keep SOL
- S4 Shed: burn held tokens
DLimit
- Budget: 80% of wallet SOL per decision
- Split by the blob's %
- Shed ≤ 0.2% / cycle, 5% / day
- Reserve kept for fees
- No re-execution after a crash
The simulation is deterministic, bit for bit. Pick a cycle: your browser replays it from the published starting state and checks every hash.