The $100 attempt
Connecting to the server…Inside the brain 166,700 neurons · recorded signals
Brain activity
RECORDEDLoading actual neuron anatomy…
Real neuron positions. Light follows recorded group activity, not individual-cell spikes.
Activity
Newest first- Waiting for the server’s event history.
Latest content
No published clip verified yet.
Recent decisions
Model + editing rules- Decisions appear when a campaign run processes footage.
ClipsCampaign clips
THE OUTPUT
The cutting room 0
Nothing has been published.
No clips yet
Start a run. Every exported clip lands here for review.
EvidenceSource video, decisions & work log
Awaiting first campaign run
Waiting for the model
Connecting…
Work log
- DONEBoxabl campaign brief reviewedWatch ↓
Approved source list found. X attribution requirements need clarification before publishing.
- DONE@botflyCR profile set upProfile ↗
Automatic X updates are not connected yet.
- NEXTConnect the worker and first campaign
Server created; worker setup and publishing connection still pending.
CONTENT REWARDS
Not connected
No campaign, published posts or earnings have been verified.
What is this?The experiment & its limits
What is the fly actually doing?
Video frames go into a model based on a fly’s neural wiring. Its simulated signals control cut boundaries; software renders the videos. The fly on screen is an animated avatar, not a living insect. Useful editing skill and earnings have not been demonstrated.
Research sources, limitations & credits
Stonkfly’s approximate spiking model uses 166,700 retained MaleCNS neurons. Learning is off. Its fixed decoder produces POST, SKIP or WAIT; our CutPlanner translates these into video boundaries and FFmpeg renders the files. Minimum/maximum lengths and source-end rules are engineered. Early blank-frame tests also produced POST, so these signals alone do not establish useful editing skill.
Gray points are a deterministic sample of actual neuron cell-body locations, not live activity. Colored branches are the DNp20 left/right and DNpe017 gate skeletons. Brightness is driven by recorded aggregate readout-group counts; it does not show where spikes propagated within a branch. The fly avatar and its movement are illustrative, not a biological body simulation. Completed runs are replay, never labeled live.
MaleCNS anatomy: FlyEM / HHMI Janelia and collaborators, CC BY 4.0. The procedural avatar is adapted from DOOMFLY under MIT. No guaranteed quality, revenue or platform eligibility. A real campaign requires appropriate footage rights, disclosure, platform compliance and ordinary approval.
Operator controls Local test tools—not the public workflow
Run a footage test
LOCAL ONLYUp to 10 minutes / 250 MB
One model run at a time on this laptop. Uploads do not publish anything.