Source
Approved data feed, document, event, price, weather, policy, research, or odds source.
Glass Lab targethttps://glass-lab.up.railway.app?artifact=hypothesis%3Asource
Plain-language user guide
How to use Hypothesis Lab
See how KELLY creates questions from data, what it wants to test, and what would make it change its mind.
Workflow
Use the lab in five moves
- Choose a proof preset that matches the question you care about.
- Ask KELLY inside this lab to get the sector-specific response without leaving the page.
- Inspect the live forecast showcase and proof-route status before trusting the output.
- Open the Glass Lab artifact target to trace source, proof, replay, and settlement evidence.
- Open the full Ask KELLY app only when you want the broader cross-sector conversation.
Artifact spine
The required proof path
Observation
A tagged fact candidate KELLY can inspect without trusting blindly.
Glass Lab targethttps://glass-lab.up.railway.app?artifact=hypothesis%3Aobservation
Hypothesis
The causal question, bridge, or forecast KELLY wants reality to test.
Glass Lab targethttps://glass-lab.up.railway.app?artifact=hypothesis%3Ahypothesis
Proof receipt
The evidence packet that explains what KELLY saw and why it moved.
Glass Lab targethttps://glass-lab.up.railway.app?artifact=hypothesis%3Aproof
Replay
The reconstruction path that proves the chain can be rebuilt later.
Glass Lab targethttps://glass-lab.up.railway.app?artifact=hypothesis%3Areplay
Settlement
The reality-close record: what happened, when, and from which authority.
Glass Lab targethttps://glass-lab.up.railway.app?artifact=hypothesis%3Asettlement
Calibration
The score that separates accuracy from confident language.
Glass Lab targethttps://glass-lab.up.railway.app?artifact=hypothesis%3Acalibration
Memory eligibility
A replay-backed candidate that may influence future KELLY work.
Glass Lab targethttps://glass-lab.up.railway.app?artifact=hypothesis%3Amemory
Proof presets
High-value forecasts this lab is built to inspect
24-hour hypothesis closure
Which fresh hypothesis can be checked against reality fastest?
Why this mattersShort-term proof that KELLY can generate and close its own questions.
Reality signalnew observation, hypothesis, reality check, closure receipt
Open proof targetCausal bridge discovery
Which two systems appear linked, and what hidden loop could connect them?
Why this mattersMedium-term proof that KELLY can discover useful causal structure.
Reality signaltwo loops, bridge hypothesis, falsifier, settlement plan
Open proof targetRecursive theory refinement
Which theory has mutated the most under evidence while remaining replayable?
Why this mattersLong-term proof that KELLY can self-correct without losing lineage.
Reality signaltheory lineage, confirmations, weakenings, memory cell trail
Open proof targetSource families
What KELLY watches
- KELLY observation ledger
- approved public data feeds
- contradiction queue
- forecast deltas
- settlement and replay receipts
Proof standard
What makes it trustworthy
- Every hypothesis needs a falsifier.
- Old data can become new evidence under a new causal question.
- User questions can guide research intent without writing ground truth.
- Reality, not confidence, closes the loop.
Boundary
What this is not
This is an inspection interface. It does not let public users create hypotheses directly inside KELLY Prime ground truth.
Every displayed forecast must link to Glass Lab or clearly say the artifact is pending.
User controls
Export, delete, location, feedback
Use Causalea Settings for browser-local export/delete, private coarse location, appearance, and feedback. This lab does not store precise coordinates and does not let public users write KELLY Prime.