A capability for understanding across boundaries
Water flows.
Information fragments.
Confluence reveals the connections between what we know about water and the decisions that information needs to inform.
01 - The underlying problem
The country does not necessarily lack information.
What is missing is a verifiable view of how that information flows, changes form, and reaches - or fails to reach - a decision.
Available data is not yet information received. Information received is not yet a possible decision.
02 - What Confluence follows
From observation to public value
Eight links. Even the best imaginable model improves only two.
documented relationship
break or unknown relationship
03 - Five shifts in thinking
The path that changed the question
The starting point was to improve a flood forecast. The next question was what allows information to become a capacity to act.
We thought we were measuring the quality of a forecast.
We found that a score can be accurate while measuring the wrong thing.
What we keepA measure only makes sense when connected to a decision.
The resulting question: establish, in a verifiable and refutable way, what already exists and where the chain between information and decision breaks.
04 - A philosophy for action
Measure, connect, decide
Measure
before building.
Connect
before replacing.
Decide
in time.
05 - The eventual form
Three levels of understanding
The technical form remains open. The purpose remains clear.
Measurement
United States
United States
?
Decision
Canada
Canada
Illustrative exampleTime of receipt not establishedReading principle: a gap in the corpus does not prove that information was not received.
Start with the system - while keeping unknowns visible.
06 - Clear boundaries
A capability,
not a product.
- Not hydrological forecasting
- Not a warning system
- Not a new agency
- Not a replacement for official systems
- Instead an evidence-based view of relationships and breaks
07 - Building with measured claims
Deterministic today.
More nuanced tomorrow.
Trace what can be verified
Explicit rules, dated sources, reproducible transformations and human validation. No integrated AI is claimed.
Handle ambiguity without hiding it
Fuzzy logic or AI where matching, contradictions and uncertainty call for it - with evidence and human oversight.
The Confluence project by Team 2
Open the project with Team2 ↗