▲ FUNCTIONAL BUILD v0.2 — all sections collapsed on load · live scoring · JSON export, clipboard copy & print all wired · for road testing
UReDI v3 · Dataset Rescue Scoring

UReDI Calculator

Urgency · Recoverability Difficulty · Importance  |  a structured priority instrument for non-expert operators

Tom Karl's framework evolved from his December 2025 scoring document (1.0), which summed Urgency, Recoverability Difficulty, and a ten-part Importance score additively, to UReDI (2.0), which kept the Importance backbone but introduced a multiplicative Priority roll-up and added structural uniqueness (Rs).

UReDI attempts to quantify how much a dataset is worth and how hard it would be to recover.

Paired with R*, which answers how likely we are to lose a dataset, and the Keeling Curve Foundation's Long-Term Observations inventory, which answers which measurement series must keep running, UReDI completes an ensemble. Value, risk, and continuity are three different questions, and run together they catch what any one framework alone would miss, making the combined system far more useful than the sum of its parts.

A

Dataset & Curator

required ▼

Identifies the dataset under assessment. The persistent identifier is the anchor that lets the same dataset be recognised across operators and runs.

Light type-detection (DOI / ARK / handle) will normalise this entry so one dataset reads identically for every operator.
B

Operator

recommended ▼
These fields are optional, but strongly encouraged. They are what make a score comparable across operators and across repeat runs. The experience rating in particular is required to test whether different operators land in the same place on the same dataset.
7 / 10
Logged with every run. Without it, cross-operator variance can't be measured later.
★

Worked Examples — Calibration Reference

6 datasets ▼

Six datasets scored in UReDI v3 Appendix B, with the expert score for every component. Use these to calibrate your own judgement, especially on the subjective components.

U

Urgency of Loss

1 component ▼

How likely is near-term loss? Urgency is a single direct score and enters the Priority roll-up unchanged.

RD

Recoverability Difficulty

4 components ▼

If the dataset were lost, how hard would it be to recreate? Structural uniqueness (Rs) carries two-thirds of the weight; the three cost components share the rest.

// reconstruction cost
RDc = ( ETC + SC + PC ) / 3
// final recoverability difficulty
RD = (2/3) · Rs  +  (1/3) · RDc
// rounded to ordinal for the triplet
RD* = round( RD )
I

Importance

10 components ▼

How critical is the dataset to science, the public sector, and markets? Importance sums ten components (raw 10 to 50), then normalises to a 1 to 5 score.

// sum of ten components
I = SAS + PCS + PRS + TL + TR + SE + SR + I_int + D + Q
// normalised importance
I* = round( I / 10 )