This page describes one possible approach from the project team. It should not be used to make health, legal, environmental, appraisal, or purchase decisions. Suggestions and criticism are welcome through the project’s GitHub issue tracker ↗.
Two calculators, eventually
Immediate impact
A single-facility view: what potential effects might be most relevant at a chosen distance from one data center?
Likely inputsFacility scale, electricity demand, cooling and water evidence, backup generation, noise, vibration, land context, and distance.
Coming soon · no live score yetAddress impact
A cumulative view: what might happen when multiple facilities and regional infrastructure affect the same address?
Likely inputsEvery included facility, distance and direction where relevant, shared utility or watershed context, and uncertainty in each source record.
Coming soon · no live score yetWorking theories
Use ranges, not false precision
Each category would have a lower and upper bound. The lower bound would use documented facts; the upper bound would include clearly labeled conservative proxies when important facts are missing. The overall screening scale would likely be bounded from 0 to 100, but the thresholds and labels are not decided.
Distance should be category-specific
Sound and vibration may use steep distance decay informed by engineering propagation, while electricity and water may retain a regional floor because grids and watersheds do not stop at a parcel boundary. Air effects should use dispersion concepts, not a generic inverse-square shortcut.
Multiple facilities should accumulate
Two facilities 20 miles away should not look identical to one facility 20 miles away. A future address model could combine bounded contributions with a diminishing-returns rule, then separately show shared regional effects so cumulative impact is visible without pretending the inputs are independent measurements.
Thresholds need evidence
Minimum and maximum thresholds should come from defensible benchmarks where possible, not arbitrary UI settings. A missing value should widen uncertainty, not silently become zero impact. We still need to decide how thresholds vary by facility type, operating status, and measured versus reported data.
Community impact dimensions
The dimensions are inspired in part by Erin Brockovich’s community-focused environmental advocacy and Community Healthbook. That inspiration does not endorse this project’s categories, assumptions, or calculations. Wildlife and land context may be reported in the write-up, but a separate wildlife score could double-count effects already represented by land use, water, construction, traffic, zoning, and flood context.
Backlog additions
Heat production and illumination should be considered as separate categories. Waste heat may matter for local microclimate or equipment exhaust context, while illumination or light pollution may affect sleep, dark-sky conditions, and nearby residents. Neither is currently scored.
- Research usable heat-output and cooling-exhaust proxies.
- Research nighttime lighting, glare, shielding, and facility design evidence.
- Decide whether these belong in immediate impact, address impact, or both.
- Find public measurement and permitting sources before assigning ranges.
Open a GitHub issue with the assumption, evidence, and a concrete example. The goal is a model people can inspect and challenge, not a mysterious number that looks authoritative.