Research
Living reports computed continuously from the same data that powers the live outage map — they recompute as events complete, so the numbers are always current. Every report publishes its methodology, sample sizes, and update dates; aggregate findings are publicly viewable and citable under stated terms.
How often do utilities push back their restoration times?
Utility-posted restoration times graded against actual restorations: how often the
posted time slips later, by how much, and how that compares to our independent estimates on the same events.
Current findings (71 completed outages with a posted time, from the 600 most recently graded, about 3.2 days of outages; data updated Oct 8, 2026): of the 61 whose posted time we could follow outage by outage, the utility moved it later at least once in 62% (2.5 later revisions per outage on average) · on the 68 where both estimates were on record at the same readings, the utility's posted time was off by ~1 h, ours by ~2 h (ours was the closer estimate on ~37% of them). Full analysis →
More reports publish as the dataset matures — per-utility reliability tables arrive when each named utility has at least 10 graded events. See also: how our own estimates are graded · data sources & licensing.