{
  "slug": "prolonged-grid-blackout",
  "question": "What are the odds of living through a prolonged power-grid blackout lasting days or longer?",
  "category": "other",
  "no_reliable_estimate": false,
  "perceived": {
    "description": "Public worry about a \"grid-down\" event is loud but oddly bimodal. One mode is the prepper-and-thriller framing — a cyberattack or solar storm that plunges a continent into darkness for weeks — which most people rate as remote and cinematic. The other mode is the mundane multi-day outage after a hurricane, ice storm, or derecho, which most people treat as a weather nuisance rather than a \"blackout\" at all and therefore underweight as a lifetime probability. The result is a perception focused on the dramatic cause (will hackers take down the grid?) while the far more likely path — a regional weather-driven outage lasting one to several days — is mentally filed under routine inconvenience. No major survey isolates worry about a multi-day outage from any cause; the closest proxies (Chapman's cyberterrorism and power-grid-failure items) measure the catastrophic framing, not the ordinary one.\n",
    "rough_estimate": "Most adults rate a multi-day outage as a weather nuisance, not a lifetime probability; the catastrophic 'grid-down' framing is what draws worry",
    "kind": "intuition"
  },
  "native": {
    "display": "~5% per year that a US household experiences a power outage lasting 24 hours or longer",
    "numerator": 5,
    "denominator": 100,
    "unit": "per household-year (outage lasting ≥24 hours, any cause)",
    "population": "US households, multi-day (≥24h) electricity outage"
  },
  "normalized": {
    "lifetime_us_adult": 0.5,
    "display": "~1 in 2 lifetime (US adult experiencing a ≥24h regional blackout)",
    "log_value": -0.301,
    "assumptions": "This entry isolates the cause-agnostic exposure question — the probability that a typical US adult personally lives through at least one prolonged (≥24-72 hour, up to weeks) regional grid outage over an adult lifetime, from ANY cause: severe weather, equipment failure or cascading fault, cyberattack, or solar storm. It deliberately does not condition on the cause, which separates it from cyberattack-infrastructure and solar-storm-carrington-event, where a blackout is the consequence of one specific trigger. Step 1, native annual rate: the 2023 American Housing Survey (Census/HUD) found 1 in 4 US households (33.9 million) had at least one complete outage in the prior 12 months, and ~70% of those (23.6 million, or ~18% of all households) had an outage lasting 6 hours or longer. Narrowing from the 6-hour threshold to the ≥24-hour threshold the question fixes: multi-day outages are the weather tail of that distribution. Do et al. (Nature Communications 2023) found 62.1% of 8+ hour outages co-occur with extreme weather and that 70.5% of US counties saw at least one 8+ hour outage over 2018-2020, with a county median of 2 per year. Taking roughly a quarter to a third of the ~18% of households with a 6+ hour outage as reaching ≥24h gives a central native rate of ~5% per household-year (1 in 20). Step 2, lifetime: the per- person annual rate is strongly location-dominated and autocorrelated — your grid, your weather region, and the fact that adults rarely relocate set the rate — so naive compounding of a population-average annual rate over 59 remaining adult years (1-(1-0.05)^59 ≈ 0.95) overstates the population-average lifetime probability by Jensen's inequality. Instead the lifetime figure is bracketed two ways. Lower anchor, catastrophic regional events only: the 2003 Northeast blackout reached ~50 million customers (DOE) and Winter Storm Uri in 2021 shed 23,418 MW (FERC/NERC, the largest firm load shed in US history) affecting millions for up to four days; major regional multi-day events of that class recur roughly every few years and each exposes tens of millions, giving a lifetime exposure to a headline-scale event of ~0.3-0.4. Upper anchor, all weather-driven multi-day local outages (residents of hurricane, ice-storm, and derecho regions accumulate these across decades): ~0.6-0.85. The central estimate of ~0.5 (1 in 2) sits between these, with a band reflecting the heterogeneity between grid-reliable low-risk areas and the hurricane/ice belt.\n",
    "uncertainty": {
      "low": 0.25,
      "high": 0.8
    },
    "scope": "us_adult_lifetime"
  },
  "sources": [
    {
      "url": "https://www.census.gov/library/stories/2024/10/power-outages.html",
      "title": "About 1 in 4 Households Experienced a Power Outage in the Span of a Year",
      "publisher": "U.S. Census Bureau (2023 American Housing Survey, sponsored by HUD)",
      "source_type": "govt_report",
      "statistic": "1 in 4 US households (33.9 million) had a complete power outage in the prior 12 months; ~70% (23.6 million) had an outage lasting 6 hours or more",
      "excerpt": "\"About 33.9 million or 1 in 4 households nationwide reported they were completely without power at least once in the 12 months before they were interviewed for the 2023 American Housing Survey (AHS). … Approximately 70% or 23.6 million of the households reporting an outage said at least one outage lasted 6 hours or more.\"\n",
      "source_date": "2024-10-02",
      "source_accessed": "2026-06-21",
      "archive_url": "http://web.archive.org/web/20260619065127/https://www.census.gov/library/stories/2024/10/power-outages.html",
      "calculation_notes": "The AHS is the cleanest household-level base rate for outage exposure: nationally representative, self-reported, and duration-bucketed. The publicly summarized figure tops out at the 6+ hour category (~18% of all households per year). This entry narrows from that 6-hour threshold to the ≥24-hour threshold the question fixes by treating multi-day outages as the long weather tail of the 6+ hour distribution (see Do et al.), estimating roughly a quarter to a third of 6+ hour household-outages reach ≥24h, i.e. ~5% per household-year. The AHS sets the upper envelope: if 18% of households see a 6+ hour outage in a single year, multi-day exposure over a 59-year adult life is common, not rare.\n",
      "independence_note": "Household survey methodology (self-report, nationally representative sample), independent of the utility-reported SAIDI data and the county-level outage-detection methodology in Do et al.\n"
    },
    {
      "url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC10147900/",
      "title": "Spatiotemporal distribution of power outages with climate events and social vulnerability in the USA",
      "publisher": "Nature Communications / Do V, McBrien H, Flores NM, et al.",
      "source_type": "peer_reviewed",
      "statistic": "62.1% of 8+ hour outages co-occur with extreme weather/climate events; 70.5% of studied US counties (73.7% of population) had at least one 8+ hour outage over 2018-2020, county median 2 per year",
      "excerpt": "\"62.1% of 8+ hour outages co-occur with extreme weather/climate events, particularly heavy precipitation, anomalous heat, and tropical cyclones. … 8+ hour outages were 3.4x more common on days with a single event and 10x more common on days with multiple events.\"\n",
      "source_date": "2023-04-29",
      "source_accessed": "2026-06-21",
      "archive_url": "http://web.archive.org/web/20260313115103/https://pmc.ncbi.nlm.nih.gov/articles/PMC10147900/",
      "calculation_notes": "Do et al. analyzed ~231,000 1+ hour outages and ~17,500 8+ hour outages across 2,447 counties (73.7% of US population) from 2018-2020 using utility outage-tracker data. The key contribution for this entry is the bridge from \"any 6+ hour outage\" to \"prolonged multi-day outage\": 8+ hour outages are weather-driven (62.1% co-occur with extreme weather) and the multi-day tail is the hurricane/ice/derecho tail of that same distribution. The finding that 70.5% of counties saw at least one 8+ hour outage in just three years independently corroborates that long-duration outage exposure is common over a lifetime, supporting a lifetime estimate near 1 in 2 rather than the low single-digit annual rate alone would naively suggest.\n",
      "independence_note": "Uses county-level utility outage-tracker detection, methodologically independent of the AHS self-report survey and the FERC/NERC and DOE event reconstructions.\n"
    },
    {
      "url": "https://www.eia.gov/todayinenergy/detail.php?id=66744",
      "title": "Hurricanes in 2024 led to the most hours without power in the United States in 10 years",
      "publisher": "U.S. Energy Information Administration",
      "source_type": "govt_report",
      "statistic": "US customers averaged 11 hours of electricity interruptions in 2024 (~2x the prior decade's average); major events accounted for 80% of those hours; major-event interruptions averaged ~9 hours vs ~4 hours/year in 2014-2023",
      "excerpt": "\"U.S. electricity customers experienced an average of 11 hours of electricity interruptions in 2024, or nearly twice as many as the annual average experienced in the decade before. … Major events such as Hurricanes Beryl, Helene, and Milton accounted for 80% of the hours without electricity in 2024.\"\n",
      "source_date": "2025-12-01",
      "source_accessed": "2026-06-21",
      "archive_url": "http://web.archive.org/web/20260618035145/https://www.eia.gov/todayinenergy/detail.php?id=66744",
      "calculation_notes": "EIA's SAIDI data establishes two things this entry relies on. First, the trend: average interruption hours doubled from ~5.6 (2022) to 11 (2024), driven almost entirely by major weather events (80% of 2024 hours), which is the era/climate factor behind the rising-trend multiplier. Second, the structure: routine non-major interruptions hold steady at ~2 hours/year, so the multi-day outages this entry counts are overwhelmingly the major-event tail, not everyday flickers. SAIDI is an average-hours metric, not a per-household incidence rate, so it is used here for trend and attribution rather than as the native probability — that comes from the AHS.\n",
      "independence_note": "Utility-reported SAIDI/SAIFI aggregated by EIA, independent of the AHS household survey and the academic county-level analysis.\n"
    },
    {
      "url": "https://www.lloyds.com/insights/risk-reports/business-blackout",
      "title": "Business Blackout: The insurance implications of a cyber attack on the US power grid",
      "publisher": "Lloyd's of London / University of Cambridge Centre for Risk Studies",
      "source_type": "reputable_reference",
      "statistic": "Hypothetical Erebos cyberattack on 50 Northeastern US generators leaves 93 million people across 15 states (including NYC and Washington DC) without power; ~$243bn to >$1trn US economic impact",
      "excerpt": "\"The attackers are able to inflict physical damage on 50 generators which supply power to the electrical grid in the Northeastern USA, including New York City and Washington DC. … it triggers a wider blackout which leaves 93 million people without power. … The total impact to the US economy is estimated at $243bn, rising to more than $1trn in the most extreme version of the scenario.\"\n",
      "source_date": "2015-07-06",
      "source_accessed": "2026-06-21",
      "archive_url": "http://web.archive.org/web/20260312044639/https://www.lloyds.com/insights/risk-reports/business-blackout",
      "calculation_notes": "This source supplies the cause-agnostic tail: a cyberattack is one of the four causes this entry folds together, and the Lloyd's/Cambridge Erebos scenario quantifies the worst-case cyber path (93 million people, 15 states). It is a scenario study, not a frequency estimate, so it does not enter the probability arithmetic — it bounds the severity tail and demonstrates that the multi-day regional blackout this entry counts is reachable by cyber means, not only weather. No publicly fetched figure supports a \"1:200 return period\" attribution, so that claim is deliberately not cited here.\n",
      "independence_note": "Insurance-industry scenario modelling by the Cambridge Centre for Risk Studies, independent of all three empirical outage datasets above.\n"
    }
  ],
  "comparison_anchors": [
    {
      "label": "Data breach exposure (lifetime, US adult)",
      "lifetime_us_adult": 0.95
    },
    {
      "label": "Identity theft (lifetime, US adult)",
      "lifetime_us_adult": 0.6
    },
    {
      "label": "Carrington-class solar storm reaching Earth (lifetime)",
      "lifetime_us_adult": 0.531
    },
    {
      "label": "Death in a car crash (lifetime, US)",
      "lifetime_us_adult": 0.0108
    }
  ],
  "personal_factor_multipliers": [
    {
      "factor": "resident of a hurricane-prone Gulf or Southeast Atlantic state (FL, LA, TX coast, NC)",
      "multiplier": 2.5,
      "notes": "EIA (2024) attributes 80% of 2024 outage-hours to hurricanes Beryl, Helene, and Milton; Do et al. (Nature Comms 2023) identify Louisiana and Arkansas among counties with the most frequent 8+ hour outages. Hurricane-belt residents accumulate multi-day outages at multiple times the national rate."
    },
    {
      "factor": "rural customer on a long radial distribution feeder",
      "multiplier": 2,
      "notes": "EIA SAIDI data and the USDA/cooperative reliability literature show rural radial feeders have longer restoration times (no network redundancy, longer line miles per customer) — roughly 2x the duration of urban networked service after a storm."
    },
    {
      "factor": "Texas / ERCOT islanded interconnection resident",
      "multiplier": 1.5,
      "notes": "ERCOT's near-isolation from the Eastern and Western Interconnections removed the import buffer during Winter Storm Uri (FERC/NERC 2021: 23,418 MW firm load shed, largest in US history), exposing ERCOT customers to a multi-day blackout that interconnected neighbors avoided."
    },
    {
      "factor": "resident of a high-reliability urban networked grid (e.g. dense city center, undergrounded)",
      "multiplier": 0.4,
      "notes": "EIA SAIDI by state spans from ~72 minutes/year (District of Columbia) to >1,800 minutes/year (Maine); dense undergrounded urban networks with redundant feeders sit at the low-duration end and rarely see multi-day outages outside catastrophic events."
    },
    {
      "factor": "rising-trend / climate exposure for a young adult (full 59-year horizon ahead)",
      "multiplier": 1.4,
      "notes": "EIA (2024): average interruption hours roughly doubled vs the prior decade, driven by major weather events; a 20-year-old faces six decades of an upward-trending hazard, raising cumulative exposure above the static-rate estimate."
    }
  ],
  "short_label": "Prolonged grid blackout",
  "myth_framing": "calibrated",
  "outcome_severity": "moderate_harm",
  "exposure_pattern": "recurring",
  "outcome_type": "inconvenience",
  "valence": "negative",
  "caveats": "The headline ~1 in 2 lifetime figure is the probability of personally experiencing the outage, not of being harmed by it. Most multi-day outages are an expensive inconvenience: spoiled food, no heat or cooling, lost work, and a few days of disruption. The mortality tail is real but far smaller — Winter Storm Uri was associated with at least 246 deaths in Texas (and some estimates run higher), concentrated among the elderly, the medically power-dependent, and households without alternate heat. That tail, not the typical event, is what the death-from-heat or death-from-cold framings would capture. Two structural caveats apply to the number itself. First, it is deliberately cause-agnostic: weather drives the overwhelming majority of multi-day outages (Do et al.: 62.1% of 8+ hour outages co-occur with extreme weather; EIA: 80% of 2024 outage-hours from hurricanes), while cyberattack and solar storm are low-frequency tails included for completeness — the Lloyd's/Cambridge Erebos scenario shows the cyber tail is reachable but it is a scenario, not a frequency. Second, the lifetime figure is bracket-derived rather than compounded: multi-day outage risk is dominated by where you live and is highly autocorrelated, so the population average masks a wide spread — a hurricane-belt rural customer may be near-certain to see several multi-day outages, while a customer on a dense undergrounded urban network may go a lifetime without one outside a catastrophic regional event. The ~5% native annual rate is itself an estimate that narrows the AHS 6+ hour figure to the ≥24h threshold using the weather-tail structure in Do et al.; it is the largest single source of uncertainty in the entry.\n",
  "quality_score": {
    "d1": 5,
    "d2": 5,
    "d3": 4,
    "d4": 4,
    "d5": 5,
    "d6": 5,
    "d7": 4,
    "d8": 5,
    "avg": 4.625,
    "scored_by": "claude-code-8d",
    "scored_at": "2026-06-21",
    "methodology_version": "1.2"
  },
  "reviewer": "8d-eval-2026-06-21",
  "last_reviewed": "2026-06-21",
  "reviewed": true,
  "generated_at": "2026-06-21",
  "image": {
    "alt": "A simplified city skyline in silhouette with most windows dark and a few faintly lit, flat vector illustration, muted tones."
  },
  "attribution": "Likelier — https://likelier.app",
  "license": "https://creativecommons.org/licenses/by-sa/4.0/",
  "support": "https://buymeacoffee.com/kgluszczyk?via=likelier&utm_content=api-fear-single",
  "canonical_url": "https://likelier.app/prolonged-grid-blackout"
}