Fourteen Days in the Dark: What Gary, Indiana Tells Us About Who Gets Rescued

By Brandi Richard Thompson Former Federal Emergency Management Official

On August 11, a derecho crossed into northwest Indiana with a 99 mph gust recorded at the Gary/Chicago International Airport. It took down roughly 300,000 NIPSCO customers, about sixty percent of the utility's entire electric system, in a single afternoon. In Lake County, initial reports put the figure near ninety-five percent of the county without power.

The same storm crossed into Illinois. ComEd restored nearly all of its customers within days, and eventually sent Illinois crews across the state line to help NIPSCO finish the job.

In Gary, power was not fully restored until August 25. Fourteen days.

Sustained national coverage did not arrive until roughly August 21, day eleven. By that point, 22,000 Gary households had already spent more than a week in the dark, and the restoration curve was well past its steepest point. The Black press was there from day two. Capital B Gary published on August 12. The national desks arrived when the story was nearly over.

I spent more than twenty years in federal emergency management, most of it at FEMA and DHS, running crisis communications across a region of 51 million people through more than a hundred presidential disaster declarations. I have watched this sequence enough times to know it is not a coincidence and not a scandal. It is a system operating exactly as designed. The design is the problem.

I. The math of who this happened to

A two-week power outage is not one event. It is actually a different event depending on who it happens to.

If it happens to a household with two vehicles, a credit card with room on it, a friend in an unaffected suburb, and enough savings to absorb four hundred dollars of spoiled food and six nights in a hotel, it is a miserable, expensive inconvenience. That household will tell the story at Thanksgiving.

Gary is not necessarily that household.

The census figures are not ambiguous. Gary's median household income is $38,731, less than half the national median of $80,734. The poverty rate is 33.1 percent. Child poverty is 49.0 percent, a number I confirmed two independent ways before sharing it here, because it is the kind of number people assume is a typo. Nearly half the children in this city live below the federal poverty line.

Thirty percent of Gary households receive SNAP. Nearly thirty-five percent include someone with a disability. Sixteen percent of the population is over 65, and 37.6 percent of households are a single person living alone. That is the statistic that should stop you, because it means that in a citywide blackout, more than one in three Gary households has no one else in the house to notice if something goes wrong.

And 21.7 percent of Gary's renter households have no vehicle at all. More than one in five renting families could not have driven to a cooling center, a relative's house, a hotel, a working refrigerator, or a pharmacy. The entire American preparedness apparatus, the checklists and the go-bags and the "have a plan to evacuate" language I have written and approved and delivered from a podium, assumes a car. In Gary, for a fifth of renters, that assumption is simply false.

Layer these facts and the picture gets clearer. A two-week outage in Gary is not an inconvenience story. It is fourteen days of an elderly person alone in an apartment with no air conditioning, no way to leave, insulin warming in a dead refrigerator, and a month's grocery budget spoiling on the shelf with no realistic prospect of replacing it.

II. Gary did not decline. Gary was decided.

There is a lazy way to write about Gary, and most national coverage takes it: the Rust Belt city that fell, the cautionary tale, the ruin photography. The passive voice does an enormous amount of work in that framing. The jobs left. The population declined. The tax base eroded.

Jobs do not leave. People move them. Tax bases do not erode. Legislatures write formulas.

Gary was built in 1906 by U.S. Steel, named for its chairman, as a company town. It grew through the Great Migration to a peak of 178,320 residents in 1960. The population is now roughly 67,000, a loss of about 62 percent. That trajectory is usually presented as weather. It was policy.

1940. The federal Home Owners' Loan Corporation graded Gary's neighborhoods. Eight percent received an ‘A’. Forty-two percent were graded ‘D’, redlined. The surveyors wrote their reasoning in plain language. The Midtown district, then roughly eighty percent Black, was graded D and described as "an old mixed area formerly alien," with homeownership at twenty percent. The Horace Mann area was graded A and marked "Restricted." Under the field labeled "Negro %," the surveyors wrote: "None." Black Oak, graded D, has a trend field that reads, in its entirety: "Shacks."

That is not a metaphor for structural disinvestment. That is the federal government putting structural disinvestment in writing and filing it.

1967. Richard Hatcher was elected mayor, one of the first Black mayors of a major American city. He took roughly seven percent of the white vote. Over his twenty years in office, Gary lost approximately fifty thousand residents.

1971. Merrillville incorporated, explicitly to prevent annexation by Gary. Roughly a billion dollars in businesses and banking relocated there. Downtown Gary went from about 500 retail stores in 1969 to about 40 by 1980.

The steel. U.S. Steel's Gary operations employed roughly 30,000 at peak. Today the figure is around 4,000, a decline of about 87 percent. The plant is still there. The jobs are not.

2008. And here is the piece almost nobody writes about, because it is a tax formula rather than a photograph.

Indiana capped property taxes at one percent of assessed value for homesteads, two percent for other residential, three percent for commercial, enacted by statute in 2008 and written into the state constitution by referendum in 2010. Amounts above the cap are not deferred. They are simply never collected.

For a city where assessed values had already collapsed and municipal costs had not, the effect was catastrophic and entirely predictable. According to an analysis filed with Indiana's own Distressed Unit Appeal Board, Gary's certified tax levy in 2010 was $62.9 million. The circuit breaker credit, the money the caps forgave, was $34.8 million. The city collected $28.1 million. It lost 55 percent of its property tax levy in a single year, and roughly half of it every year in that window.

The state's own fiscal monitor, reviewing Gary in December 2009, found that property taxes made up about eighty percent of the city's General Fund operating revenue, projected a structural deficit of roughly $22 million a year, and wrote the sentence that should end every conversation about why Gary's recovery looked the way it did:

Annual property tax revenues would be less than the present annual cost of providing police, fire and emergency medical services.

Read that again with a blackout in mind. Indiana structurally defunded Gary's capacity to respond to exactly this, sixteen years before it happened, and put the arithmetic in a public filing.

So when a two-week outage hits and the city cannot surge its own response, cannot stand up cooling centers at scale, cannot canvass door to door for the elderly living alone, cannot absorb the cost of feeding thirty thousand people, that is not municipal incompetence. That is a policy outcome arriving on schedule.

III. The utility, and what the ratepayers had just bought

In September 2024, NIPSCO filed for a rate increase of roughly $368.7 million. The Indiana Utility Regulatory Commission approved about $257 million of it in June 2025: a 16.75 percent residential increase, about $23 a month, phased in through the first quarter of 2026.

Rate cases like this are argued on reliability. The company had proposed roughly $769.5 million in electric transmission and distribution upgrades, technology improvements, and safety and reliability initiatives. That is the pitch: pay more now, and the system holds.

The phase-in completed in Q1 2026. The system failed in August 2026. Gary residents had finished absorbing a reliability-justified rate increase five months before the largest outage in the utility's history.

During the outage, NiSource, NIPSCO's parent, announced $140 million in dividends.

Governor Braun has directed the state's utility consumer advocate to petition the IURC to investigate NIPSCO's vegetation management, right-of-way maintenance, and the part that matters most, whether money regulators approved for infrastructure improvements was actually spent as intended. A proposed class action was filed in Porter County Superior Court alleging that failed tree trimming and line inspection made the outage "predictable, obvious and preventable." NIPSCO disputes the characterization and says its vegetation management spending has more than doubled since 2016. Mayor Eddie Melton, who worked at NIPSCO for fifteen years, put it more plainly: "It does not take that long to turn the lights on."

I want to be careful here, because precision is what makes this argument survive contact with people who do not want to hear it.

The derecho intensified as it crossed into northwest Indiana. Some of the difference between ComEd's days to restore power and NIPSCO's weeks is storm intensity, not corporate character. Say so. The argument does not need the exaggeration.

And on August 15, at roughly 12:20 a.m., two NIPSCO employees scouting restoration work in Gary were fired upon. Their vehicle was struck twice; one round lodged in a headrest. No one was injured. NIPSCO paused Gary restoration and resumed around 4 p.m. the same day with additional security coordinated by Gary Police and Indiana State Police. Mayor Melton called it unacceptable, and it was.

That incident is real, it was wrong, and it has been picked up by outlets arguing that it explains everything. It does not. The pause lasted part of one day. The outage lasted fourteen days. Gary still had 22,000 customers dark on August 19, four days after work resumed, and by August 25 more than seventy percent of the county's remaining outages were still inside Gary's city limits. A half-day security hold does not produce that curve. The dates are the argument.

That final figure is the one I would put in an after-action report, because it is a fact about restoration sequencing and it requires no one to have intended anything. Utilities restore in an order. That order is driven by circuit topology, by customer density per repair, and, as one peer-reviewed analysis of a Midwestern utility found, by proximity to critical facilities like hospitals and fire stations. A city that has lost 62 percent of its population, has 28 percent of its housing units sitting vacant, and has fewer institutional anchors than its neighbors will be restored last by a neutral algorithm. Nobody has to decide to discriminate. The map does it for you.

That is what structural means. It is not a synonym for malicious. It is worse, because there is no one to fire.

IV. What the media hierarchy actually is

There is no study I could find that measures the volume of national coverage Gary received against comparable-scale outages elsewhere. I am not going to assert one. But the publication record is not in dispute, and it says enough.

Local Chicago-market outlets covered it from day one. Capital B Gary, a Black-press nonprofit, published on day two. The Chicago Sun-Times, day nine. NPR's first sustained national piece, day eleven. NBC Nightly News and the network desks, day thirteen. CNN, Essence, and Ebony on day fifteen.

The disparity framing was raised by Gary's own clergy, by Mayor Melton, by Bernice King, and by the Black press before it appeared on national television. Reverend John Jackson called it "Katrina 2.0" and said what everyone in the city was already saying: "There has been not enough attention to Gary in this moment." Bernice King was more precise: "It is part of a pattern. Black communities have too often been expected to endure conditions that would be treated as intolerable somewhere else."

I have run the media side of enough disasters to explain the mechanism without any conspiracy in it, which makes it more damning rather than less.

A disaster becomes nationally legible when it produces one of four things: a body count that crosses a threshold, footage of dramatic physical destruction, a recognizable place, or an aggrieved constituency that news organizations already treat as their audience.

A blackout produces none of those. There is no footage of a blackout. Darkness photographs as nothing. The harm is distributed across thousands of interiors, a warm refrigerator, an oxygen concentrator that stopped, a child who did not eat dinner, and none of it makes a frame. The deaths, if they come, arrive weeks later as heat stress and cardiac events and missed dialysis, and they are recorded as medical, not disaster. A slow, invisible, interior catastrophe in a place that has already been written off as a ruin is, in newsroom terms, not a story. It is a condition.

And this matters far beyond feelings, because coverage is not decoration in the disaster system. Coverage is an input. Attention drives political urgency, political urgency drives declarations and resource allocation, and declarations drive money. When a disaster is not legible, it does not merely go unwitnessed. It goes underfunded. The people in the dark pay for the fact that darkness does not film well.

V. And when the help does come, the formula is already tilted

Here is the part I am obligated to say precisely because of where I worked.

The disaster assistance system in this country is not neutral, and this is not an outside critic's claim. FEMA's own National Advisory Council, a statutory body rather than an advocacy group, wrote in November 2020 that FEMA's programs "provide an additional boost to wealthy homeowners and others with less need, while lower-income individuals and others sink further into poverty after disasters," and that "through the entire disaster cycle, communities that have been underserved stay underserved and thereby suffer needlessly and unjustly."

The research is consistent with the confession. Junia Howell and James Elliott, examining a national panel from 1999 to 2013, found that in counties receiving substantial FEMA aid, white residents accumulated roughly $55,000 more wealth while Black residents accumulated roughly $82,000 less. Not merely that aid failed to close the racial wealth gap, but that more federal disaster aid in a county was associated with a wider gap. Separate work analyzing FEMA's Individuals and Households Program from 2006 to 2018 found that damage cost, not household vulnerability, was the dominant predictor of how much aid was disbursed, and that where the program appeared to overperform, disbursement fell as the share of Black and Asian households rose.

The mechanism is not malice. It is arithmetic. Assistance formulas key on property damage, and property is precisely where two centuries of inequality are already stored. A household with a $332,700 home and full insurance generates a large, well-documented, formula-legible claim. A household in a city with a $94,700 median home value, thin or no coverage, and losses that consist of food, medication, wages, and a fried window air conditioner generates almost nothing the system knows how to see.

Which brings us to the most concrete injustice of this particular disaster, and the one I put at the top of the resource sheet we built to go with this piece.

FEMA does not reimburse spoiled food. That is FEMA's own published position. NIPSCO's published policy is that it does not reimburse food spoilage or property damage from storms, which it classifies as acts of nature. So the single largest, most universal, most immediate loss suffered by tens of thousands of Gary households, the contents of the refrigerator, is covered by neither the federal government nor the utility whose system failed.

The routes that remain are homeowners or renters insurance, where a typical sublimit is around five hundred dollars and many policies pay only if the failure occurred on your own property, and SNAP replacement benefits, which required every affected household to file an individual affidavit by August 31. Nothing was automatic. In a city where 30 percent of households receive SNAP, an individually-filed, four-day-notice affidavit is not a benefit. It is a filter.

That is what it looks like when a system is not designed for you. Not a locked door. A form.

VI. What preparedness has to mean now

I build preparedness systems for a living, and I am going to say the uncomfortable thing about my own field.

Most preparedness guidance in this country is written for a household that has a car, a credit card with available balance, a second location it can reach, a smartphone with a data plan, and seventy-two hours of financial slack. Evacuate early. Keep two weeks of supplies. Have a plan to relocate. Every one of those instructions is sound, and every one of them silently presumes resources that a third of Gary does not have.

Preparedness written for people who already have options is not preparedness. It is a lifestyle product.

Gary is the case study for what the field has to become, and I would put it in four lines.

One: assume help is late, and build for the gap, not the response. The federal system is a reimbursement machine that arrives in weeks and pays on documentation. It is not a rescue. Anyone who plans around cavalry arriving on day two is planning for a disaster that does not exist. Plan for day fourteen.

Two: documentation is a survival skill, and nobody teaches it. The single highest-return action available to a Gary household right now costs nothing and takes ten minutes: photograph the inside of the refrigerator before emptying it, itemize the loss, keep every receipt, screenshot the outage timestamps. Insurance, FEMA, the state relief fund, the utility regulator, and any eventual settlement will all demand the same evidence, and the households that have it will be made whole in a way the households that do not simply will not. That gap is not about who suffered more. It is about who had a camera roll and knew to use it. We can close that gap with a piece of paper, and we should have already.

Three: the neighbor is the infrastructure. In a city where 37.6 percent of households are one person living alone, more than a third contain someone with a disability, and a fifth of renters have no car, the operative unit of resilience is not the household. It is the block, the congregation, the barbershop, the mutual aid table. Roughly forty Gary churches organized around utility accountability during this outage. The Salvation Army, the YWCA, Catholic Charities, the Food Bank of Northwest Indiana, and dozens of congregations fed people for two weeks. That is not a charitable footnote to the real response. That was the response, and any preparedness framework that treats it as a supplement to institutional capacity has the hierarchy inverted.

Four: the complaint is part of the plan. The most underused tool available to Gary residents is not a generator. It is the Indiana Utility Regulatory Commission's complaint portal. The Governor has directed a formal investigation into whether infrastructure money was spent as approved. That proceeding will produce a record, and the record will be built from whatever is in it. A docket full of Gary addresses, dates without power, medication lost, a specific elderly parent in a specific hot apartment, is materially harder to dismiss than a press conference. Teaching people to file is preparedness for the next outage, which is the only kind of preparedness that compounds.

VII.

The lights came back on in Gary on August 25. That same day, the President signed a major disaster declaration covering Lake County among twenty others for individual assistance, and eleven counties for public assistance.

The people who spent those fourteen days at 85 degrees in a dark apartment, throwing away food they could not replace, are now being asked to prove what they lost to at least four different institutions, on four different forms, under four different sets of rules, at least one of which will not cover the thing they actually lost.

We will not fix that with a better checklist. But the checklist is what we can put in someone's hand this week, and the analysis is how we make sure that the next time this happens, and it will happen again, to a place that looks a great deal like Gary, the argument is already made and the evidence is already sitting in a docket.

Fourteen days is not a natural disaster. Fourteen days is a decision, made in stages, over a century, by people who mostly did not think of themselves as making it.

The work is to make the next fourteen days someone's decision on purpose.