Imagine a school that lets every student drop their worst test of the year before the report card is printed. The reason for it makes sense. One bad day should not define a whole year of work. But not every bad test qualifies — only a test that is bad enough, and the bar for that is set by how badly the same student has done before. That is not a school. It is the standard method for measuring whether your electricity works. Last year one Indiana utility reported a hundred and seventy-eight minutes without power per customer — under three hours for an entire year. The number before the subtraction was five hundred and thirty-two minutes. And right now, about forty thousand homes and businesses in northwest Indiana are ending their second week in the dark.

On the morning of August eleventh, twenty twenty-six, a derecho — a long, straight-line windstorm — crossed from Iowa through Chicago into northwest Indiana. The strongest measured gust, ninety-nine miles an hour, hit the airport in Gary.

The utility there is NIPSCO, which serves about half a million customers. The storm took out more than sixty percent of them. In Gary alone, crews had to replace around three hundred poles.

Then came the part people actually remember. The automated outage system, the one that tells you whether anyone knows you are out, collapsed on the first two days. When the utility published its first list of restoration dates, Gary was not on it at all.

Seven people died across the state in the storms and the flooding around them.

Schools in Gary have been closed since August thirteenth. And in Gary, Munster and Highland, full restoration was promised for August twenty-fifth — two weeks after the wind stopped.

The argument that follows an event like this is already underway. The governor says the company needs to step up. A class action filed this week alleges the utility let trees grow into its lines instead of cutting them back on schedule. The company says the damage was unprecedented. All of that may matter. But the state's own regulator already publishes something nobody disputes: this utility charges the highest electricity rates in Indiana and has the worst reliability record in Indiana. Both were on the record before the storm. So the question is not whether anyone noticed. Everyone noticed. The question is why noticing changed so little — and that comes down to three rules which together decide what a dark hour is worth.

Start with how reliability is measured. The main number is simple: total minutes without power divided by total customers — the average time one customer spends in the dark in a year. Every utility reports it twice: once including major storm days, once with those days removed. The second version is the one that gets quoted.

Here are both, for this utility, last year. Including storms: about nine hours. Excluding them: three. Two-thirds of the dark sits outside the headline number — and this is from before the storm we have been talking about even happened.

The shape over time is the real story. Track the storm-free line back to twenty twelve and it has barely moved — three hours or under, year after year. The line that includes storms swings from under three hours in a good year to ten in a bad one, and the last two years sit near the top of that range. By one measure this system has been steady for fourteen years. By the other, it has never been steady at all.

Which days come out is decided by a formula written by engineers. It takes the daily outage numbers from the previous five years and draws a threshold near the top of that spread. Any day that goes above that line comes out of the number.

Here is the part that matters. That threshold is not a fixed standard of severity. It is built out of the utility's own past five years — so the same storm can count as an exceptional day for one company and an ordinary Tuesday for another, depending on how badly each has been performing already. The bar moves with a history the customer never sees.

The regulator's report gives the reasoning, and it is sound: removing those days strips out interruptions the utility has little or no control over. That is true of the wind. It is not true of the days that followed. And as long as enough customers are still out, those days cross the threshold too — so the storm and the response to it can leave in the same subtraction.

The second rule is about where the money comes from. A regulated utility does not profit on electricity the way a shop profits on bread. It earns an approved percentage — here, nine and three-quarters percent — on the capital it has invested. Build something, and you earn that percentage on it every year, for decades.

Trimming a tree is not capital. It is an operating expense — it keeps the branch out of the wire this year, and produces nothing you earn a return on.

Indiana then widened that gap on purpose. There is now a fast lane for money that goes into steel — and tree trimming is not allowed in it.

The lane was built to fix something real: waiting years for the commission's next full review of rates meant that replacing century-old equipment kept getting postponed. So a two thousand thirteen law changed the order. File a multi-year plan, get it approved, and you can raise rates every six months as the work goes in.

Tree work does not fit in that lane. When another Indiana utility tried to put its tree budget into the same kind of plan, the commission struck it out as ineligible.

So look at what the lane carried. The plan approved here is worth one point four billion dollars: substations, lines, transmission, metering. None of it is improper. It is the shape the rules reward — and the leading cause of outages here, by the company's own account, is trees.

The third rule is the shortest, because it is an absence. In Indiana no standing rule makes a long outage cost the utility anything. The regulations require reports — file your reliability numbers every March, keep the data seven years, report your tree-trimming budget once a year. They attach no payment to an hour without power.

So the price appears only afterward, and only through politics: an investigation, a lawsuit, a governor's statement. After a storm in two thousand one, a different Indiana utility agreed to pay a hundred dollars to every customer out thirty-six hours or more — a one-time arrangement, negotiated from scratch.

Put the three together and the loop is complete. The scoreboard subtracts the worst days, using a bar set by your own worst years. The money follows what gets built, not what stays standing. And a dark hour has no price until someone makes it a scandal. So what would it take to break that loop?

Three designs try to break it, and every one of them has a price.

The first design is to make the hour cost something. Michigan, one state over, has been doing it since twenty twenty-three. Since twenty twenty-three its commission has required an automatic bill credit — no application, no phone call — worth about forty dollars for each day you are out. Michigan customers now receive about seven times what they did before the rule.

The trade-off is steep. The credit only starts after a threshold, and the bigger the storm, the longer you have to sit before it starts — so in a storm the size of Indiana's, the first four days in the dark cost the company nothing. Forty dollars does not cover a lost freezer, a hotel night, or a missed shift. And the targets those utilities must hit are anchored to their own past results — the same trick again, with the goal built out of the company's own history. Michigan still has some of the slowest restoration times in the country.

Britain built the same design with sharper teeth. A customer cut off in a severe storm is paid ninety pounds — about a hundred and twenty dollars — once the outage passes twenty-four hours, then another forty-five pounds for every six hours after that, up to a limit. It keeps climbing while you sit in the dark, which buys urgency in hour thirty and hour sixty, when a flat payment has already stopped caring.

Its trade-off showed up the moment a real catastrophe arrived. During Storm Arwen the payments hit their ceiling while people were still without power, some for more than six days. The networks waived that ceiling themselves, by choice, after the fact — the Indiana problem again, in a country that thought it had designed its way out. At the extreme, the payment went back to being a decision somebody made afterwards.

Both versions of the first design share one limit: they pay you for the dark, they do not make it rarer. The second design leaves compensation alone and goes after the scoreboard.

Instead of an average across half a million customers, publish a headcount: how many were out more than a day, more than three days, more than a week. An average dissolves two weeks in Gary into a rounding error. You cannot average away forty thousand homes.

The trade-off is that this one is almost free, and almost free is the problem. A truer number changes nothing by itself, because a number with no money attached to it is still only a number. It arms the argument. It does not settle it.

The third design goes back to the second rule: the pole earns, the branch does not. So either let preventive work travel the same lane as steel, or tie part of the company's approved profit to what customers actually got. Indiana has already looked at tying profit to results, in a report its commission sent the legislature last year.

The trade-off is money. If tree trimming earns a return, you don't just pay for the trimming — you pay for it plus a profit on top, every year. Same saw, same branch, more expensive. And every widening of what counts as eligible invites the next argument about what else belongs in that lane, which is exactly why regulators keep striking items out of these plans.

Three designs. Pay for the hour. Count the people instead of averaging them. And let the work that prevents the outage earn what the steel earns. None of them is free.

None of this requires anyone to have behaved badly. The utility followed a measurement standard written by engineers, invested through a channel built by legislators, and earned a return set by regulators. The regulator published the bad numbers itself. The governor pushed. The lawyers filed. Everyone did their part, and forty thousand homes and businesses are still ending a second week in the dark. So when your utility next asks to raise your rates, there are two questions worth having ready. Which reliability number is in front of the people deciding — the one with the worst days taken out, or the one you lived through? And if it gets worse next year, what happens to the company? If nothing happens to the company either way, then that number was never a measurement. It was the same report card, with the worst test missing. ---