Decide the restoration order before the typhoon arrives
Most regional networks restore service in an order improvised by tired people at night. That order should be agreed in advance with the towns a network serves and approved by its board every year, with the cost of putting some customers later stated plainly.
Every operator in Kyushu has a typhoon plan. Most describe how to stage crews, where the generators are kept and who calls whom. Far fewer state, in writing, the order in which the network will come back when it cannot all come back at once. My position is that this order is the most consequential operating decision a regional network makes in a disaster, that it is usually made by the wrong people at the wrong time, and that it should be agreed in advance with the municipalities the network serves and approved by its board each year, with its costs stated.
I hold this view because I have made the decision badly, and later made it better under pressure, and I would not want a colleague to repeat either experience.
What gets decided at two in the morning
In March 2011 I was running access-network field operations for a fixed-line carrier in eastern Japan. When the tsunami destroyed our plant along the coasts of Miyagi and Iwate, we had a short list of hospitals and government buildings to restore first. Beyond that list, the order was set by whatever was possible on a given morning: which roads were open, which crews had fuel, which customers rang the operations room most often. Large business customers had account managers who knew my mobile number. Households did not. Nobody chose that order. It emerged, and when I reviewed it months later I could not defend parts of it.
Five years later, in the early hours of 16 April 2016, the second of the Kumamoto earthquakes cut service to 118,000 households on the network I was then responsible for at Hanagiri Networks. I had extended the list by then, but it still covered only hospitals and municipal offices. By dawn the question was whether to send our first available crews to the business parks, where we had contracted service levels and credits to pay, or to the evacuation centres and the community channel's disaster feed, which carried municipal notices to older residents with no other dependable source. I chose the second. Our enterprise sales head argued, correctly, that we were breaching contracts. Our finance team estimated the credits. The president backed the decision on the second morning.
I still think the choice was right. I do not think it was right that I made it alone, after a night without sleep, with the commercial argument in the room and the municipalities' view only in my head. A decision with consequences of that size belongs to people who can weigh it calmly and answer for it afterwards.
A list the towns and the board have signed
Since 2017 Hanagiri has kept a written restoration order. It sets out, in tiers, what we restore first: hospitals and dialysis clinics, municipal disaster headquarters, the community channel's playout and its links to municipal offices, and designated evacuation centres. Welfare facilities and the homes of residents registered as needing help to evacuate come next. After that we restore households and businesses by the number of connections a crew-day can bring back. Contracted business customers sit in that third tier unless they are part of the response themselves, such as fuel depots or supermarkets a municipality has designated.
We review the order each spring with the fourteen municipalities in our area, using their disaster-prevention plans and their own lists of priority facilities. The review is not ceremonial. Towns move evacuation centres, open welfare facilities and sometimes ask us to raise a site we had placed low. Last year one town asked us to put the monitoring link for a water-treatment plant in the first tier, and we did. After the review the order goes to our board with an estimate of what it would cost in a severe event: the credits we would pay, the business customers we might lose and the revenue we would defer. The board approves it as policy, not as an operating detail.
Two things change when the order is signed in advance. The first is speed. Crews can be staged ahead of landfall on the forecast track with a clear list of where they go first, and nobody in the operations room has to argue a principle while the wind is still rising. The second is accountability. When a business customer's chief executive rings me after a storm, I can explain that the order was agreed with the towns and approved by our board, and send him the policy. Most accept that. A few do not, and they are entitled to choose another provider.
The customers who will wait longer
The objection I take most seriously is commercial. A regional operator earns a large share of its margin from business customers, and they pay for service levels precisely because they want to be restored quickly. If we tell them in advance that evacuation centres come first, some will buy from a national carrier. We lost two enterprise contracts after the 2016 earthquakes, worth about ¥180 million a year, and paid ¥310 million in credits.
Our answer has been to tell business customers before they sign. Our contracts describe the restoration tiers and offer diverse routing, at a higher price, to customers whose resilience should not depend on our order at all. That has cost us some sales, and our enterprise team still regards it as a handicap in competitive bids. It has also won us customers, including a hospital group and two logistics firms, who would rather buy from a provider that has thought about the question.
The second objection is that a written order takes judgement away from the field. In practice it does the reverse. The order ranks categories; it does not tell a crew leader which pole to climb on a flooded road. Field supervisors keep the authority to depart from it for safety or practicality, and they record why. During the typhoon of late August 2024 they departed from it eleven times, and our board saw each departure in the review that followed.
The third objection is that no order can anticipate the actual disaster. That is true, and it is why we review the order every year and rehearse it with municipal disaster staff. A plan that fits last year's flood will not fit next year's earthquake. What it can do is settle the principle, so that the people working through the night are solving the problem in front of them rather than writing policy.
I would put one question to the board of any network operator, utility or transport company in a region like ours. If the service failed tonight across a third of the area, in what order would it return, who decided that order and what would it cost the company? If the answer is that operations would work it out, the board has delegated one of its largest decisions to whoever is awake at two in the morning. I have been that person, and I would rather the decision had been made in daylight.