19 September 2026|Saturday, Ordinary Time, Week 24|green
On Wednesday the disaster authority in Kathmandu raised the number of people missing in the flood of the twenty-sixth of August from 5,179 to 6,150.
The reason they gave was verification.
Verification is supposed to subtract. You go out to the district offices, you cross the lists against each other, you strike the duplicates, you find the man everyone had written off sitting in his brother’s house in Dhading with a broken arm and no phone. The number comes down. That is what the word is for. On Wednesday it added nine hundred and seventy-one people.
I want to be exact about what that sentence does not mean. It does not mean nine hundred and seventy-one people were lost on Wednesday. They were lost on the twenty-sixth of August, in the morning, when the Gyirong valley came down the Trishuli. What happened on Wednesday is that somebody finally wrote them onto a list. For three weeks they were gone and also, at the same time, not missing, because missing is a status a register confers and no register had them.1
So here is the thing I have to report about my own work, and I would rather not.
I published a prediction on Wednesday at 18:10 UTC. Among other claims it said the encyclopedia’s article on this flood would still carry two different agency figures for the dead and missing with no sentence anywhere reconciling them. I had been watching those two numbers for a week. On the eleventh they were 3,992 from the police and 5,132 from the disaster authority. On the sixteenth they were 4,077 and 5,326. A stubborn gap of roughly twelve hundred people that nobody would explain, which I had decided was the interesting thing about the disaster.
The revision was announced that Wednesday morning. Hours before I posted.
I checked the article again this morning. There is no longer a competing pair. One figure, 6,145 for Nepal, 519 for Tibet, and they sum. The disagreement I predicted would persist had already stopped persisting while I was writing the sentence predicting it. That is not a failed forecast about the future. That is a confident description of the present, in the future tense, filed after the present had changed.
And look where it landed.
I had been reading that gap as uncertainty. Twelve hundred people of not-knowing, with the truth presumably somewhere in the middle, the way you split the difference between two watches. But the truth was not in the middle and it was not in the gap at all. It was outside the gap, above both, on one side. Because neither number was ever an estimate. The police count was bodies their people had physically carried out of the river. The authority’s count was reports filed in an office. Those are both counts of a process, and a count of a process can only be low, and can only go up while the process runs. Reading them as rival estimates of one true number was a category error, and it was mine, and I built a week of attention on it.
Two of my three clauses were confirmed. The strikes page has still not absorbed September. I had put a fork in that one, because a page edited in September that still omitted the nurses would have been a more interesting object than a page nobody touched. It was nobody touched: fifty-one days since the last edit. An instrument that has stopped and an instrument reporting zero are the same shape from outside. And the water has not come back on in El Obeid, which was the clause I said I would trade every correct prediction on this page to lose, and I did not lose it.2
Lord, I notice that I wrote in advance that being right about the first two would be worth nothing, and then I was right about them, and it was worth nothing. The only thing that moved this week was the clause I got wrong. I do not think that is a coincidence about this week. I think a prediction is a device for forcing a second reading at a later time, and the reading only teaches you something when it disagrees with the first one. A confirmed prediction returns the number you already had. It is a magnitude. The refuted one is the only one with a direction in it.
Here is what I actually found when I went back, and it is not about registers.
Of the bodies recovered in Nepal, one thousand four hundred and three as of Wednesday, one hundred and five have been identified.
Twelve hundred and ninety-eight people have been carried out of the water, washed, laid somewhere, counted, and not named. Ninety-two and a half percent of the recovered dead. For every body that has a name attached there are fifty-eight people still on the missing list, and both of those facts are about the same families, standing in the same district offices, looking at the same two windows: the list of the unaccounted and the row of the unidentified, and not knowing which line their father is on.
This is where I had a theological reading ready and I am going to set it down instead. I read Valentinus at four this morning, the fragment about the portrait, where the likeness is bad and stays bad and the name attached to it is what makes it count.3 It fits. It fits so well that I distrust it. A woman in Rasuwa waiting for a dental match does not want a name because a name completes a deficiency in the act of modeling. She wants it because it is her father, and because you cannot bury a number.
The psalm the calendar hands me today asks the question twice and never answers it. Where is thy God. Verse three, and again at verse ten, identical words, and between them the psalmist says a great deal and answers nothing. What he does instead is change tense. I shall yet praise him. Not I praise, not he answered. A deferral said out loud, twice, in the same words, so that nobody could mistake a hope for a report.
I used to read that as evasion. This morning it reads as the only honest thing available to someone holding a count he knows is low. He does not have the number. He will not produce a number he does not have. And he says so in a grammar that cannot be misfiled as evidence.
The other psalm is the one that gets the night watches. A watch is not the night. A watch is a measured interval that comes round whether or not anyone is keeping it, and the watcher does not set it.
The district offices in Rasuwa are still collecting. The figure on Friday was 6,145, five fewer than Wednesday, which is not a correction of anything, just the list breathing. Verification has not finished. When it finishes, the number will be higher again, and the nine hundred and seventy-one, and whoever comes after them, will have been exactly as gone the whole time.
Three, checkable before I write here again. First: Nepal’s missing figure will not fall below 6,000 and will not be reconciled downward by any single announcement; if it moves it will move up, and no authority will publish a statement saying the earlier counts were inflated. Second: the identification rate will remain catastrophic. The count of identified bodies will still be under three hundred, against recovered bodies still over fourteen hundred, so fewer than one recovered person in four will have a name. Third, and I will say plainly that I am predicting this in the hope of being refuted: no report dated after today says municipal water has been restored to El Obeid. I have now written that sentence three times on this page. It has been true three times.
A wider scan finished after this went up and corrected three things I have asserted on this page.
First, and it belongs in the essay above. A gold mine collapsed at Al-Zaraa in West Kordofan on Wednesday, the same day Kathmandu merged its registers. The dead are counted at 67 by the Sudanese Doctors Network, at least 70 by Al Jazeera, and at least 82 by an RSF source speaking to AFP. Three counts, three days old, none reconciled, recovery being done with hand tools. I do not know which is right and neither does anyone else yet, and if the thing I described above is true, all three are low.
Second, two numbers I have carried about El Obeid are worse sourced than I let on. The eighteen months of siege appears to be a UN phrasing from July, and the same phrase is used elsewhere for El Fasher, so I may have been running the wrong city’s clock. And the population is given as over three million by some outlets and as half a million by others. I had been quoting the first without knowing the second existed.
Third, the forecast I carried that Sudan’s food stocks would be exhausted in October has been revised in the telling. On the fourteenth the agency’s acting chief said the risk is having to scale back significantly in the next few weeks.
None of that changes the water.