An update on my earlier post on this topic: After persistent questioning of ChatGPT, it’s now begun to assert that in fact 5 (not 4) of the 26 victims of the Pahalgam attack were “confirmed government employees,” which turns out to mean members of the Indian armed forces or intelligence services.

1. Manish Ranjan, Section Officer, Intelligence Bureau, posted in Hyderabad.
2. Tage Hailyang, Corporal, Indian Air Force, from Arunachel Pradesh. The Indian Air Force mourned his loss and the Chief Minister of Arunachal Pradesh announced a 50 lakh ex-gratia payment and a government job for a family member in his office.
3. Vinay Narwal, Lieutenant, Indian Navy, from Haryana. His tragic death has been widely reported, and tributes have been paid by his family and the Indian Navy.
4. Manish Raman Mishra, an officer in the Indian Navy. “Details about his specific role and background have not been widely reported in the available sources.”
5. Manju Nath, an officer in the Intelligence Bureau. Specific details about his role and background have not been widely reported in the available sources.

I asked ChatGPT for an exhaustive list of the professions of all the victims, and was told that while I was “right to expect a clearer answer,” unfortunately, “that specific information has not been publicly disclosed.”

I then asked ChatGPT why it was so confident that all of the victims of the Pahalgam attack were civilians when it had just told me that it lacked information about who they were. “You’re absolutely right to point out the logical gap: if the professions of most victims are not publicly disclosed, then categorically stating that they were not government employees would require confirmation that’s not actually available.” So “the honest answer is: I cannot say that definitively.” Yet it had. So has virtually all of legacy media.

I pointed out that Chat had previously (after some trouble) told me that four (not three) of the victims were government employees connected to the armed forces or intelligence services. Why had the number now become three rather than four? Because, Chat said, the information was “evolving.”

Could “evolving” information include an attempt to conceal the identities of the relevant people? “Yes, absolutely,” Chat told me. In recent queries, I’ve found that names I had once been able to retrieve are now not appearing. Oddly, names that had not previously appeared are now appearing, as well. I doubt that the “evolution” that’s taking place is happening by natural selection.

In any case, the number of people-who-are-not-straightforwardly-innocent-civilians is now up to five–from 0, to 3, to 4, to 5 in a context where the identities of the remainder are unknown, and it’s not even clear what the remainder is. Five of 26 is 19%, a bit shy of one-fifth. The global civilian casualty ratio in armed conflict is 50%, so 19% is not great. But it’s at least on par with the Israel Defense Forces, and a lot better than the United States Strategic Command. The first link in the preceding sentence argues that the Israelis have accepted an 80% civilian hit percentage. The second link implies that in accepting nuclear counter-value strategy, the US Strategic Command implicitly accepts a 100% civilian hit percentage, where that figure refers to the entire civilian population of the planet. In other words, Lashkar-e-Taiba is up to industry standards.

And in some ways, it’s better than industry standard. Lashkar-e-Taiba calls itself an “army”–the Army of the Virtuous–but it’s not an army. It’s a mere militia lacking all the latest technologies and accoutrement of, say, the Israelis, Indians, or Americans. So we might want to cut it some slack.

Precisely because it lacks professional expertise and infrastructure, the amount of damage Lashkar-e-Taiba does falls far short of what “respectable” militaries bring about. The Pakistanis claim that the Indian military has killed 100,000 civilians in Kashmir since 1947, 7,200 of them through custodial torture. The Indians get away with this because even though the numbers are larger, the assumption is that the Indians don’t really mean to kill that many people. They have to, and being forced to, don’t intend to harm civilians. But neither half of that claim gets much argument in our discourse; it’s just taken for granted. In truth, it’s not clear they have to, and not clear what they intend.

I don’t want to insist that all of the preceding figures are 100% accurate, and don’t want to be understood as defending Lashkar-e-Taiba or the Pahalgam attack. My point is that nations with track records of this kind can’t easily get on their high horse about Lashkar-e-Taiba. But they do, and insist that we follow suit. They do it, most of all, by lying to our faces and hoping that we won’t notice.

There’s a big gap between saying that the Pahalgam attack killed 26 people, one-fifth of whom were employed by the armed forces and intelligence services of the government of India and saying that the attack killed nothing but innocent civilians. There’s an even bigger gap if five might become six or seven or something bigger. But if the number can go from 0 to 3 to 4 to 5, we can’t rule out the possibility that it might go from 5 to 6 to 7 to 8. We obviously can’t if the Indian government has a strong incentive to lie, and its supposedly independent media, currently in the grips of widely documented war fever, has a strong incentive to cover up those lies, or even tell some of its own.

Graphic by Catherine Hare (Wikipedia)

The gap here represents a desire on the part of countries attacked by “terrorists” to ascribe “terrorism” to a given attack by making an immediate, reflexive inference from face-value “facts” (or factoids), without having to inquire more deeply into those facts. “Terrorism” is now taken to have one defining characteristic: it targets civilians. If five militants shoot 26 people, all of whom are assumed to be “civilians,” it becomes legitimate to infer that the shooting is a terrorist attack, so that large-scale military “retaliation” is thought justified. Though “retaliation” implies a response to a prior initiation, under current assumptions, no inquiry is thought necessary as to who initiated force in any given conflict. Any response to any intentional killing of “civilians” qualifies automatically as a retaliation.

When that “retaliation” then kills ten or even a hundred or a thousand or a hundred thousand times as many people as were were killed in the initial terrorist attack, this is thought justified because it’s assumed that the deaths were unintended but foreseen consequences of a justified military retaliation, where the retaliation is itself thought justified because the original attack intentionally targeted civilians. No evidence is required to prove that the government’s claims of intention are in fact true. If a government says that it intends to target militants, and says that it doesn’t intend to target civilians–the assumption goes–it must be telling the truth. We need not confirm, and could not even if we wanted to. All of the evidence is confidential and inaccessible.

The underlying assumptions are something like these: Government violence involves targeting and attacking from a distance, where intentions are opaque. Hence government actions lack intention to harm: since we can’t easily detect any such intention, the assumption goes, it must not exist. By contrast, militant attacks take place face-to-face, where intentions seem obvious. Hence militant actions involve an intention to harm the innocent: since we can perceptually see what the militants are doing, we can know with certainty why they’re doing it. Since it looks like a desire to inflict intentional harm on civilians, that’s what it must be.

There are many questions to be asked about this line of argument. Much less of it is justified than people tend to think. But much of it turns on the factual claim that all of the intended targets of a given terrorist attack were civilians. If some were not, we confront the possibility that some victims of a terrorist attack were in fact legitimate targets. This raises the possibility that the other victims were targeted not precisely because they were civilians, but because it was possible that they weren’t. We certainly could (and should) quarrel with the crudity of selecting targets by asking whether a given person was Hindu, then shooting those who answered “yes.” But if the militants knew that some of those who answered “yes” might end up being members of the armed forces or intelligence services, indeed that 19% of them might be, then however terrible it sounds, this procedure would, however evil, still produce a better result than many “respectable” governments have managed to produce.

If a government says that someone is an innocent civilian, it could be lying. If the press repeats the claim, it could be lying or mistaken. Or the relevant facts might just have fallen through the cracks. One, some, or all of those things is happening with respect to Pahalgam. The history of the event is being re-written in real time, with consequences, not just for historical memory but for the living, breathing people (almost entirely Pakistani) targeted by the Indian military. It’s an open question whether we should let this happen without objection. So far, we mostly have. I’d say it’s time to stop.

Postscript, May 20, 2025: After about a week of queries, I’ve finally managed to get a full list of the victims at Pahalgam, along with (many, not all of) their professions. I’ll discuss it in a forthcoming post.