I found Alex de Waalโs interview with Noam Dworman increasingly frustrating for one basic reason: what, exactly, could falsify his claim?
If large numbers of people die, that demonstrates famine. Fair enough. But what if they do not?
Then we are told that famine deaths need not be recorded as starvation. People weakened by malnutrition often die of diarrhea, respiratory infections, other diseases, or the collapse of chronic medical care. That is perfectly true. Malnutrition substantially increases vulnerability to infection, especially among children.
But this cannot mean that starvation becomes empirically invisible. In severe historical famines, many people die through diseases potentiated by malnutrition, but one also sees severe wasting and deaths directly associated with hunger and malnutrition. And there is a demographic signature. Young children are biologically particularly vulnerable. A recent review of historical famines notes that, in most famines of the nineteenth and twentieth centuries, young children and older people were the main victims in absolute numbers. Even the Dutch Hunger Winter, which de Waal invokes as a complication, saw extraordinarily high infant mortality and a very large increase in mortality among children aged one to four.
This makes Michael Spagatโs Gaza Mortality Survey especially interesting. I do not regard that survey as beyond criticism. Quite the opposite. It has serious methodological questions of its own. Among 9,729 people enumerated in its households, 52 were reported as imprisoned. After weighting, that produced an estimate of about 9,400 imprisoned Gazans, while BโTselem reported 2,219 Gazans held by Israel at the end of December 2024. That is more than four times the known figure, a rather striking warning about representativeness or interpretation of the questionnaire.
But here is the problem for the famine argument. If we nevertheless take Spagatโs survey seriously as evidence about mortality, its demographic pattern is extremely awkward. Children were much less affected than adults, much much less affected.
De Waal can point to historical exceptions and say that children do not always show the largest proportional mortality increase. That is true. But exceptions cannot be used to make the expected demographic pattern irrelevant. If a survey produces almost no indirect child mortality, that is evidence that requires explanation. You cannot invoke the survey when it supports excess mortality and then effectively disregard its age distribution when that distribution cuts against the starvation hypothesis.
There is another potentially independent test: refeeding. De Waal explicitly warned that thousands of severely malnourished children might require hospital treatment once food became more available, because ordinary feeding could itself be dangerous. The IPC Famine Review Committee likewise warned that refeeding syndrome might become a risk for significant portions of the population.
Yet when food availability subsequently increased, where was the corresponding population wide wave of refeeding syndrome and deaths? This does not by itself prove that serious malnutrition did not exist. There may also have been isolated cases. But if severe starvation existed on the enormous scale being described, the physiological consequences of restoring food should themselves provide another observable test. The apparent absence of the predicted phenomenon on anything like the anticipated scale should cause us to update our assessment. Instead, it seems simply to disappear from the argument.
And then the definitions themselves start moving. Later in the interview (min 55), de Waal appears willing to treat the lived or psychological experience of famine, fear of hunger, desperation over food, competition for food, social breakdown, as "famine", even without mortality.
This is where the method stops looking scientific. Scientific definitions are supposed to reduce ambiguity, draw boundaries, and limit the analystโs freedom to reinterpret inconvenient evidence. Here the opposite happens. Food insecurity, hunger, starvation, famine, famine related mortality, and even the subjective experience of famine are progressively collapsed into one another until the categories cease to constrain the argument at all. Instead of the evidence testing the definition, the definition expands to absorb whatever evidence happens to exist. That is not conceptual precision. It is methodological elasticity, and it gives the interpreter almost unlimited discretion to protect the original conclusion from refutation.
The same problem applies to prediction.
If mass starvation deaths occur, the warning was correct: Israel starved Gaza.
If the deaths are officially attributed to infection or other disease, the warning was still correct: malnutrition caused or aggravated those deaths.
If the excess deaths cannot actually be demonstrated, then the data are too poor to know.
If children, normally among the groups most vulnerable to severe nutritional stress, are barely represented among the nonviolent deaths in the available household survey (problematic as it is), then historical exceptions show that this need not matter, contrary to what de Waal himself said during the war.
If the expected wave of refeeding complications does not appear after food availability improves, that apparently does not matter either.
If mass mortality never materializes at all, then the original warning was still correct because de Waal, the UN, and others raised the alarm and catastrophe was averted.
And if none of these things can be demonstrated, then even the psychological experience of famine can apparently be brought into the definition.
So I genuinely want to know: what observation could have made de Waal say, โMy assessment was wrongโ?
Not โthe data are incomplete.โ Not โthe deaths were really caused by other diseases.โ Not โchildren are not always the most vulnerable.โ Not โour warnings prevented the deaths.โ Not โpeople still experienced famine psychologically.โ
What evidence, actually observed in the real world, could count against the hypothesis?
Because that is the crucial scientific question. A scientific claim has to risk being wrong. Evidence is meaningful only if some possible evidence could force you to revise the conclusion.
If deaths count as confirmation, the absence of deaths counts as successful prevention, deaths from unrelated recorded causes can be reclassified as starvation related, missing deaths are attributed to incomplete data, an unexpected demographic pattern can be explained away by historical exceptions, the absence of a predicted physiological consequence does not matter, and subjective experience can substitute when the measurable outcomes fail to appear, then the evidence is no longer testing the hypothesis. The hypothesis is controlling the interpretation of the evidence.
At that point, it is not merely difficult to falsify. It has been constructed so that it cannot lose. And that is the opposite of an empirical scientific argument. It increasingly seems that de Waal is prepared to make almost every possible rhetorical move, treating Gaza as a unique case whenever necessary, rather than seriously entertaining the one possibility his framework never seems to allow: that he and his favorite NGOs and UN agencies might simply have been wrong.