The chlorine substitution treats rival objectives (immigrant welfare, GDP per capita, civilizational persistence) as interchangeable illustrations of "name an objective and compute." The opener's own live fight is compassion versus fiscal responsibility, which are competing objectives. If that is the fight, "nobody computes" misreads a large political remainder as a missing calculation.

Chlorine arrives with a forced name (keep people from dying of the water). Immigration does not. Shared prosperity, safety, and stability can be declared together and still not pick a level. The leftover is whether those three close enough of the fight that the rest is engineering, or whether declaring them is the compassion-versus-fiscal fight already quoted.

There is no counting rule for a large leftover versus incomplete computation. A case where the three goals are shared and a number still is not picked can always be recoded as "they did not really share the goals" or "they still have not computed," so the leftover cannot appear.

The prevalence claim is supported by three measurements: a voter-economist gap on what-would-happen questions, partisan clustering of unrelated positions, and briefing-induced convergence in deliberative polls. None of those is the leftover after people who already have models compute under a named objective.

Caplan's voter-economist gap shows the public holds a different model. It does not show that economists, once they have computed, agree. The page's own empirical list — trade, unemployment, whether the economy is improving — remains a fight among people with models. If those surviving fights are the ones called values in legislatures, the gap does not measure the leftover.

Clustering of unrelated positions is also predicted by coherent value packages and by two-party bundling. A package that wants abortion rights, gun rights, and no wage floor points the other way from the two US parties. Clustering is therefore not a measurement of empirical error.

Deliberative-poll convergence after a shared briefing shows that sharing a model produces agreement. It does not show the model was how the policy actually works rather than a constructed briefing. The experiment does not separate those.

OGAS and Cybersyn are planners of answers: a nationwide planning network and a real-time factory coordination system. Section VI says the proposal is not that. Killings of planners do not explain why the maintained architecture in VI would be killed. They explain why a different object died.

The same page lists computations that finished: runway lengths, drug trials, the Taylor Rule, chlorine levels. "Every attempt was terminated" is then a claim about the Leibniz-to-OGAS lineage, not about computing policy. The immune-system story explains a failure mode section VI says it is not in.

Section VI says current policy computes most answers as zero. Section VIII says fertility has been tracked for decades and pension insolvency is a computed answer the system refuses to act on. Those are two diagnoses. Pricing and publishing the cost of inaction repairs the second and is the same publication the pension forecast already is. It does not repair the first.

An informational flag is a new copy of a number that already has copies. If the problem is missing models, the job is to build them. If the problem is ignored models, a new publisher of the same forecast is not a different machine. The page writes both as one fix.