The mechanism for indexing worlds is in principle unbounded: it can "bind" worlds over arbitrary structural distances and across an arbitrary number of intervening operators. The expressive power of this mechanism is at odds with the discovery of several constraints that severely limit the indexing options for worlds in specific environments. Attempts to directly constrain the indexing mechanism to avoid overgeneration in these cases (e.g. by restricting its application to certain syntactic environments) often do not go beyond a restatement of the facts and therefore rarely provide much insight. In this talk, I want to highlight the virtues of an alternative approach to indexing constraints that goes back to ideas by Uli Sauerland and Orin Percus (though my particular interpretation of the approach might at points diverge from what these authors had in mind). We start with the assumption that world-indexing is in principle free and unconstrained, and that any limits on indexing emerge as a side-effect of independent properties of the intensional system. Specifically, they emerge from the interaction of two components: (a) an ontology of individuals that treats them as world-bound and (b) a mechanism that relates individuals across worlds (a mechanism for "Quinean de re"). The strategy is to reduce all world-indexing constraints to compositional pressures imposed by the ontology: individuals need to match the world of the predicates that take them as arguments. Mismatches are typically resolved by the de re mechanism, but crucially not in all cases. The hypothesis is that illicit examples of world indexing are those where a mismatch cannot be resolved. I will show how this strategy generalizes to novel environments, predicting the emergence of world-indexing constraints that have not previously been observed. Finally, I will discuss potential extensions of this strategy to the temporal domain. Time-indexing faces the same dilemma: an unbounded binding mechanism conflicts with systematic indexing constraints. Since time-indexing constraints seem to be (more or less) parallel to world indexing constraints, a unified treatment is needed. Extending the general strategy to times means adopting an ontology of individuals as (time-bound) stages. I will discuss potential implementations of this extension, point out some facts about temporal modifiers that such an approach might explain and address the promise (but also the overall difficulty) of executing this program.