The sentiment expressed by Gallego (2016) with regard to the status of intermediate copies is well-entrenched in the minimalist thinking about chains, so well and deeply that it comes even to be used as a hallmark of minimalism in an argument against chains, as in Stroik and Putnam (2013):
... an A-chain will have only two properties and only two links: a foot-link that has its CAT feature (and its thematic coefficient) interpreted by the C-I performance system and a head-link that has its Case feature interpreted by the SM performance system. The fact that each performance system sees/interprets only a single link suggests that the chain itself has no visibility to the performance systems, which further suggests that chains, too, should be dispensed with. (Stroik and Putnam 2013: 137)
A-chains are the extreme case, obviously, A-bar chains being in a better position to retain some of chain-like properties even in frameworks which otherwise abhor chains, expressing a dependency which may be understood along the lines of variable binding in logical calculi. Even here, though, intermediate chain links are suspect (modulo reconstruction sites; see e. g. Franks (2014), who rejects internal merge for wh-movement in favour of ‘feature-driven multiattachment,’ or Dikken (2009), who argues for elimination of intermediate movement sites at the edges of C-phases). Larson and Hornstein (2012), discussing derivational accounts of locality constraints on movement, remark that the idleness of intermediate links at interfaces makes it impossible to state successive cyclicity requirements as following from demands of interpretive components, and note briefly that ‘The alternative is to assume that chains in their entirety are CI objects, not just the operators and variables. This is a coherent option but one, we believe, that is not generally assumed. It is certainly incompatible with certain chain uniformity assumptions commonly made in the syntactic literature which require the elimination of “traces" in intermediate CP positions...’ (Larson and Hornstein 2012: 5 n. 10). If the option is coherent, it deserves at least some exploration, its lack of popularity notwithstanding. Some interpretive flesh should be imposed on the bones of the notion of ‘chains in their entirety,’ and the fact that they consist of several occurrences should be similarly accounted for in interpretation- related terms, clearly, but the possibility of having chains ‘in their entirety’ fulfill interpretive roles without being dismembered and mutilated seems worth closer scrutiny. It does not, on the other hand, mean reversing the directionality of explanation with regard to syntactic mechanisms and properties; chains arise in virtue of the operation merge occurring in the internal mode, and their consisting of occurrences at the edges of successive phases is a result of syntactic requirements. Thus, locality constraints do not emerge because the C-I component needs them—they arise because of the way that the operation merge and phasal derivations work, thus falling under the category of those properties which fall out in some natural way from the computational process' and not those ‘imposed by legibility conditions’ (Chomsky 2000a: 113). So do chains. Yet their being born that way does not mean that they should be treated as mere side effects of syntactic computation which do not enter interpretive processes, by-products of a purely syntax-internal relevance. If an analysis of the impact of the presence of chains on the C-I component along the lines sketched above proves viable, the interpretive component is partly turned into a snowflake, with regard to the interpretation of chains at least. Their presence, forced by labeling considerations together with requirements of minimal computation, makes the interpretive side of the derivation work in tandem, thereby also satisfying MC; furthermore, far from being required to undergo some sort of impoverishment, they are interpreted with respect to all their links. A procedure employing counterpart relations to link interpretively links of chains gives rise to objects—not objects in the diluted sense of ‘members of a domain,’ but Gegenstande in a wide sense of the term. Modifications and refinements of the basic interpretive machinery may then enter the scene for various purposes, but chains remain a crucial syntactic mechanism which provides the basis for ‘rich perspectives for interpreting and referring to the mind-independent world involving Gestalt properties, cause-and- effect, “sympathy of parts,” concerns directed to a “common end,”psychic continuity, and other such mentally-imposed properties’ (Chomsky 2011: 276), a uniquely human perspective on the world.