In his article 'Consonant cluster neutralisation and targeted constraints', Wilson (2001) proposes a far-reaching revision of Optimality Theory to accommodate targeted constraints, which compare candidates differing only in certain specific ways. Targeted constraints, it is argued, can explain why cluster-simplification processes affect the first member of a cluster but never the more marked member of a cluster. In this remark, I show that this argument encounters difficulties once it has been embedded in a fuller picture of constraint interaction. Some general properties of the targeted-constraints model are also discussed.