Proposed resolution:

This wording is relative to N4582.

1. Insert the following paragraphs after [alg.copy]/4:

   template<class ExecutionPolicy, class InputIterator, class OutputIterator>
     OutputIterator copy(ExecutionPolicy&& policy, InputIterator first, InputIterator last,
                         OutputIterator result);
   

   -?- Requires: The ranges [first, last) and [result, result + (last - first)) shall not overlap.

   -?- Effects: Copies elements in the range [first, last) into the range [result, result + (last - first)). For each non-negative integer n < (last - first), performs *(result + n) = *(first + n).

   -?- Returns: result + (last - first).

   -?- Complexity: Exactly last - first assignments.

2. Insert the following paragraphs after [alg.move]/4:

   template<class ExecutionPolicy, class InputIterator, class OutputIterator>
     OutputIterator move(ExecutionPolicy&& policy, InputIterator first, InputIterator last,
                         OutputIterator result);
   

   -?- Requires: The ranges [first, last) and [result, result + (last - first)) shall not overlap.

   -?- Effects: Moves elements in the range [first, last) into the range [result, result + (last - first)). For each non-negative integer n < (last - first), performs *(result + n) = std::move(*(first + n)).

   -?- Returns: result + (last - first).

   -?- Complexity: Exactly last - first assignments.

[ 2016-05 Issues Telecon ]

Marshall noted that all three versions of copy have subtly different wording, and suggested that they should not.

[algorithms.parallel.overloads]/2 says that "Unless otherwise specified, the semantics of ExecutionPolicy algorithm overloads are identical to their overloads without."

There's no "otherwise specified" for the ExecutionPolicy overloads for std::copy and std::move, so the requirement that they "start[] from first and proceed[] to last" in the original algorithm's description would seem to apply, which defeats the whole point of adding a parallel overload.