As I explained, one of major changes we will do is getting rid of Boost.Iterator (we already removed Boost.MPL, Boost.TypeTraits, and several more).
I think the questions that should be asked is a) "what will be of the most value to users?" b) "how will this expand the potential user base?" c) It the above are positive, the next question would be "how much effort would be required. So I generally answer the above by required the oldest version of C++ which supports the features that the library needs. This maximizes the breadth of the potential audience while minimizing the effort required. By the same token, if one isn't going to add features that users are requesting, there isn't much point in investing more effort in the library. So if it were me, I'd say - move it to C++ 11 or perhaps 14 to simplify the code, and maintenance, and leave moving to C++17 or higher when there will be some positive return on the effort. Robert Ramey