FTP should be a focused conference, that concentrates on techniques for, and implementations and applications of, first order reasoning systems. This focus differentiates FTP from CADE, which covers a very large field, and attracts papers from many very diverse subdisciplines. I think FTP should be a conference where any member of the program committee can competently review any submitted paper, and where the majority of participants can understand the majority of accepted papers, at least on a superficial level, at first reading. In my opinion, a mix of stand-alone FTP conferences, IJCAR, and FTP co-located with other conferences (whether in the frame of FLoC or, as in 2003, RDP) works very well, on the one hand allowing the different deduction conferences to maintain their identities and different foci, while on the other hand supporting cross-fertilization and a single community.
I've also been nominated as a CADE Trustee, and have a similar position statement for that election.