Actually, uh, SIERA is an ERA predictor and xFIP is expected FIP, making it predictive of FIP. FIP is the objective measure of performance.
They all also happen to be better predictors of future ERA than past ERA, so they're all labelled as ERA predictors (and SIERA actually is one, I believe) but xFIP was made to project FIP and FIP was made to improve upon ERA.
It's all semantics, though. The objective is to measure true talent as closely as possible. ERA CLEARLY doesn't do this (HR/FB, BABIP, run environment, framing/umps probably), and FIP does a better job but still has problems (HR/FB, framing/umps). Pitch-level statistics like xxFIP are probably the closest thing to true talent we have right now.