Do you know how meaningless a poll 4 years out from an election is?
Maybe at predicting who the winner will be, but their usually alright at determining who will run, and what the basic "tiers" will be. People like John Bolton who consistently poll at 0 and 1% are less likely to run and less likely to take off once they become candidates (see Jim Gilmore). Meanwhile, even those who polled well in 2013 (Rubio, Christie, Cruz, Paul) did go on to be at least competitive candidates before Trump steamrolled them. So, perhaps they're not useful at determining who the winner will be, so much as who the losers and non-entrants will be, and what people have an appetite for.