In 2012, Gary Johnson won 0.99% and Jill Stein 0.36%.
In 1980, Libertarian Party candidates Ed Clark and David Koch received 1.07%. Unlike Johnson and Weld, Clark and Koch were actually libertarians.
I can't see how Johnson and Stein are going to win anywhere near a combined 8.5% of the national vote. My guess is that "Johnson" is often used as a socially acceptable or polite way of saying "Trump" when speaking with a young coed pollster.
My own inexpert SWAG (scientific wild *** guess) is that Johnson "might" double Clark's performance (rounding up to 2.2%), and that of the remainder, 2/3 will vote for Trump and remaining 1/3 will not vote for president or will stay home. Since there is absolutely nothing that is libertarian about Hillary Clinton, I can't see her picking up many votes from this group. Another guess is that Stein triples her performance from 2012 to 1%. Of the remainder, 1/2 might vote for Clinton, 1/4 might have heard her comment about Clinton starting WW III and vote for Trump and 1/4 stay home.
This poll (along with the LA Times poll) is encouraging, and I'm starting to believe that Trump might actually be able to pull it off.
Short version: I don't think third-party voters could have much effect on the election, even if they voted for Trump or Clinton -- because most of them are in deep red or blue states.
On Wednesday, I'll look through the results and see if I'm right.