To make his argument, Beutler has to switch between two contradictory risk assessments:
1) First, he concedes that "some people clearly view Clinton's persistent polling lead...as a kind of liberation" to protest vote - what explains the vote for Stein or Johnson is the "widespread assumption that [Clinton] can't lose."
2) But then, he worries that votes predicated on this logic "would send a message that a coin-toss between a fit and unfit candidate is an acceptable risk for the country."
I don't see how these perceptions are compatible. If, as in 2000, the Democrat and Republican candidates were consistently polling within the margin of error of each other, one could understandably interpret a protest vote as sending a message of non-preference between the two candidates. Evidently, Beutler gets that a lot of people don't see 2016 that way: certitude about Clinton's victory is "clear" and "widespread". But in that case, why would anyone understand a narrow Clinton victory as sending a message of non-preference? Why wouldn't we just adopt Beutler's initial explanation: that narrow margins just reflect an assumption that Trump would lose, and that voters saw no risk in supporting a third party?
As so often happens with this sort of discourse gaming argument, I can't help but detect a certain double-standard at work. We, the savvy readers, are perfectly capable of looking at the situation and figuring out that most people reject Trump, and that third party numbers would be much smaller if it seemed likely that he would win. But other people may not be able to get this - they would look at narrow vote margins, and simply get "the message" of no particular preference against Trump. That's why we, the savvy, have a responsibility to try to game the system by voting for our second preference instead of our first preference - we need to puppetmaster the voters into seeing things our way, but we can't rely on them thinking about them the say way that we do.
There's a pretty simple alternative to this elaborate scheme. If the vote margins are narrow, one can simply stick to Beutler's first (correct) analysis, and insist that they were just "self-limited by a widespread assumption" that she'd win. If the right understands the outcome as proof that Trump was a viable candidate, one can simply reply that he was so unviable that everyone assumed he would lose and proceeded accordingly. This interpretation doesn't just allow people to vote for their preference on the understanding that voters are capable of understanding complex democratic outcomes; it also has the advantage of being correct.