I've been mulling over my options for the presidential side of the election in ten days. There are two effects to a vote, the message that it sends and the effect it has on electing the next president. People tend to discount the former and focus on the latter.
That's not a bad approach, but it ignores the message effect. A protest vote (assuming that you're not like my friend Rocky who sees his Trump vote as a protest vote) ignores the pick-the-leader effect, which is often short-sighted.
However, protest votes for president are somewhat different than protest votes for legislative seats. The Republicans having 53 senate seats rather than 52 will make a difference in what laws get passed, while the difference between getting 280 electoral votes and 296 won't make a dent in who wins the White House.
If you're in a very swingy state, your vote well might make a difference, as if your state sends candidate X from 260 to 276, your state is the fulcrum. Kevin's Ohio and Rocky's North Carolina are very hotly contested, with 538 putting the odds of a Trump win at 51% and 36% respectively as we go to press. Those are places where a few extra votes here and there might make a difference.
Trump's a longer shot in my home turf of Michigan, being back 6.2% per RCP and 6.4% per 538. 538 has Trump's odds of winning at 14%. Yes, polls are notoriously imperfect, increasing so as folks opt not to talk to strangers on the phone, making gathering representative samples of the population problematic. However, they're the information we have at our disposal beyond what we see anecdotally with friends and family, and the anecdotes may really not be representative.
However, OH and NC are on the bubble as the race currently sits. If the polls hold up, Clinton will win comfortably 324 is 538's over/under for Clinton, while 333 is RCP's no-tossup number. For Trump to have a shot, he'll have to pick up a few percent of the vote and get Clinton down under 270. If Trump is to win, current on-the-bubble states like FL, OH, and NC will slide into his camp and longer-shot states start to become on-the-bubble ones.
War-gaming the map to a Trump win would require flipping FL (0.7% back per RCP), NC(2.4%) AZ(1.5%) and NV(1.6%) and holding onto leads in OH, IA. That gets Trump to 266, assuming he holds onto Utah, where he's neck-and-neck with Evan McMullan in recent polling.
Getting that last state will be tricky, as the states get less swingy from there. New Hampshire seems to be the most likely one to get, with a 5.2% deficit for Trump per RCP. PA is 5.0% back, MI is 6.2%, WI is 6.7% and MN is 5% back, although that seems to be redder than I would expect for a traditionally left-leaning state whose niceness might argue against Trump. Michigan is thus the fourth state on that get-Trump-over-the-top list, although I don't quite trust the poll set for MN or PA, making Michigan a notch higher on the to-do list.
If Trump has a 19% chance of winning per 538 and Michigan has a 30% chance of being the swing state given that scenario, the Michigan swing vote has a 6% chance of being determining the outcome. 538 has a table giving MI a 12% chance of being the tipping-point state; I don't quite get their model, for I don't see the 18% for FL, but it's at least an alternative figure to bat around.
Using my top-of-the-head math, we Michiganders have a 6% chance of being the swing vote. A protest vote has a 100% chance of being heard. If you're voting to pick the next president, it's a binary choice, as a commenter here called it. If you are trying to send a message with your vote, you have more than two options. The Libertarians and Greens are on most ballots nationwide. In Michigan, we also have the Natural Law and US Taxpayers parties as options. That gives six ovals to choose from on the ballot. Write-ins are a seventh (Evan McMullin has filed a slate of electors and will have write-in votes counted) and abstaining from voting for president is an eighth.
So, that binary choice turns into an octal (eight-options from geek-speak) one if you're factoring in the protest vote. If you don't feel comfortable with either of those binary choices and see them as too evil to comfortably vote for, you can always send a message with your vote; given that there's only a 1-in-15 chance of it making a difference in who gets in the White House, that might be worth exploring