It's a forecasting tool!  Yes, you can make a market for anything -- "who will be elected" is actually one very common (the most common?) use for prediction markets.  What's interesting about these markets is that they produce very accurate predictions (in the aggregate) for who will win the election.  Maybe more interesting than political markets are policy markets -- for example, will GDP go up or down if Hillary wins the election?  Will rates of violent crime in city X increase or decrease if it passes a law making panhandling illegal?

If you have a widely-used, liquid prediction market, you can simply look up the odds of...well, just about anything!  If you ask me, I think that -- if this takes off -- in 5 or 10 years, we'll look back to now and think, "Wow, we were living -- and somehow making plans! -- without knowing the odds that anything was going to happen!"  And that will seem just completely absurd to us.

All that said, prediction markets are actually a very general idea, and you're correct that you can create markets that are more like traditional gambling.  Personally, I find these gambling-esque use cases to be not terribly interesting; after all, if you want to wager on something like "where will the roulette wheel stop?" or "who's going to win the Superbowl?", there's already any number of casinos that would be happy to have your business!