Fake doors. I saw @jesskah speak about these at an Ignite talk, and she did a great job. Here's the link (http://bit.ly/lQpnOY), but essentially it's putting up a "door" that leads to nowhere in your application, and you can see how many people try to check it out. Kind of like baiting for fish.
Obviously you don't want to do this on a large scale, but check out the talk, it's only 5 minutes and she can explain it much better than I can.
Testing is what you should be doing for everything, and that doesn't always mean BEFORE implementation. Some of the best feedback you can get is through implementing and THEN validating. If it works, keep it. If it fails, lose it.
If you're worried about sinking time into an idea that you're not sure of, talk to people. Build out a single landing page with the core ideas on it and see if people bite. Test. Test. Test.
My first step is to write all my ideas in a small notepad as soon as I have them and then I get them out of my head. Then after a month I check these ideas and see if any of them look even better with this distance in time thus validating it with myself.
I never thought of that, but yeah, things do not have to be very complex or pretty to test it out.