That's exactly what he expects them to accomplish: a lot of junk.
The thing is, junk that exists is much better at convincing top-notch coders to join than vaporware that doesn't. It shows them you're serious. More to the point, a good coder will instantly pick out a dozen ways in which your prototype sucks, and want to make them better. Because that's something virtually all top-notch software engineers I've met share: the urge to make something, once they've seen it, as good as it possibly could be. If they can't see it, they have nothing to work with. If they can, you might be able to snag them.
Everybody can't code, but everyone can make mockups. Use a tool like mockingbird (gomockingbird.com) and literally sketch out every single page/screen. When you make the mockups, you should be sure to know where every link/button goes, otherwise there's a page missing in your mockups!
Just this task alone would stop 95% of people with vague ideas. It also shows the coder that you have taken it as far as you reasonably can without learning to program.
A prototype is halfway. :-) All the way would be learning to code well enough that you could launch the product yourself.
If you can communicate your product vision in static mockups, that's great. But for most products, there's a world of difference between seeing something working - even if it's held together with string and wires and only works for the inputs you're demoing - and seeing a bunch of pictures.
The thing is, junk that exists is much better at convincing top-notch coders to join than vaporware that doesn't. It shows them you're serious. More to the point, a good coder will instantly pick out a dozen ways in which your prototype sucks, and want to make them better. Because that's something virtually all top-notch software engineers I've met share: the urge to make something, once they've seen it, as good as it possibly could be. If they can't see it, they have nothing to work with. If they can, you might be able to snag them.