That sounds good on paper but in reality writing code for new product is based on faith. You don't know what will work and figuring that is both the most risky part of writing software as well as most lucrative (when solved).

And the only way to find out if something will sell is to build and try to sell. The method you advocate (get sales before you have a product) just doesn't work.

"A list of verifiable people that you don't know that want to pay you money"

How do you propose to "verify", or even find, people that you don't know?

"Someone that can write a check at BigCo."

Would you write a check for something that doesn't yet exist? How do you know, from just a description of the product, whether it's going to be moral equivalent of Vista (a massive flop by Windows standards) or 7 (a massive success)?

You don't, which is why no one pays for ideas or description of future products.

"Someone that is beta testing your product live."

That does require that code has already been written, doesn't it?