Advertising models will result in users paying more for the advertisers' products, either by increasing volume or tolerating an increased price. If it does not, the advertisers will stop advertising and the software will go away. And every other business model (go ahead! Try me! :) can be similarly reduced to money flowing from users' pockets to yours.
Choosing an advertising model is not a good fall-back strategy for a product idea that people won't pay for: it's useful only as an optimization strategy - that is, worth pursuing if the advertisers will pay more than the users.
But if you don't think people will pay real dollars for what you're building -- I say stop building it and go to a movie.
Advertising models could be seen more as a competitive strategy of companies that have efficient infrastructure to target ads to software users, thus reducing the price of the software for the end user and squeezing out competition.
So, I think the lesson is to always be thinking of advertising models because if your competition is and you're not, you could find yourself priced out of the market.
