The Innovator's Dilemma is full of inspirational stories of markets discovered retrospectively. An example off the top of my head is honda motorcycles being used as off-road recreational vehicles - this market was discovered by a sales rep observing customers doing it. This could only be apparent some time after the product was created, and sold.

Another is Xerography being rejected by IBM after careful (and accurate) market assessments. The market significance only appeared after Xerography was actually in use. (from The Billions that Nobody Wanted.)

So when pundits (not cdixon) tell you it's impossible to discover a market retrospectively, they are simply wrong. But they're right that it's risky; and that it's much safer to launch with buyers waiting (of course!). I think the key take-home is that: a business needs customers. (Sounds obvious? Yet it's a perennial downfall of engineer-lead startups.) Finding customers (aka marketing) is at least as much work as creating the product in the first place. To put it in perspective: if creating the product takes a few prototypes over a few months or years, it's probably reasonable to put comparable effort into marketing: a few markets and 4 P's over a few months or years. The danger is when you exhaust yourself engineering it, and collapse at the finish line - just when the second marathon starts, of marketing it. (of course it's even better to run both marathons simultaneously... to torture the metaphor).