Not once has any of these people succeeded with this approach. On the other hand, Joel at Bufferapp who I've been watching since the early days was very public. Got out there to find out if people wanted the product pre-MVP and when there was an MVP he let people use it. Today they're based in SF, have strong revenue, a team, venture funding and hundreds of thousands of users.
The people with "stealth companies" that succeed are serial founders with money behind them. Most of the time they don't keep it under-wraps, they just don't share it online. If you ask them what they're doing they'll tell you and they're constantly getting out there talking to partners, investors, etc. Many of these companies fail such as Color.
Unless you're starting a new national airline that could be stopped if a competitor found out what you are doing and went to congress or partners to block you from entering the market.
I recommend talking to competitors if you're put in a situation with one. This is something I've done and they've been very helpful, on one occasion one helped me out without intending to by talking about why one feature probably wouldn't work without a tweak because they found out it doesn't work without X.
In the end it's about execution. Get as much as you can out of the door as you can and learn as much about the market as you can.
I actually wrote an article about this a new months ago: http://brack.in/post/32461803751/stealth-is-bullshit