Stealth-mode or Open ?

The increasing number of start-up launching services on the web and especially within the 2.0 trend is bringing a new issue: should web services be developed in stealth mode or not? Thanks to both Mark Fletcher and Paul Kedrosky for their disagreement which i summarized herebelow.
Stealth mode is when a company is operating in secret for some length of time before launching their product or service. Is success of a web service inversely proportional to the secrecy that surrounded its development ? There are both supporters and opponents to this approach. Here are some good reasons for not going Stealth but rather go fast and be open as early as possible under the assumption that most web services should not take more than 3 to 5 months from conception to launch:
First mover advantage gives the opportunity to define the new space and have competition compared to you.
There is no such thing as a unique idea. It is indeed highly probable that someone else has already thought about your wonderful web service, and is actually way ahead of you.
It forces you to focus on the key features of the service thus shortening development cycles.
Being perfect at launch is not only impossible and unnecessary but probably detrimental. The rule should be ship early, ship often since the service can be continuously updated and fixed.
The sooner something gets out there, the sooner it will start getting feedback from users. In a way, users could be seen as the most valuable asset of a web service if not the only one. They act as advertizers, providers of new feature ideas, quality testers and most of all help find out extremely quickly if the service is actually useful or not. But the price to pay to develop a strong base of passionate users is the high level of customer support responsiveness required.
Final point, launching quickly probably doesn't require a lot of money and ressources, thus avoiding an heavy and disturbing fundraising process happening too early in the company's life cycle.
But other could argue that Stealth mode is the way to go because :
Even though first mover advantage is important, there are many advantages to be follower when it requires heavy infrastructure investment. As an example Google was not the first search tool in the market.
Entrepreneurs stay in Stealth mode sometimes not so much because they think their idea is unique but rather because they think there is no need to prematurely advertize.
Launching early with the key functionality of the site carries the risk of not reaching an acceptable quality level.
Shipping early and often is great but doesn't necessarely means ship wide. Opening can be progressive with a viral approach from a predetermined and limited list of early testers. Thus avoiding to upset many people not ready to handle bugs and a too minimal level of features.
Feedback from users can be obtained without exiting strealth prematurely.

I leave it up to you to favor or not the Stealth mode vs Open approach from a start-up standpoint. As you may have guessed, i would rather recommend to go Open mode but with the difference of a progressive launch such as Joost when it come to opening it to testers and users.