I'm of the opinion that the "launch" is a highly overrated concept. Unless you have lots of existing traffic or a press embargo, your launch will consist of turning on the product, then watching as nothing happens. Later on, if you're lucky enough to get press somewhere, you'll get a spike of traffic that will be gone the next day.
As an experiment to this end we developed Omniref [1] entirely in public, just to see what would happen. We never hid the thing, even when it was horribly broken (we even had organic search traffic before we were ready!) It made no difference. What works is getting users, one at a time.
The bottom line is that product/market fit trumps everything. Better to just put the product out there and try to get one daily user as soon as possible than to worry about "launching". It's not as gratifying to the ego, but you learn more, and that's the whole game.