Stewart Butterfield has an interesting perspective on iterating a game in public in this IndieGames.com article:
We thought we'd be able to iterate our way to the promised land, and that wasn't the case
As a co-founder of Flickr I'd say he knows a thing or two about successfully iterating a product. I was surprised by his take and it makes me question my assumption that the launch-early-and-iterate method is always the right way to go.
But then again, look at the run-away success of Minecraft: the quintessential example of iterating in public. Mojang was obviously able to make that system work for them.
I guess the biggest takeaway here is that there isn't one right method and that every project is different. Finding the right balance between launching a game early vs. polishing is something that really defines good vs. mediocre games and game studios.