I'm not sure I follow. How can getting (for free) a long list of people who may be interested in the product once it launches be a bad thing?
I think the OP's main point is that instead of having a landing page you should be concentrating on product/market fit by describing the product. I think startups with these coming soon pages are already doing that, just not on their public website.
Perhaps I wasn't clear. Collecting email addresses is fine, but make sure you're really getting people who are genuinely interested in your product. My point is that people shouldn't focus on hoarding emails. The focus should be on testing certain message, certain angle of the product, whether they like the future vision, etc...
So you're saying startups should be A/B testing their coming soon landing pages. That sounds like good advice. I'd be surprised if that feature were not on the todo list of the LaunchRock guys...
I think the OP's main point is that instead of having a landing page you should be concentrating on product/market fit by describing the product. I think startups with these coming soon pages are already doing that, just not on their public website.