It is a little contradiction to the statement "My philosophy is that they're right if their feature request is right only if it works for 80% of your customers.".
Apparently sometimes a small minority of customers still can make all the difference.
I think it's important to distinguish "platform compatibility" from the kinds of feature requests he's talking about, because I don't think the take home lesson is necessarily "cater to some minorities' requests".
Adding Mac support didn't make DropBox any more complicated for the end user -- there were no new corner cases or potential conflict scenarios, no new steps in the setup, no additional configuration complexity, basically no cost at all in terms of end-user mental load in order to extend their product beyond the Windows platform and reach some important taste-makers.
This is distinct from the end user confusion created by adding, say, multiple folder support or any of the other complex features and configuration options that users will clamour for. When you don't carefully curate those kinds of features, you end up sticking your users in tyranny-of-choice territory in a way that simply broadening your platform support doesn't.
Adding Mac support didn't make DropBox any more complicated for the end user.
I suspect that's not strictly true. Mac OS X and Windows file systems are subtly different in features (hard/soft links, aliases, .lnk files; files with multiple streams; various Finder annotations to files), and I suspect that once you sync to both Windows and Mac you will run into weird issues. Of course, syncing to multiple platforms is probably only something "power users" would do, and they can tolerate some teething problems.
Macs are a small minority of the total computer market. They're a huge percentage of the trailing edge, however--the tech-savvy people who are willing to try a new service like Dropbox or Syncplicity.
Leaving out Mac support in this case is like opening a coffee shop that sells $4 lattes, but can only be reached by people driving beat-up old Fords and Chevys.
Macs are a small minority of the total computer market. They're a huge percentage of the trailing edge, however--the tech-savvy people who are willing to try a new service like Dropbox or Syncplicity.
I agree with your first paragraph but would note that I think you're shooting for "Leading edge" rather than "trailing edge."
Also, imho, having a Linux client early helped quite a bit with gaining street credibility amongst hackers/nerds. I know this is the reason I started using it instead of competing services.
There's some real gems in here from the founder of Syncplicity.
For instance, not having a Mac client meant that the tech press wouldn't touch it. Then, they wouldn't write about it. That's pretty enlightening.