#2 is not true. Stories posted automatically via the API won't show up in friend's newsfeeds. But apps can popup a "feed form" which allows users to preview the content the app wants to post to their wall (this can include an image or embedded swf). If the users posts it then that content will show up in all their friend's newsfeeds. I think this format is actually much better for apps.
I'm not sure how common it is for apps to be making that much, but I know one of the top apps is pulling in $5,000+ a day, and it's by an individual developer.
I was making around $150 a day from an app with 100,000 monthly active (90% were CPA ads with the rest being socialmedia), and there are a fair amount with over 5 million monthly.
I think the idea of an embeddable widget that flips through all your photos (like the videos you describe that inspired the site) would be really cool. I think if a widget like this is done right you could get some good viral growth from MySpace.
I'd think about capitalizing as much as possible on the timelapse concept and stray away from being just another photo sharing site. Right now, users profiles are just a feed of images, which doesn't seem that compelling..
Just a guess, but they may feel that having the beta label on a downloadable app tends to turn people off, whereas people don't mind beta web apps (its the norm and they won't screw up your system).
I think an option to filter out (or just see) self-referential posts would be great. I actually like AskHN posts a lot, and an option to just see those posts would probably encourage people to post more.
While you can probably improve a karma system with decay or a more complicated algorithm, there's something to be said for having a system simple that your users understand. I'd make a wild guess that karma wouldn't be as interesting to users if they can't figure out how their actions directly affect their score.
I interviewed a bunch of reddit users for my senior thesis. One thing I found was that the karma system really encouraged participation (not too surprising but worth considering if your building a social news site).
Check out http://www.thesixtyone.com. It's a great example of how you can instill a social media site with a really fun, game-like quality. Not sure if you want to go in that direction, but still worth checking out.
As far as spamming and moderation, I wouldn't worry about it too much if your just getting started. I probably wouldn't even do email validation to begin with since you want the barrier to be low when your trying to attract your first users. You can always add that stuff later when the need arises.
In the interview you mentioned that your monthly burn rate was “in the low four figures". Does that include salaries for your team? Just curious what your cost breakdown is, how much of that goes towards servers, etc (if you don't mind revealing that).
I think it would make sense to allow people to use your app (start building a list of artists) before registering. Just store their list to a cookie and show a little notification that tells them to enter their email and a password to save their artist list and start receiving notifications.
Maybe that person just got unlucky and had a bad app tester.