Working on discussion boards. Like tumblr, for forums.
We already run some forums based on vBulletin, XenForo, Vanilla, etc... and have ~300k monthly unique visitors, ~40k registered members, ~15k active (each month) users, and mailing lists of ~25k.
Forums being one of the oldest things on the internet (usenet) and one of the most resilient.
But we're not making this forum or that forum... those are muffins and we're building a muffin making machine. A hosted platform to build great forums.
The whole business came out of user demand. Frustration at the stagnation of forum software and what users have to do, and how they have to compromise to make a community work on such software.
We (the users and I) started to ask "What if we..." instead of "How do we work around this limitation?". Where we've ended up isn't technically impossible... it's just a lot of work. But they're really excited by it, as am I, and the first mockups have all been built with a lot of feedback from them and their response has largely been:
That mock-up thing is rad as fuck! Hurry up and get some bank details or something up so I can blast some patron money in your direction!
The fundraising is to help support the building of it (stipends for people to work on it, rent for a small office, etc), and the users themselves are asking for hack weekends so that they can come in, and pool 20 people's effort into pushing it along.
But still, the YC application was borne out of hope that we could move even faster. We make a profit already from affiliate fees, and there's a lot of money in hobbies... we just want to get the first real version out of the door and into the hands of the users.
* The LaunchRock page is just that... because we're already working with users most of what we've focused on has been in-person stuff from our office near London Waterloo. We haven't yet felt a pressing need to spend time on a pretty landing page.
If you compare it to tumblr, I guess it has to do with the way a user can follow multiple ones at the same time? As such, would it be a little like "Reddit for forum?"
That simple thing means that things like classifieds are currently threads. Events, are threads. Articles, are threads. Reviews, are threads.
It drives them mad. They can't find anything, cannot do anything with the information.
They can't even compare classifieds, or find events and map them.
They're forced to then smash up their community's content and litter it around the web. Some on eBay, some on Facebook or Meetup.com. Which is a real problem for them: discoverability/search is poor, data ownership is a mess, permissions are confused, etc.
They want a basic form of all of this stuff within the forum that they're using. They want to bring it back home.
None of the hosted forum software solves this. And because of things like the built-in search, it's really hard to hack a vBulletin (or something else) instance into having this. It should've been there from day one.
But if you want programmer insight and are ever in London, UK... I run open office hours all the time. Just drop me an email and pop in.
File store remains an open question we've deferred on by not yet implementing file upload ;) The master is likely to be Riak and CloudFront as the CDN for it.
API is written in Go.
Front-end is Python and Django, though we have a number of volunteers who are asking for hackathons to help implement the front. Some of these know Ruby and Rails and not Django... so it may be that we actually offer two flavours of front-end... the official one which we will update in Django and an alternative one in Rails.
We've had people already ask things like, "Why not Mongo and Node?".
But we just don't go there. Such questions can be asked until the cows come home, and we prefer to just get things done today than engage in non-pragmatic debates.
How do you store extended data (hstore?) that you talk about in the upper comments?
We've not answered of those problems in the ways you've hinted at though. Maybe one day there will be a High Scalability article that explains it... it takes more space than a HN comment would provide.
A good plugin architecture, and ecosystem of plugin writers, would probably be key, at least down the road. My recent freelance projects have been in Wordpress and Magento and IPBoard ... all powerful frameworks with a rich amount of plugins available. Makes for tedious dev work though -- "here's a plugin that does 90% of what the client wants, let's just randomly change PHP code I don't understand until I've implemented the other 90%". If you can fix that problem, that'd be something special.
They do want the ability to extend the data structure and query things in their own way, but mostly they want the ability to pick up a forum and deeply integrate it into their projects, or to take their project and deeply integrate it into the forum.
What I've got are users, designers and agencies telling me:
1) Give me an API that does everything so that we can make our own mobile apps for it.
2) Give me full and unrestricted access to the entire front-end so that we can do whatever we want with it.
3) Give me some way in the future to extend or associate data in your API.
Which ultimately has shaped the design as: You create a forum/site on the platform, and we'll give you a dedicated API for your forum. We'll also give you a simply customisable forum front-end in Django. You can take the Django front-end and host it yourself if you want, and butcher it to your heart's content... making it whatever you want it to be, and we continue to solve the storage, searching, problems via the API.
In essence, there's two parts to it. One part extensible (the API), and the other part 100% within your control (the front end web app).
But if you're not technically inclined and just wanted a forum already, then you got that in 30 seconds and it was really simple to change the look and feel so that your community has it's own strong identity.
We actually see a far bigger market in the space of the non-technical wanting forums. I've setup or helped so many forums over the years, and honestly... most would just love to get something running today, not have to learn how to be a sysadmin, and to not have to worry about security issues and updating the forum software.