While the post was meant as a follow-up for the people helping us test, you are right that we're not clear about the product. Our homepage http://heartbyte.com gives you a sense of how the audience votes during the event with their mobile devices. Also, a link was added to post :)
We have some rough edges on our web presence to be sure. This may help a bit on the UI front: http://heartbyte.com/s/hn
As for the tech, given the requests we need to handle from our distributed queue, we do clustered stream processing in near realtime. Our broadcast customers want a line on their screen within a second or two.
Correct, just replicate between standalone Couch and Cloudant seamlessly. Replication is a powerful feature of CouchDB, and we also use it internally to synchronize partitions if they diverge. It is pivotal in how we achieve our multi-datacenter capabilities, too.
seiji, we are planning on releasing the apps separately as well (esp. rexi) and accompanied with blog posts. Also, we will transition the copies of apps that we have been using to their public projects (ibrowse, mochiweb, etc). But we wanted to get this open first. Thanks for the feedback.
It will soon. We have an 8-node cluster running in PROD right now. A Dynamo-like layer provides consistent hashing, partitioning, read/write quorum constants and the like.
I can only guess how they're going to do, business model-wise, but big data doesn't necessarily mean big companies or web properties as you suggest. Small startups can do compelling things if they make sense out of big data. It's a tedious and often impossible job now for just about everyone. Companies displaying competence will do quite well for at least the beginning phase of this movement, and have customers both big and small.
My gut says that it will become commonplace to need to handle tera or petabyte-sized data. 640K turned out to be a wee bit shy of our memory needs. Traditional RDBMS's will turn out to be a wee bit shy of our analysis needs.