I had a chat with Frank Smith, Production Editor at Contentinople and he wrote up an assessment of our project.

He's absolutely spot on with regard to the challenges we face in establishing copyright precedent. I'm hoping that the existing structure of syndication to local affiliates will provide much of the necessary business terms. But it is possible that my next six months will be spent working out individual deals with content publishers to whom we bring an audience.

As to the proof in our pudding, we gave put it to a taste test at Techcrunch 40 last week and were pleased to have 100+ people give us an average rating of 4.5. The last week of development is all about making boxing it up for the general public. ffwd - just add favorites and friends for an instant video treat

ffwd (pronounced "fast forward") is a platform to discover video content based on your general tastes and choose what to watch according to your mood. Use the ffwd recommendation network to make personalized, ranked, and filtered discoveries from all the choices coming available — broadcast TV hits to the most obscure user-generated clips — and surf through them smartly.

ffwd provides a cross-source viewing experience and profile which remembers your tastes. ffwd transforms your social network into a social viewing experience, identifying content you would collectively enjoy. Lastly, ffwd guides you through program choices by calibrating your surfing stream according to your response to previous options.

Try ffwd.com as your video viewing start page and spend less time deciding what to watch and more time enjoying your discoveries. Please, come by to learn more and get your invite to our launch party.

integrated start page

member profile

discover page

ffwd player

recommendation form

recommend video message

Come check out a preview demo at TechCrunch 40 today (Monday 17th).

It's almost a week until we're supposed to debut our little baby to the world, and folks are starting to get nervous. But we've been churning out code for more than two months now, going through an architecture redesign both on the software and system level. One of the things we wanted to do this round was to address issues of scalability. So for the next several posts, I'll try to discuss some of the things we've tried to accomplish in this area.

One of the first things we took a look at was offloading all our images to AmazonS3, but after doing some benchmarks on it, it seemed to be pretty slow at serving up images. Compared to having them hosted locally, the load times were about doubled. Ouch! We really don't have a lot of images in our pages, because a lot of the design is pure CSS, but in certain cases we need to load a bunch of images at the same time, as is the case for our shows and channels picker. So for now we're going to stick with our own local hosts for these, and start evaluating migrating over to a CDN like MirrorImage or Limelight in a few weeks.

However, AmazonS3 is not something to just be disregarded completely, and for what it is, it works quite nicely as a cheap, reliable storage system. Our hosting company doesn't currently support mounting shared SAN storage, so until then, we'll be using S3 to store all of our user-uploaded images. This seemed to be a more reliable alternative than using some kind of rsync script on the server side to synchronize uploads between our load-balance web servers. All we have to worry about is just making sure the URLs are pointing to the right location on Amazon's servers, and we're good to go.

I spent some time integrating S3 into our PHP uploader built in CakePHP, and for all intensive purposes, it was pretty simple, thanks to another open source library out there called PHP-AWS, which provides PHP5 classes for connecting to all of Amazon's web services. Everything is well documented and easy to integrate, and we were able to get something up and running within a day. I'll try to follow up on this with another post in a few weeks, to let you know how everything turns out.