Why Cloud TV

By the early 2000’s, Cable TV was long overdue for an a rich graphical interface. But cable companies couldn’t convince its subscribers to buy a new setup for $500. It would cost the top 3 cable companies about $ 6 Billion each o provide those set-tops.

Update: the business case for Cloud TV no longer exists. Read more in this post of mine

The Idea

Render the user interface in the cloud and send a MPEG Video stream to EXISTING cable boxes, saving money and deploying rapidly

The Challenge

The number of servers needed to render 20 million video streams by brute-force encoding of the user interface was very high – about 10,000 servers for each cable company just for the menu

Caching scaled videos and encoded stills, reduced CPU by about 10%. The MPEG-2 format was too rigid, requiring a lot of bit-shifting to stitch the videos and stills together to make an user-interface video

The Breakthrough idea came to me when I was spackling a wall to prepare it for painting

If we could eliminate the bit-shifting, we could cut the CPU needed by 75%. We do this using spackle codes as shown below. The logic of calculating the spackle codes is quite sophisticated, but all possible spackle codes can be computed offline and cached, so you can combine any two MPEG-2 streams with a simple lookup