I have often wondered about the networking costs of the big sites like YouTube and Facebook. For the amount of bandwidth they need to provide ubiquitous access to all those large files, and the servers that store all those files, it quickly turns into some serious money and this Slate article discusses the economics.

So rather than hosting all the content themselves, and the network connections to access the content, why don’t they let the content sources host the information. All they have to do is provide a directory to where the content is. (Didn’t Knapster and some of those previous file-sharing entities do that?)

If I go out and post a video on my server on my ISP, I could send the link to it to YouTube and if someone wants to watch it, they follow the link and stream the video directly off my server. If several million people want to see it, then my ISP service will probably shut down access to my server after I hit my bandwidth cap, or they will ask me to pay more for increased bandwidth. Then it is up to me to decide what I am willing to pay for.

YouTube’s main role would be to develop the applications that I would have on my server to allow the streaming and to provide some coordination between all the videos that people post. (Presumably they would get some advertising revenue from people browsing their site to see what videos are available.) If the video is not available because the server isn’t on line, that’s not YouTube’s fault.  (Actually, YouTube could do a quick query to see if the video is available before providing the link to the browser.)

I suppose YT could come up with a way to disguise the IP address where the video is coming from to stop people from bypassing YT for future videos.

Certainly sounds better than losing half-a-billion dollars a year.