One of the key features of the current explosion in consumer video upload sites is that it is largely made up of short form programming. In general clips are three minutes or so long.
The way this has been turned into a new medium is fascinating, but does remind me of a long running argument about viewer attention spans that started with MTV's promotion of the short form slot. The theory was that shorter content resulted in attention deficit and therefore made a generation hyperactive. However, ironically, Hollywood movies have steadily been getting longer over the past couple of decades and television seems to have made an art out of three hour reality programmes where nothing much happens.
This seems to be a case of the media fitting the medium - upload sites can currently only cope with short form uploads without overloading their servers, and the fact that videos are uploaded in raw format and encoded on servers (far more time consuming than encoding locally and uploading the files subsequently) also limits film lengths. Expect a whole raft of tools such as Microsoft's Movie Maker (there is a new product coming along in Vista, I understand) that allow local encoding of content and will be linked to upload sites. To date, this has proved to be a bit complicated and less popular than the short form sites, but is surely set to change as long form consumer content takes its place alongside the short form.
The medium, after all, is the message...