Om Malik nailed it at the end of one his posts on the 250GB cap. He lists Comcast's stilted usage examples for email, music, etc., and asks "How many HD movies will fit under the cap?"
That's the right question, but not because he is worried he'll bump the cap watching video.
The reason it's the right question is that the cap has more to do with Comcast's long-term business plans -- and the cable/telco industry's plans -- than about slapping down a handful of network hogs.
One fact before we continue. As I've written before, while everyone squabbled over BlueRay or HDDVD, the filesharing world settled on the x264 implementation and a container format for it. In that compressed format, a high-def film averages 10-15 GB.
But here's the real deal:
Comcast's cable TV and on-demand video business competes directly with Internet video (both legal and gray-market). By making the cap high enough that it won't immediately impact YouTube watchers but -- over time -- will put a serious dent in customers' ability to get HD video over the Internet, Comcast is simply crippling one product so that customers will have to buy another product from them. Don't even think about mentioning broadband competition -- if you define competition as more than one entity offering substitutable goods (i.e. similar bandwidth, QoS, etc), then most of the U.S. has zero competition.
The 250GB cap is a way to push customers to buy the same content twice. To be specific, you'll pay to see a movie broadcast on cable TV in HD, and then pay again for some special plan that lets you watch the same content later on Hulu. Or you'll pay for Comcast "on-demand" to watch a show on your TV. But if you want to place-shift that show to your laptop or the gym (on your phone), you'll have to pay to get enough headroom to download the same video.
This move is not just about trying to squeeze out more money. It's yet another attempt by network operators to neglect their core business and instead try to get into a content business, so they can do two things badly instead of just one.
Meanwhile, Steve Gillmor writes that bandwidth caps will mean the end of BitTorrent downloading and a shift to streaming. Steve's a smart guy, but streaming is one of those perennial non-trends like speech recognition and thin-client computing. It's only conceivable to someone who has a computer connected to their TV or who lives on their laptop.
Having played a key role in several home-entertainment / media-tech startups, two of which have been successfully acquired, I feel confident saying streaming is only a little piece of a solution. For the foreseeable future, streaming will never be more than that. It neglects too many consumer needs and desires, is too inflexible, and is too warped by DRM and platform-lock-in issues.
Moreover, the latest round of DRM around HDMI and high-def media playback in general means that even in-home "streaming" -- i.e., from a PC to a TV or from one PC to another -- is facing setbacks just as Joe tech-savvy consumer finally starts to realize how to use a home server or network attached hard drive to get and watch shows.
And the set-top boxes, increasingly tied in tightly with content delivery (Comcast DVR, DirecTV TiVo expiring your pay-per-view movies, etc.) are more closed than ever. Instead of encouraging an ecosystem of streaming, they force customers into an ecosystem of downloading.
I think we'll see more people gaming the 250GB cap ... using scheduling software so they can maximize the utility of the 250GB and hoard downloads. Not to mention leaving the torrent client running so that every time I visit someone else's network -- an office, a coffeeshop, etc. -- I'll keep the downloads slowly but surely coming in.
One final note: the marginal cost of moving a kb of data around a big network -- that is, the cost of the power and cooling for the switches and the amortized cost of the cables and gear and ops guys divided over their lifetime data transfer -- is more or less zero when the network isn't already near peak bandwidth somewhere. So even if there is a problem, the volume-based cap is the answer to a different problem.