“ With internet TV you have a network configuration problem that is possible to solve. A couple of the examples above - where everyone wants to watch the same thing at the same time - are the easiest to solve. If the last link has the bandwidth - the phone line for DSL, the coax for cable - then the problem comes down to getting the content to the node before the last link - before the subscriber line. For DSL, it means getting the content to the local phone switch (possible up to 50,000 people), for cable it means to the end of each subscriber loop (100 people). If you cache all video in the last node, then delivery is only a problem when everyone wants different content. Then the link to the last node has to be wide enough to transer all the content simultaneously. We do not cache much today. Rather we rely on getting content from the source with little or no near subscriber caching. With the collapse in the cost of disk storage - falling by a factor of 4 in the last year - the cost of caching in the last node will fall quickly. It might already be cheap enough to allow caching without raising the cost to the subscriber. However, in Canada we have no competition beyond a small oligopoly which has no interest in lowering price. So while they could do this for $40 a month, they want to charge $10 per show in hopes of getting hundreds of dollars a month. In short, they hope for a repeat of the cellular telephone windfall profits. Once spoiled by cell phone profits, communication companies and the compliant CRTC (a total captive of the industry) will never consider delivering value to customers. (And I really hate that each cable package still contains many channels that I never watch but still have to pay for.)

“ Yes, it is possible to use caching, or other technologies designed to conserve back and middle-end bandwidth to solve some of the problems of serving 10 million simultaneous users of the same video feed ... but ... why would you do that when you have a much better designed system for that application already in place? And of course, if those ten million people are not all watching the same thing at the same time the problems become worse. The Olympics Broadcast consortium reported peak usage of 133,000 live streams - and 22 million television watchers (during the men's final hockey overtime) all of whom were watching a much fatter bitstream than the Internet users. People have trouble appreciating the huge difference in scale between the use of television and the use of Internet video. In part, that's because they are never measured the same way. Converting streams served to hours viewed is not simple. When you attempt the math though, the conclusion is the same as Mr. Switzer's. The Internet is superb at doing what it was designed to do. - the cost of making the Internet do what it was never designed to do (replace television) is beyond the value of doing it by about two orders of magnitude.

“ The heritage cable system was designed as an analog broadcast standard definition TV system, which it does very well. Overlay digital TV, internet and voice, as we now have, this system is a bit of Frankenstein. Thus the 'good old' cable system is due for an upgrade anyway. One of my last IPTV architecture projects for a telecomm vendor (which was about 9 years ago, so no HDTV traffic analysis) we figured on a few hundred multi-cast streams could be supported in an optical metropolitan network with some caching. Then the telecomm bubble burst and the carrier network infrastructure upgrades slowed down considerably. Note that multi-cast is not the same as broadcast - it uses Internet protocols to better manage network bandwidth. With the advent of HDTV, both network infrastructures are badly in need of an upgrade anyway. Network bandwidth technology steps up by orders of magnitude, Ethernet has gone from 100M to 1G to 10G in the last 10 years. 10G is now mature, and 100G is on the horizon, so the requirement for 2 orders of magnitude upgrade to the network is achievable and on top of that, multiple wavelength technologies are getting cheaper too. As already pointed out, the problem is not in the 'last mile' technology, whether its DSL or coax or even wireless for that matter, but the network behind them. I think that mobile broadband will place an even more pressing need for network upgrades, so I believe that the carriers will have to make the investment anyway.

“ Jay its a good perspective and I buy in . Broadcast distribution networks are better for mass distribution of the same content at the same time . The concern is that as anywhere,anytime viewing fragments audience and splinters time (in the sense of concurrent viewing) the internet will increasingly be a disruptive and important platform to deliver what broadcast distribution cannot do as well , at least to date. Unless all elements of the traditional system can buy into a collaborative approach to an on-demand world (eg some form of revenue sharing), the health of the traditional ecosystem will remain at risk . There has to be an incentive somewhere in the content value chain to create new business models for content or a large chunk of what supports mass distribution will be eroded to the point where the distribution platform and the content it supports are no longer sustainable. Just because one platform is better for something than another does not ensure long term success.