Cuban on Streaming HD: It Won't Work

Too bad it is already happening on great sites like www.revision3.com.  Strictly speaking, it is not streaming, but you can download what you want and watch whenever you want.  That is the type of TV I am looking for anyway.  I rarely watch a show when it is broadcast, that is why I own a PVR...

 There will be performance issues to overcome and the ISPs will have to find more ways of monetizing their networks, but it will inevitably happen.

 

John

Hmmm...wonder why anyone should trust Cuban's opinion on this subject because he wouldn't want to do anything to rattle his relationships with traditional media companies.  How can someone who came from technology ever say the phrase "you can't" -- that's the beauty of the internet.  The problems that prevent us from doing what we want today are only future solutions people are currently working to solve.  Who remembers the day when you had to hook your rotary phone receiver to access a modem?  I'm sure the company that made that device said the same thing about other emerging technologies.

Cuban's a businessman protecting his livelyhood...looks like he'll say anything to shut down the rise of the internet over television. 

The telcos said the same thing about voip in the mid 90s.

In the extreme Mr. Cuban may right. If “everyone” wants to watch a streamed (i.e., unicast and individually controllable), HD version of the Super Bowl, things would probably fall apart. But, for that matter, Mother’s Day is still a problem for some phone companies. These are really corner cases, the vast majority of people will be able to view HD video streamed over the internet directly to TV-attached set top boxes.

However, Mr. Cuban is wrong in thinking that the problem is in the “internet.” He seems to believe, like many in Washington, that this sort of video actually goes over the internet. It doesn’t. Content owners and storefronts pay content delivery networks (CDNs) to bypass the internet and deliver traffic directly to the broadband service provider. These CDNs are well engineered networks, many with distributed caching, capable of providing the service quality demanded by HD video. The real problem is in the broadband network.

The broadband network is really a single undifferentiated “channel” that thinks MPEG-4 traffic carrying HD video is no different than SMTP traffic carrying email or HTTP traffic carrying web pages. While this is kind of dumb, it’s what we have to thank net socialists for.

More advanced broadband providers are experimenting with additional “channels” on top of basic broadband that deliver video (or VOIP, gaming, or VPN) traffic with the quality of service it is engineered to expect. This quality of service is typically expressed not as a priority but as a specific amount of bandwidth.

This seems to offer the only path forward that satisfies all parties. Consumers are happy because they can buy the service quality they want. Broadband operators are happy because it affords them the new revenue streams needed to justify further investment in broadband infrastructure. Even net socialist should be happy because the basic broadband internet connection still treats all packets precisely the same.

It is one thing to proclaim that HD over Internet won't work - it's another thing to create the technologies that will make it happen. The main obstacle of making it happen today is the limited bandwidth to support a mass market video service with the adequate quality of service. I'm not even going so far as to look at HD only, although this is definitely a growing consumer demand, but even a standard definition TV service is not feasible with the current unicast (streaming) methodology.

Furthermore, if we take a closer look we can see that all networks today are engineered to support peak time demand, and from a business standpoint it really doesn't make sense to over invest in network capacity to be able to support only a small portion of the daily hours. What's with all the unused off peak capacity. There is a real opportunity to monetize all that non-utilized bandwidth. All it requires is a change in the mindset to make it happen.

Users today are seeking to express their personality and this is also reflected in their video consumption. The Internet has introduced the means to easily find the content which meets a user's personal taste. As more users jump onto the bandwagon it will become an even greater burden for network providers to cope with the growing bandwidth demand to satisfy their customer's needs. In contrary to Mark Cuban's opinion, the younger generation is transforming their PC into their preferred video playback device. One simply cannot deny that a new generation of consumers is evolving, which is technologically more apt and which is relying on the Internet for most of their social engagements and entertainment requirements.

So how can we support this ever growing demand for more personalized and high quality content via the Internet?

Well, the technologies are already available and only need to be implemented.
A major step would be for network providers to finally realize that they have to move past unicast to make good on the promise to deliver high quality video (content) over their networks. By activating the multicast capabilities, already inherent on large portions of the network, networks providers can expand their service portfolio to introduce a high quality video service capable of supporting HD (and I'm referring to full HD), without heavy CAPEX investments. In addition this video service could be personalized to each users individual taste - now wouldn't that be nice.

Already today more and more users are delighted by the advantages of a PVR-like experience. As a matter of fact, users with a PVR or video-from-storage based consumption preference watches less and less linear TV. Titles are ready and waiting on storage to be viewed anytime the user so decides. Time-shifting is big today and more users are unchaining themselves from EPG slavery.

Still users must browse the EPG to find the titles of interest to them and manually manage their storage capacity. TiVO was the first to introduce a more personalized way of automatically getting content titles matched to your personal preferences. There's a dispute on how much personalization is actually needed to create a satisfactory, user friendly yet simple way for users to get only (or most) of the content they want.

A new technology and actually methodology to enable a video service which is in line with the changing consumer experience is Multicast-to-Storage. User requests for specific titles are aggregated to form content request groups, which are used to create a delivery schedule that covers most of the requested titles. The titles are then pushed directly to the user's device storage (PC, Set top box, Gaming console, you name it) via multicast which replicates the file at the outer most point in the network rather than sending it via individual dedicated streams (unicast) for each of the title requests.

The huge advantage is that the user subscribes to the titles he wants and then waits for them to arrive. The network providers can now utilize all that unused bandwidth (and there's lots of it) to deliver these titles in best quality. The actual access bandwidth of each user becomes irrelevant in terms of video quality as consumption is from storage without buffering, reduced frame rates and other diminishing factors that reduce the user experience.

So, Streaming HD may not work due to unicast delivery constraints - but this doesn't rule out that today’s networks are capable of providing a personalized HD video service over the Internet.

From what I read on his blog, he's missing the idea of why we watch Hulu - and it's because we don't have a TV (he says Hulu and others should sell their service through the cable companies, like some extra channels).

But I don't think cable companies will want this, and I'm not sure about users either.

Maybe he's just saying that to keep the people buttering his bread.

With the advent of Adaptive Streaming (AS) HD over the Internet became real.  There are several sites already offering HD content and YES they are streaming content directly to your desktop.  Technology is available for both plug-ins, Flash and Silverlight, and it can support 720p and 1080p at higher bitrates.  You do need a quality broadband connection of 5MB down but that type of connection or better are now readily available from your local Cable or DSL provider.  The magic of AS content is in ability to chunk data in small chunks and use “built-in” capabilities of client side plug-in to adjust video quality according to the current capabilities of the end-user’s Internet connection.  We will definitely see more and more of HD content on the Internet as various content owners start utilizing AS technology to achieve the highest and most appealing quality video experience for their customers.

The messages on this board are okay, but nobody has in a fundamental way addressed the real technology issues, the ones Mark Cuban are pointing to,  but i.e.  deep packet technology in the backbone of the Internet or even the last-mile problems.

Personally, I think that people typically underestimate the ability for technology to develop its own workarounds. Remember when DSL was capped by "laws of physics" at 128K, or whatever it was?  The Phds were going red in the face telling you we'd be stuck at 128K. I belive Cuban's "Super Bowl Paradigm" is extreme. There is plenty of ways to develop the content business over broadband without the Super Bowl. That's just one "mode" of the content business,  one type of event -- the big-ticket, live event. Movies are a totally different story. You do not have the stream movies and most of the bandwidth concerns could be alleviated with storage and/or caching. For example a settop box that contains a hardrive with 99% of the movies you might want to watch, so that when you want to buy them you just access a key that unlocks them.

So what is it? Mutlicasting, storage, adaptive streaming, caching, a combination of all of them? What to do about the Super Bowl?

 

 

Over next several years until Internet connectivity is reliable enough in terms of been able to deliver 10MB on consistent basis there won’t be a “silver built” that will solve last mile issues. Rather a combination of technologies will be used.  Multicast on public Internet won’t see a light any time soon because of several reasons, the main being that networks cannot charge for content delivery because in multicast environment virtually unlimited number of users can tap into single broadcast so network pipes don’t get congested.  In addition, to serve multicast content you don’t need large server resources.  More likely you will see a progressive download to hard drives for the on-demand content while content is preloaded for playback.  Note that Adaptive Streaming can take advantage of such process as well, where data chunks are preloaded into client’s cache and played back accordingly.  This can be used in both on-demand and live scenarios.  However, in live environment to do this you are delaying live feed for a few minutes but in return you get higher quality (actually you can get true HD quality) and network PVR capabilities.  Adaptive streaming is a truly revolutionary technology and best of all it’s still in infant stages.  

 

As far as Super Bowl, you encode content in HD but you also offer additional bitrates using Adaptive Streaming technology so your live video feed offering looks more like 300K, 500K, 750K, 1MB, 2.5MB and 5MB.  Next you preload as much of content on users’ machines and you let the technology serve the quality accordingly.  If you can live with a 2 minutes delay, you will get high quality and no buffering today.