Hacker News new | past | comments | ask | show | jobs | submit login
The Great Internet Video Lie (blogmaverick.com)
45 points by astrec on Jan 27, 2009 | hide | past | favorite | 30 comments



When I first saw the title, the lie that lept to mind was: "People want to stream." They don't. Streaming sucks and is a hack around content producers pathological belief that streams are somehow safer than normal downloads because they can't be captured. Or something. (It doesn't make any sense when you get down to it.)

They do in some cases want their video "now" (not in all cases), but they'd be just as happy if the "stream" stuck around in some form for a while rather than evaporating immediately.

Of course, I read the article, and at first it seems I was wrong. But... if you dig deeper, there's still an interesting aspect to my initial reaction. Delivering live streams to 10,000 users? Yeah, you're in for a world of hurt. But one hour of 1mbps video as a simple file is ~450 megabytes of data, and delivering that to 10,000 people an hour average is more feasible; there's more tricks you can play if you're just distributing a file and not a stream, not the least of which is torrenting. Some people might not get it in an hour, but it's a lesser problem to come close to an hour download time than to get everything necessary for the stream to work.

If you're not trying to stream, the problem is significantly reduced back to something manageable.

Whether this actually helps, I don't know exactly. But I certainly don't think it's as dire.

Interestingly, I observe that the big video sites seem to not stream anymore, they all let loading run ahead if possible. Maybe this is the reason?


Even conceding all of Cuban's points, this is really only relevant for events people want to watch live. Meaning sports and some news, mostly. Concerts, maybe, but what's the market for watching a live concert without being there? Even bands like the Stones, I think, don't do live broadcasts of their shows, do they?

But in any case, all of the fiction shows on television, and even news commentary (e.g. the Daily Show), work fine on the download model. And that is exactly what's happening, $1.99 from iTunes or free with commercials from the network's web site or Hulu the next day, in HD (don't think it's 1080p, but looks pretty good on my monitor).

So, again, even if Cuban is right about everything he says, it demonstrates a much diminished role for cable TV going forwarded, limited mostly to live events.


Speaking for myself, I am completely uninterested in a live stream of a concert at our current tech level, because the compromises to the audio and video to successfully stream are at their worst for a concert. Give me a real, DVD-quality MP4 (clocks in at around 1-2GB) after the event, thanks.


Here's an idea if you want to do almost streaming and run into the problems that the article adresses.

Use a modified version of bittorrent and delay the stream for one or two minutes. That should be enough to distribute the torrent chunks to all viewers. You could make it as an add-on to media players or as a stand-alone product depending on your income model.

Ta-da: problem solved :-)

Disclaimer: I don't know enough about torrents or streaming technology to assess whether this would actually work, but I don't see any conceptual barriers. Feel free to correct me...


Bittorrent isn't sequential, so it'd be tough to say with any certainty that you have established a buffer. I'm not sure if this is something that's recently changed or not though, I'm no expert either.


Bittorrent isn't sequential by design, but you can tweak a torrent client to ask for pieces in order (e.g., Pando, using a bittorrent-like protocol, can do this) -- though if you have a moving target in terms of the set of blocks you are interested in things a get a little more complicated -- a streaming-specific p2p protocol would likely perform better (see comments on this thread for examples).


Streaming can be useful if:

1. You just want to have a preview 2. You have a fast connection and you don't want to wait for the download / you just want to watch it once 3. You just want to click and watch


BitGravity has an API they call Advanced Progressive that allows progressive download with stream-like seeking.


p2p bit torrent streaming (built into browsers) would solve this. Hopefully this technology becomes more mainstream and then Cuban's point would be less valid!

I saw this one p2p streaming client on TC called rayv


"People want to stream." They don't.

Huh?

http://widgets.alexa.com/traffic/graph/?r=1m&y=r&z=1...

^ (youtube vs. piratebay vs. isohunt) have no idea how to save links in alexa, so that will probably disappear.


Per my last parenthetical comment, youtube doesn't actually "stream", and it's also fairly easy to download youtube videos directly, last I knew. Interesting tidbit, no?

But my point isn't that people aren't flocking to "streaming" sites. It's that if you gave them the choice between a forced stream, like the way Netflix streams movies, or a file that downloaded as quickly as possible, persisted, and could be viewed as it came in if your connection is good enough, they'll take the second one every time. The only way "streaming" wins is when it has something that isn't available that second way. (Or people don't realize it's available, or, gasp, don't actually want to break the law to get it.)


Mark forgot to mention a "Full disclosure" he founded HDnet, which is a HD television network.

I call FUD. Mark's argument is that point to multi-point streaming of live video events isn't feasible technologically because of bandwidth limitations. His argument is a straw man.

There are problems with point to multipoint video streaming. That doesn't kill internet video. Fewer and fewer people are watching live events as they happen. Many people want the content, but they want to watch it asynchronously, not live. People are increasingly recording the shows that they want to watch on DVR and then watching them whenever they want. Internet video is all about watching content whenever you want to, not about watching the content live. That's where Mark is wrong.

Who here watched the inauguration live? I watched it on Hulu 1 hour after it happened. Much more convenient than trying to remember to hit the DVR record button.

The boon that YouTube and internet video has been for content creators is that they can get any distribution at all. Before YouTube and video streaming, if you had video content and you wanted to distribute it, you could either try to get a TV station to pick it up, sell it to a movie studio, or market it directly yourself via direct sales.

It's actually _possible_ now for a content creator to distribute their video on YouTube and get it seen by one million people. That's the promise of internet video, not that one million people are going to watch your content all at once.

And, it's that sort of distribution is making a lot more cottage video industries possible like Peepcode: http://peepcode.com/ Where a couple of guys can make a living selling educational video streamcasts. That would never have happened 5 years ago.


To be fair, Cuban was also involved in broadcast.com and Red Swoosh, so he has experience with Internet video and traditional TV.

In the comments he says that on-demand streaming with a lot of simultaneous viewers is just as difficult as live.


Yes, on demand streaming is difficult, but it's not "the lie of internet video".

Perhaps he's aiming his blog post at TV executives. If they believe that they are going to be able to stream the Superbowl instead of having it on broadcast TV, then yes, internet streaming is a lie. But there are only a couple of hundred content producers that would be true for. Think producers and distributors of hit television shows or event broadcasters.

But, for the vast millions of content video content producers around the world that now have an audience and distribution, internet streaming is a dream, not a lie.


This may be true from the producer perspective. It's difficult to get your message to millions of people at one time using the Internet.

From the consumer perspective, however -- the internet is the clear winner, if you must choose one. There are millions of streams available at any one time. That's the beauty of it. That's why Internet video will win in the long run or at least be a viable option for the individual over television. If you are willing to sacrifice mainstream content, ditch the TV and go all Internet. I haven't had a TV in over 6 years.

But Mr. Cuban is right, TV will be around for a long time. It will remain the winner for mainstream content, but the Internet isn't about mainstream content it's about access to content that isn't popular enough to warrant distributing to the main streams via traditional wireless routes.

Anyone here also TV free? How many years?


Cuban's point is largely specious--the cost of distribution is high. The Internet doesn't magically make things cheap. Granted.

However what the Internet does do is level the playing field (at least compared to traditional media distribution methods). This means that cranky old white men clutching yesteryear's business models will have their lunch eaten sooner or later.

Even though scalability is an issue, it hasn't been solved because the demand hasn't been high enough yet. How many events have to scale that big? In this case, the long tail of content (meaning anything that needs to scale to more than a million simultaneous views) is the vast majority of content. Hulu doesn't have any trouble streaming the most popular television shows. How many Superbowls and Obama inaugurations actually happen?

Furthermore, the solution for massive scalability already exists. It's called BitTorrent, and Cuban was remiss in not mentioning it. BitTorrent makes scalability a function of total Internet bandwidth. It also transfers the hardware costs to the users (who don't see it as a cost). The bandwidth is still more expensive than television broadcast, but I imagine reaching everyone in the US via television is not exactly cheap either.


I've been TV-free for about 3.5 years, and only had broadcast TV (in Spain) for about a year before that. So the last time I had cable was in 2004.

Sadly the time savings afforded by not having a TV are often eaten up by the Internet nowdays, but at least it requires slightly more activity on my part.


> Anyone here also TV free? How many years?

Never owned one.


Back to p2p networks then, is it?


There are a ton of chinese live tv p2p networks ( tvkoo, ppstream, sopcast ), in theory they could stream the kind of numbers that Cuban is talking about.


Yeah, you are right. I have setup a live broadcast netwowk using TVAnts in my school's Intranet, about 3k-5k peak users. I think it's easy to expand it to 10k.

Report says http://news.sohu.com/20070320/n248849308.shtml in 2007 the peak viewer of QQLive reaches to 1650k during the Chinese New Year Spring Festival. I guess this year they may reach a much higher number.


tstream torrents from EZTV: http://trial.p2p-next.org/


I wouldn't be surprised if the P2P networks have less available bandwidth than the CDNs.


Does that matter? For every person who wants to view it, they add their bandwidth. If the average viewer is bandwidth-neutral, then it will work.

Certainly you can get an unlimited number of broadband users. My upload rate is more than enough to watch a good quality stream and upload at least as much as I download.


Cuban argues that serving many simultaneous video streams is expensive, but what's the ratio between simultaneous streams and total viewers?


.. and yet if you are to believe the NY Times (I also read something from Google that says the same thing, but was unable to find the link) that TV will be dwarfed by people using the internet in the not too distant future.

http://www.nytimes.com/2008/08/25/technology/25drill.html?_r...

Then again, Nielsen shows that the average American household watches significantly more TV than uses the internet?

http://www.nielsen.com/media/2008/pr_081124.html

I'd say that while this might be currently true - given the shift in eyeballs from the TV to the Computer Screen amongst the younger people, then costs for content producers should invariably become lower as time marches on - so cuban's point is a bit of a moot point.


the porn industry (notably cam4.com) proves this to be utter bull. The question is if you can make it pay, sex makes a good bit of money so that's where you'll see tech that is 'not feasible' years before the mainstream catches up.


Doesn't Joost address the simultaneous user problem?


Most people don't have time at the same time and will watch stuff timeshifted to one another.


Hold on. Has he never heard of ustream.tv?




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: