It’s Time For Netflix To Say Goodnight To Silverlight


In 2006, Netflix scored a grand slam when they announced a $1 million prize for anyone who could improve their recommendation engine by at least 10%. It took 3 years for a team of scientists to actually accomplish this feat, but the prize was ultimately worth far more than a million dollars in publicity and to Netflix’s bottom line. Better recommendations not only led to happier subscribers (less churn), but they also made it easier for Netflix to sell the niche content that they spend less money on. Recognizing the benefit that they received from the contest, Netflix was quick to announce a sequel, but ultimately had to suspend their plans over privacy concerns.

While a contest to replace Silverlight likely wouldn’t garner as much attention, I believe that the financial benefit to replacing this video platform could be just as significant.

Some will argue that I’m being tough on poor old Softie and that Silverlight represents some of the best video compression out there, but consider my logic for a moment. From where I’m sitting, Silverlight has two basic flaws: It’s buggy as all get out and it’s a bandwidth thief.

The screenshot posted above is a real life example of Silverlight in action. All video frameworks are prone to errors of course, but look at all the hoops Netflix makes their customers jump through just to support this buggy piece of software. If I had a nickel for every time I’ve had to restart my browser after a Silverlightning strike, I’d probably have .35 cents by now. Seriously, I have less trouble with Real Network’s codec and that’s saying a lot. Instead of putting up with these kinds of errors, Netflix should be actively searching for a more reliable alternative.

Given Netflix’s runaway success, it shouldn’t be a surprise that the big telco companies are running scared. While usage based pricing hasn’t hit the US yet, the Canadian telcos were very quick to raise rates the minute Netflix invaded their territory. When you consider how many internet service providers also sell video, it’s clear that Netflix will need a way to undercut these tactics, especially if they plan on expanding internationally. Currently, an SD movie over Silverlight clocks in at approximately 2 Gigs, while an HD movie will cost the user 3Gbs towards their cap. If Netflix could reduce the size of a movie file by 50% – 75%, without sacrificing quality, they could end the usage based meter for their customers, while also undermining a critical future component to their latest competitors’ business model.

Getting Hollywood to sign off on an outsourced video codec could be a potential problem for Netflix, but even if they were able to gradually ween their customers away from Silverlight by delivering independent films with the new technology, the benefit could still be substantial. Given how little they pay for traffic, they probably wouldn’t save $1 million on their bandwidth bill, but being able to stop telcos from nickle and diming Netflix’s members would be priceless and would help to future proof their business.

Davis Freeberg is a technology enthusiast living in the Bay Area. He enjoys writing about movies, music, and the impact that digital technology is having on traditional media. Read more at Davis Freeberg’s Digital Connection.

21 thoughts on “It’s Time For Netflix To Say Goodnight To Silverlight”

  1. What? You do realize that a DVD (SD quality) can be up to 8GB and a Blu-ray Movie file can be up to 25GB. You do realize that Netflix HD is already compressed to all hell. What magic codec do ou want them to use. In fact I wonder if you realize that Silverlight is not even a codec. As far a streaming goes, Silverlight is the hands down best until somebody gets something better. Ahhh the joys of 1st world problems.

  2. What browser are you viewing netflix. Does this error come up with all browsers? I stream all the time with silverlight (not just watching netflix) and I have never come across that problem.

  3. Seems like you are error prone or you’re not, as there are other sorts out there, but I’ll let Davis respond for himself.

    I do agree that additional compression is unlikely and undesirable. But there could be a better platform. Or not. (I’ll take Silverlight over Adobe/Flash who I no longer trust or appreciate…)

  4. Joe public has a problem with their Silverlight install they post in the forums. A-list blogger has a problem then Netflix has to switch their technology strategy.
    SL has never crashed for me and is supposed to use the available band width.

  5. SilverLight has hosed up on me a few times: Mac + Safari + SilverLight.

    I had to completely uninstall and reinstall it. It was a pain.

  6. Joe, We’re fortunate to have a soapbox. But what’s the point if we don’t use it to share our personal experiences? I prefer we speak first hand of things we like or dislike, rather than to mechanically present some company press release on command. Also, we see ourselves as emcees and leave the comments open for discussion. We’re totally cool with folks disagreeing and/or proving us wrong. And thank you for putting us on the A-list.

    Regarding Silverlight and Netflix, I can’t recall experiencing any desktop problems and prefer it over Flash for a variety of reasons.

  7. I echo the point about the quality, and today most of the streaming support is with lousy audio to begin with. There’s a pretty clear disconnect between people who are willing to watch TV on a phone and people that want real video and sound quality. I think that divergence in expectations/goals will continue.

  8. John is right, they are looking to HTML 5. I wonder what percentage of their streaming pie Silverlight makes up anymore. You are only getting Silverlight if you are using a browser to watch Netflix, and with Netflix on hundreds of devices that has to be a rapidly shrinking percentage. I am with you though, Netflix is the only thing that I have Silverlight installed for and I’m looking forward to the day when I can ditch it completely.

  9. Don’t get me wrong, Silverlight is a very impressive piece of technology, but I guess that the heart of what I’m getting at is that we can’t allow codec (or container) technology to stagnate here. Just because we can quickly download movies, doesn’t make it adequate. One of the biggest pressures Netflix will face when they are expanding internationally will be incumbent providers jacking up rates, if they can help to develop an open version of an even more efficient codec, they will help to bulletproof their business from these evil plans. As far as my being error prone, that’s probably true because I had another silverlightning crash just last night. This time all I had to do was close and open the browser.

  10. I can’t see them going to a better codec than what they have now regardless of platform. Once they put 5.1 Dolby Plus surround on all streams, the size will keep going up. I think complaining about the size of the stream is very hard to do. Ever evolving codec mean you need a platform(silverlight or flash) to keep up. HTML can’t keep up. They haven’t even decided on a codec standard. h.264, Ogg Theora, or god help us all WebM. Who do you put the trust in being able to play the latest codecs… (Silverlight/Flash or HTML). The codec being open does not mean better. Ogg Theora has noticeably worse quality vs h.264 for the same bitrate. My dad always said to me, There is Quality, Cost, Time… you can only choose two. Same here. You can’t have it all, Size, Quality, Standard, Openess, Constant Evolution.

  11. It’s also probably worthwhile to distinguish codecs from containers and platforms. Davis’ errors/bugs aren’t specific to the codec, but specific to the Silverlight platform and plugin. Here’s an old, old blog post some of the technology Netflix uses for streaming:

    Realistically speaking, I agree we probably don’t want to compress content any further. Making Davis’ allusions to caps and net neutrality all the more of a concern. As we get more content over the top, it could very well become an issue… or revenue stream. Neither a positive development for the typical consumer.

  12. Absolutely… Codecs are not containers and vice versa. The Netflix blog post states the same issues i posted about being able to do want they want within a standard. Eventually what needs to be said is that Microsoft needs to get Silverlight working much better across all platforms. I have never had a silverlight streaming issue but I realize there are issues. Netflix should move to the latest Silverlight though. I believe they are still on Silverlight 3.

  13. I’ve never had a problem with Netflix streaming in any of the browsers i’ve tried it on, granted i’m on a PC, I don’t know if you’re on a Mac. I’ve generally preferred Silverlight over Flash or other alternatives when i’ve seen both used for video streaming.

  14. Actually we’re pretty big fans of Netflix around here and support what Microsoft has done in this space as well. But this is obviously just a drive-by on your part…

    Regarding bias, It looks like you’re commenting from a financial firm who may have business dealings with Netflix and Microsoft…

    We welcome constructive discussion and even criticism, but personal attacks and comments devoid of useful contribution are unwelcome. Troll/astroturf elsewhere.

  15. It would be awesome to see a poll of netflix subscribers on how many stream to a computer, vs a netflix streaming device.

  16. Kind of an off the wall post in my book. Silverlight isn’t a codec, and uses the best codec available at the moment. Certainly there are going to be improvements over h.264 but its going to be a while, and some magical 50-75% bandwidth improvement without sacrificing quality seems unlikely at best. I’d have to call this request naive given the soapbox its being put on.

    Silverlight itself is one of a set of containers trying to do ABR streaming at this moment, something Apple, Adobe, Microsoft and many others are doing simultaneously. Basically switching the bit rate every couple of seconds based on observed bandwidth, latency, error rates to get you the best possible video stream given the connection you’re on without you having to do anything. And start up quickly. If Silverlight has some bugs I’d have to say I’m not suprised and assume Microsoft and Netflix are working on them. Take it to the forums dude.

  17. As others have said, Silverlight isn’t a codec, but a framework/platform, kind of like Quicktime, or Flash. (Really, it is MS’s attempt to compete with Flash.) The codec Netflix uses with Silverlight is, IIRC, VC-1 – the standardized form of WMV3 AP. VC-1 is one of the advanced codecs used in Blu-ray (and HD DVD), H.264 being the other one.

    VC-1 and H.264 are roughly comparable – each one has strengths and weaknesses, and each one will handle some content a bit better than the other.

    H.264’s big advantage is HW support – it is the de facto standard in advanced codecs and HW support is baked into lots of chip sets these days – VC-1 not as much. Chips designed for STB use tend to have both, but mobile devices generally have H.264 decoding in HW, but often not VC-1. (IIRC Netflix uses H.264 for streaming to at least some mobile devices, like iDevices.)

    So while VC-1 & H.264 are generally comparable in quality/compression, H.264 has the advantage of wider support.

    The other advanced codec, somthing of a newcomer, is VP8. VP8 has been around for a little while, and was originally created by On2 Technologies. Google purchased On2 and later open sourced VP8 and opened the patents for others to use. VP8 is the codec in WebM and WebP and is a highly advanced codec comparable to H.264 & VC-1. I don’t know why an earlier commentor said “or god help us all WebM” – WebM is a good spec. It pulls some of the best available technology together – the MKV container format (simplified for WebM), the VP8 video codec, and the Vorbis audio codec. For streaming video this is a very solid technology set, and existing implementations have shown it works very well.

    WebM has a lot of potential in the future of web video, and with the broad backing it has WebM/VP8 is likely to be one of the codecs standardized in HTML5 – H.264 is probably the other likely candidate as it’s the other one with any real backing.

    Ogg Theora was brought up – that’s older tech, and yes it is NOT as good as VC-1, H.264, or VP8 for the same bitrates. Ogg Theora is an open sourced version of On2’s old VP3 codec – several generations before VP8, as you can tell from the naming. It is more comparable to MPEG-2 and that generation. VP3 dates back to 2000, Theora to 2002. VP8

    BTW, we’ve all been using the VP codecs for a while. VP6 was/is widely used for Flash video as the standard codec starting with Flash Player 8. A lot of the Flash video you’ve seen is VP6.

    Flash added support for H.264 in Flash Player 9 Update 3, and a lot of newer Flash video is using this. (Flash will be adding VP8 support as well.)

    Anyway, the issue isn’t really with the codec. Switching to H.264 across the board from VC-1 may help Netflix. Since H.264 is more widely used there are more people focused on optimizing H.264 encoding, and strong arguments that the encoding tools for H.264 are better than for VC-1 at this point. But it probably isn’t a big jump. Really the differences between VC-1, H.264, and VP8 aren’t that big on the quality/size front.

    Saving bandwidth isn’t something they can easily do, their current encodes are already highly compressed. Keep in mind they do HD streaming starting at 2.6Mbps. Even at 720p (instead of 1080i, let alone 1080p) and 24 or 30fps, that’s compressed to hell and back. The same content on an HD cable or satellite channel, even assuming the same 720p resolution and frame rates, would get a lot more bandwidth. And on Blu-ray it’d probably be 1080p and 10x the bandwidth, or more, for the bitrate. They’re already pushing the codecs hard.

    It is going to take the next step change for any big jump in compression – some new idea in how to do encoding and compression. Just like the MPEG-4 AVC/H.264 generation was a major step change from MPEG-2. Or how WMV3/VC-1 was a big jump from WMV2. Someone thinks of a new way to encode the video to digital, someone else creates a new compression algorithm.

    It is like file compression algorithms – back in the day there was ‘compress’ which used the LZW algorithm, this was displaced by gzip which did a better job at compressing and used the Deflate algorithm. Then the bzip2 algorithm did yet a better job. And then 7zip came around using the LZMA algorithm and I believe that’s currently the champ. (This is all generally speaking.) There are always people (generally mathematicians) looking at algorithms for ways to improve things like video encoding and data compression. But it isn’t such a simple thing to find the next breakthrough that provides enough of an improvement to cause the whole market to make the jump.

    We’re going to be dealing with MPEG-2 for many years to come – it is ‘baked in’ to DVD and ATSC, as well as other systems. Even though it is clearly inferior to newer algorithms it isn’t so simple to remove – neither will VC-1 or H.264. Even it someone came out with a better codec today, say one that halved the size, it would still take time for that to get baked into HW and for that HW to reach the market. And Netflix would still have to support existing platforms for quite a while. Supporting the new codec wouldn’t be economic until there was a critical mass of installed users for it.

  18. Oops, I got interrupted while writing that and never finished the paragraph I was on. When I said “VP3 dates back to 2000, Theora to 2002. VP8” that last bit was supposed to be “VP8 is only from 2008, it is much more recent technology.”

  19. “From where I’m sitting, Silverlight has two basic flaws: It’s buggy as all get out and it’s a bandwidth thief.”

    Netflix is absolutely rock solid with Silverlight. It’s not buggy at all. Secondly, it’s the least resource intensive streaming technology out there, because it bases streaming on your connection speeds.

    Silverlight is far superior to Flash, and superior to HTML5, especially based on what I’ve seen of HTML5 streaming on Youtube.

Comments are closed.