YouTube Launches "Video Speed Dashboard", But The Results Don't Tell You Anything

Earlier today, YouTube launched a new "YouTube video speed dashboard" saying their goal is, "to give you insight into what your YouTube speed looks like compared to the YouTube speed of users in other regions and different ISPs." While that sounds like a nice idea, the results don't actually tell you anything in terms of your "YouTube video speed" or the quality of the video you are viewing. Since the results are not based on the data from an actual stream, YouTube's video speed dashboard is nothing more than a speed test that every other ISP has, which gives you no data on the actual performance of any video being delivered. It simply tells you the size of the pipe available.

YouTube's new speed dashboard says I am at 9.33Mbps, yet myself, and others, still get a lot of video buffering. There is no problem on my end with my ISP, the problem lies with how Google delivers video. As anyone knows, there are many factors that go into the buffering issue and while the last mile is one issue, it's not the only one. So who cares if YouTube sees that I have a 9.33Mbps connection if the 500Kbps video stream they are sending me isn't working. It's not just about the size of the pipe that matters.

Google says that "A higher YouTube video speed translates to a better and faster experience", but their new speed dashboard does not provide any real-time data or analysis of what the "experience" actually is. On YouTube's blog announcing the new dashboard they ask, "So, what can you do with all of this new data about your video speed?" The answer is easy, nothing. It's completely useless for telling a consumer anything about the video experience they are getting from YouTube.

- Why Can't YouTube's Player Auto-Detect When A User Should Get HD Quality?

- We Should Care About YouTube's Core Business, Not Their Market Share

- YouTube's Biz Blog Goes On Offensive, Says Industry Comments Are "Myths"

Comments

[John Luther]

does not provide any real-time data or analysis of what the "experience" actually is

Hi, Dan. Did you try the test video? Click the Show Test Video link under the charts. A video will play and show you live stats in a callout box.

[Dan Rayburn]

Hi John, yes, I played the "test" video, but there are a number of problems with it:

- the video file has almost no motion in it at all. 90% of the video window is taken up by a powerpoint slide with a little tiny video window in the upper corner. this is NOT what the average clip on YouTube looks like. most clips have full motion.

- the file is only encoded at 259Kbps. It is way too low of a bitrate to really be able to judge what kind of performance one is getting. of course this clip is going to perform well for almost everyone since it is at such a low bitrate. but most of the content on YouTube is not encoded at 259Kbps.

- this video is a "test" clip. meaning it is not a popular file, gets no traffic and is not representative of what the experience is for content that actually get traffic on YouTube.

- this data is not available for any content on YouTube, only for this one file. is YouTube distributing this one file, the test clip, differently than it does for content that is popular and has traffic on their network?

To me, this clip tells the use almost nothing useful.

[Lou Quillio]

We are not the audience for the YouTube Video Speed Dashboard. It's not even about YouTube. It's about this [1] and this [2] and a general public education campaign about how the last mile is the chief bottleneck for bandwidth-demanding applications. And it is. Among Google services, YouTube is the best illustration. But it's not about YouTube per se.

Since this is fairly obvious, seems like you ignored it in order to play Gotcha.

What you can "do with all of this new data about your video speed" is agitate for a fatter last mile, or at least be aware that all the stuff you want to do at once is limited foremost by the size of the pipe you get.

Put another way, Google is beginning to explain its place in data delivery. If CATV or cell or electricity service is out or spotty, we blame the end-to-end provider, because it positions as such. If YouTube or Hulu delivery is sucky, there's only so much to be done server-side. Lots of downstream business decisions figure critically in user frustration. Society doesn't understand that, and Google needs it to.

LQ

[1]: http://www.measurementlab.net/
[2]: http://googleblog.blogspot.com/2010/02/think-big-with-gig-our-experimental.html

[Dan Rayburn]

Lou, your comment could not be further from the truth. You say that this is about "a general public education campaign about how the last mile is the chief bottleneck for bandwidth-demanding applications."

Really? Is that why YouTube videos looks so bad, because people don't have enough bandwidth at the last mile to get a 500Kbps stream? No. I don't think so. It's because YouTube's delivery stinks, not the users last mile.

You say that "If YouTube or Hulu delivery is sucky, there's only so much to be done server-side." 95% of the time, it's a problem on the content owner's end, NOT the last mile. The idea that the last mile is the reason for YouTube's video quality being so poor is simply wrong. There are no facts to prove your point.

[Predrag Mandlbaum]

Google is preparing users for the upcoming commercial version of You tube ... ;-)

[Lou Quillio]

Dan: From a video encoding standpoint, Google's decisions re YouTube are what they are. Could they be more ambitious? Sure. Evidently they have reasons not to be. Among these are a less than devout commitment to Flash Video, IMO. Isn't it simplest to conclude that Google is managing user expectations until third-parties have less control over the stack? Meanwhile, YouTube.com is a global top-five popular site by every measure I can find.

Anyhow, YouTube video encode quality is *your* favorite topic, and Google isn't speaking to it every time it says something about YouTube. Seems clear to me that Google is speaking to consumers about their ISPs, and YouTube just provides an opportunity to do that.

> It's not just about the size of the pipe that matters.

Of course that's right, but it's the only topic taken-up by the Dashboard. The Dashboard fails as a technical dissertation because it isn't one.

The gist of your post seemed to be, "Why did they bother with this Dashboard-thing when it doesn't answer my questions?" My reply is that I believe you're mistaking its purpose.

LQ