Myth 2. HTML5 is Ready for Prime Time

Upon reflection, it's not surprising that HTML5 isn't used by any major sites, since it's clearly not ready for prime time. Deficits fall in four major areas, installed base of HTML5-compatible browsers, the requirement to encode in multiple formats to support HTML5, performance and a critical feature deficit for sites where streaming video is mission critical. Let's take each in turn.

Installed Base

Figure 2 shows the installed base of browsers from a report produced by NetMarketShare. As you can see, as of July 2010, Internet Explorer, which offers the least HTML5 compatibility of all browsers, still sits at over 60% as of July, 2010.

browser market share.jpg

Figure 2. Browser market share.

More disturbing is Figure 3, also from NetMarketShare, which shows how many 'Netizens don't instantly upgrade to the newest browser or browser version. The 17.58% still using Internet Explorer 6 is remarkable when you consider that this browser was first introduced in 2001.

browser version market share.jpg

Figure 3. Market share by browser version.

With Internet Explorer 9 scheduled for release in 2011, these numbers should shift significantly over the next few years. But remember - while many consumer and small business users are free to update their computers at their whim, huge numbers of computers in major corporations, schools and government offices are completely locked down, and can't be upgraded without corporate-wide fiat. In addition, for every technology innovator who updates their browsers as soon as possible, there are twenty or so in the late majority or laggard classes who couldn't care less.

Locked down desktops and non-early adapter users are why Microsoft's decision not to support WebM unless already installed on the computer is so critical (discussed more below). HTML5 proponents assume this means that all IE9 users will actually find and install WebM from another source. But what about locked down desktops that can't install additional software, or Uncle Harry, a casual surfer who thinks that web updates are the surest way to catch a virus? If you're trying to reach corporate desktops, or technology laggards, will you assume they can will find WebM? (As an interesting aside, will Adobe include WebM in the downloadable Flash Player if it turns out that Flash is the source of WebM for all those IE9 desktops that might otherwise not obtain the codec? Will HMTL5's codec-related success with IE9 depend upon the Flash plug-in HTML5-tistas love to hate?)

Ditto for Google Chromium - it's great to say that IE users can install that plug-in and gain immediate HTML5 compatibility, but that assumes that the user can (and knows to) install the plug-in. Not to mention that most of the appeal of HTML5 is the elimination of plug-ins - so now I have to install a plug-in to avoid installing a plug-in?

In stark comparison to HTML5's penetration stands Flash's audited penetration statistics.

flash penetration.jpg

Overall, while users seem to refuse to update their browsers, they stay pretty current on Flash.

HTML5 Performance

For reasons unknown, HTML5 had assumed an unproven and unwarranted mantle for superior performance. Then I actually tested HTML5 performance against Flash and found Flash performance superior in most test scenarios. Here was my primary conclusion:

When it comes to efficient video playback, the ability to access hardware acceleration is the single most important factor in the overall CPU load. On Windows, where Flash can access hardware acceleration, the CPU requirements drop to negligible levels. It seems reasonable to assume that if the Flash Player could access GPU-based hardware acceleration on the Mac (or iPod/iPhone/iPad), the difference between the CPU required for HTML5 playback and Flash playback would be very much narrowed, if not eliminated.

Apple did later open up their GPU-related APIs, and Flash's performance improved dramatically, well ahead of Chrome running HTML5, though still behind Safari running HTML5. 

Obviously, browser vendors are integrating GPU acceleration into future browser versions. However, until those browsers are released and tested, it's silly to assume that they will be more efficient than Flash.

In addition, as discussed below, GPU accelerated Flash playback can run on all browser versions that have the necessary graphics hardware, while browser acceleration will require a complete browser update. This means that it will take much longer for HTML5-based hardware acceleration to reach the masses.

The Codec Conundrum

Which gets us to the codec conundrun, which makes the BetaMax vs. VHS and HD-DVD vs. Blu-ray wars look logical. Briefly, there are three HTML5 compatible codecs, Ogg Theora, H.264 and WebM. Here's the status of current browser support from the excellent Dive Into HTML5 site.

current browsers.jpg

As you can see, no format is supported by all browsers. Here's what the situation will look like after all announced browsers are shipping, though Microsoft has announced that IE 9 will support H.264 playback, and VP8 if installed on the system. Here's the precise quote "In its HTML5 support, IE9 will support playback of H.264 video as well as VP8 video when the user has installed a VP8 codec on Windows." 

fugure browser versions.jpg

To be fully HTML5 compatible with all current and legacy HTML5-compatible browsers, web developers will have to support at least three formats. For example, here’s a code snippet from www.jilion.com/sublime/video, showing the syntax used to deploy the necessary files produced by Jilion for their HTML5 player.

julian.jpg

Yes, that’s four files, and if the website uses VP6 for Flash, they have to produce in that format as well.

Perhaps producing these extra formats would make sense if they delivered some benefit above and beyond H.264. To the publisher of free internet video, H.264 is now free in perpetuity, however, so there's no benefit there. Ogg has much lower quality than H.264 according to my own tests and many others, and now seems as dead as Latin. WebM is promising, and offers similar quality to H.264, though hardware playback support is much further along with H.264, resulting in a superior playback experience for H.264 on a wider variety of machines.

Now onto the final reason HTML5 isn't ready for prime time.

HTML5's Feature Deficit

This is pretty objective, so I'll make a quick list of key features currently unavailable in HTML5.

- True streaming - where a server streams video as needed to the player, or a player retrieves data as necessary to sustain smooth playback. Key to efficiently satisfying high volume of users.

- Adaptive streaming - the ability to switch between streams to adapt to changing connection conditions and ensure the highest possible quality for both low speed and high speed connections.

- Digital rights management (DRM) - either user authentication, or any mechanism for encryption.

- Live streaming

- DVR functionality - key to many of athletic event offerings like Sunday Night Football.

- Multicasting - the ability for one stream to be shared by multiple viewers. Very efficient in enterprise settings.

- Peer to peer delivery - the ability for peers to distribute streaming video, which can be exceptionally efficient when multicast isn't available.

All of these technologies either are or will soon be available in Flash and Silverlight; none are available in HTML5 or even that far in the development cycle.

To be fair, many of these features are irrelevant to many, many sites simply delivering video playback in a window. However, UGC aside, and that's clearly a special case, most video streams are produced and delivered by media sites like the three letter networks who need to both monetize and protect their content.

To these networks, and many other video distributors, the features currently unavailable in HTML5 represent signifcant obstacles to HTML5 adaption. As shown in the table below, in the StreamingMedia Survey, we asked respondents to rate their concerns about HTML5. As you can see, the codec issue rated a 4 or 5 for 56% of respondents, while 57% of respondents rated browser penetration a 4 or 5. The lack of adaptive streaming, DRM and complete support among advertising networks was also very concerning. These are not abstract feature deficits, they are significant obstacles to reaching potential viewers via HTML5, monetizing content or equaling the user experience available with Flash or Silverlight.

key html5 issues.jpg

OK, you say, sure HTML5 is behind, but because group standards are such a robust platform for feature development, HTML5 will certainly catch up quickly. That leads us to the next myth about HTML5.


Comments (35)

RazorX
Said this on 9-2-2010 At 02:55 pm

To add to this article. Did you know that only 2 out of the 7 demos found on Apple's HTML 5 showcase page actually used HTML5 http://apple.com/html5

It's amazing the levels of deception Apple and HTML 5 purits are willing to take to further their propoganda machine.

Said this on 9-2-2010 At 04:31 pm

Actually didn't; thanks for sharing.

Jan

Said this on 9-5-2010 At 11:12 am

Great article Jan, well done and very informative.

Said this on 9-5-2010 At 12:53 pm
Hey thanks Wade, and thanks for coming by.

Jan
Keaton
Said this on 9-7-2010 At 03:55 pm

Well written, and although clearly biased, still presented in a mostly "objective" statistics & data driven analysis. The bits on the HTML5 spec timeline I think are quite important. The worst thing with web development these days is having separate stylesheets for IE6/IE9/FF/Webkit/ etc etc .. now we will just have more layers of things that require various levels of falling back and customizations.

HTML5 fancy features aside, there is still a very large deficit with browsers and their rendering reliability of HTML/CSS. Everyone seems to avoid talking about the most common problems of web development are still right there, but with all the video codec hubub, they have gotten pushed to the side. Personally I wish video wasn't involved with HTML5, they should just focus on getting all browsers to be implemented the same way.

Said this on 9-7-2010 At 04:51 pm

Thanks .... I think. I admit that I get pretty wound up with how biased the favorable HTML5 articles are,  with little or no facts to support their claims. All I can say is that you should have seen the first draft of this article - it was all bias and no fact. 

:-)

I agree that video has consumed a signficantly disproportionate share of the focus about HTML5 - Flash is dead just grabs more eyeballs than HTML/CSS. Let it do the basics first, then worry about video.

Thanks for sharing your thoughts.

Jan

Said this on 9-8-2010 At 03:59 am

It's funny how any Apple Fanboi will jump on Steve Job's every word and talk about how CPU intensive that Flash is...

Well the latest Google Doodle uses only JS and CSS has been crashing computers around the world because of the CPU demands.  HTML5 is a progression, but it's not the entire answer.

http://www.zdnet.co.uk/blogs/jacks-blog-10017212/g...

Jan
Said this on 9-8-2010 At 07:26 am
Fun article, hadn't seen that. Interesting how the reality of these technologies seldom match the hype (you think we'd know by now), particularly early on. Everyone seemed to think that Flash would simply go away, and either not be replaced (no more flashing ads!) or be replaced by something far more efficient (because Adobe is so stupid and untalented). Few seem to consider that the replacement might be far more inefficient and potentially un-secure.

Thanks for sharing.

Jan
Said this on 9-8-2010 At 09:51 am

Jan,

Great article.  The hype machine has been rolling the past 6 months but the facts presented here speak for themselves.  It's reassuring to see the streaming community setting things straight. 

 

Thanks,


Mike

 

Said this on 9-8-2010 At 12:05 pm

Mike:

Thanks for the kind words. As you probably noticed, the hype machine has been irritating to me as well, hopefully, this will add some balance.

Thanks for stopping by and for your comment.

Jan

Said this on 9-8-2010 At 12:46 pm

Your analysis looks backward more than forward. Yes, the actual support and penetration of HTML5 video on the web is low at this point, but it's clearly the direction that the industry is going in. It's going to take a few years to get there, but I'd rather start experimenting with it now (I've got no other choice if I want to support iOS devices) than ignore it and have to make dramatic changes in order to keep up with architectural changes.

Said this on 9-8-2010 At 01:53 pm
Len:

Thanks for your note. I would say that my analysis looks at today, not yesterday. I agree, developers need to be HTML5 aware today, with a plan to evaluate if and when to cut over at some point.

Thanks again for weighing in.

Jan
Said this on 9-8-2010 At 05:59 pm

A few more myths would be.

1). Flash does not support touch. LIE!

2). Users lose browser controls like back button. LIE!

3). Not Seo friendly. LIE!

People use "flash amatuer" developers to base all of the negative on. Any one in the proffesional world knows that anything can be done with flash, and it can perform amazingly.

On a side note, looks like flash is getting native 3d support.

Said this on 9-9-2010 At 03:45 pm

Great additions, thanks for contributing.

Jan

David
Said this on 9-9-2010 At 04:27 am

One thing people need to think about, is the commitees that developed some of the older specs didn't have a financal interest involved. The current commitees are made up of people in tech companies with strong financal interests, all the decisions are weighed against the interests of those businesses which means they are slow and contentious, and often lacking in inovation.

Said this on 9-9-2010 At 11:11 am

Good point, and I totally agree.

Thanks for contributing.

Jan

Said this on 9-10-2010 At 06:17 am

A quick correction: there's no DRM in HTTP Live Streaming - or at least no DRM in the sense that anyone would recognise it. The closest approximation is (per Apple's documentation):

Media files containing stream segments may be individually encrypted. When encryption is employed, references to the corresponding key files appear in the index file so that the client can retrieve the keys for decryption.

When a key file is listed in the index file, the key file contains a cipher key that must be used to decrypt subsequent media files listed in the index file. Currently HTTP Live Streaming supports AES-128 encryption using 16-octet keys. The format of the key file is a packed array of these 16 octets in binary format.

This is in no way equivalent to the protections available when using WMRM, FairPlay or FlashAccess; instead it's more like asking the consumer "promise you'll be good".

Said this on 9-10-2010 At 09:38 am
Jon:

Thanks for the clarification and contribution.

Best.

Jan
Said this on 9-15-2010 At 09:10 am

As i've written in my blog post, HTML5 Video – It’s a long way ’til JQuery, I see this as analagous to the spread of standards-based Javascript/DOM/CSS in the early 2000's.  lf you wanted a great/custom/interactive user experience on the Web back then, you had to build it in Flash or even as a Java applet. HTML4 offered things like layers and a DOM model and CSS, but browser incompatibilities and differing implementations made actually building anything using that technology an expensive, fragile affair. It was a still for science-fair projects and proofs-of-concept, not production apps.

Eventually, libraries like JQuery came about, which were made possible by browsers coalescing around standards - both official and defacto.  That gave devleopers an abstraction layer that allows the terrific web UIs we see today.  Less work, better features, better user experience.   

At this point, HTML5 video is still in that science-fair stage. Sure, you can use it, and to support Apple devices you have to. But it still offers less functionality for more effort, and as such, remains a necessary distraction for many video publishers at this point.

One day, it will be "there" and we'll love it like we love JQuery. But until then, Flash and Silverlight will provide the best path for supporting the most users with the best experience and the least effort.

 

 

Said this on 9-15-2010 At 12:24 pm

Hey Larry:

Thanks for your insightful post. I would argue that when it's "there" Flash and Silverlight will be much further ahead, and there won't be there anymore. As I mentioned in a post on the StreamingMedia site,

"You seem to be saying that HTML5 can provide an effective lowest common denominator video playback experience. My point is that Adobe and Microsoft keep raising the floor as to what that lowest common denominator experience is, even for small corps."

But, you're closer to the firing line than I am, so your opinion has lots of weight with me, and I appreciate your sharing it.

Jan

 

 

Said this on 9-21-2010 At 12:40 am

This article defines the term "reality check," and blows away the smoke and mirrors. Thanks.

Said this on 9-21-2010 At 08:24 am
I'm blushing.

:-)

Thanks for stopping by and taking the time to write your kind words.

Jan
Said this on 9-22-2010 At 04:43 pm

Jan, I should have said the same thing about the work you and Tim did last January in your articles on adaptive bitrate streaming. They were timely and truly helpful for those of us trying to make sense of this environment and build practical systems. Tech journalism can't do better than that.

:: SH

Clement
Said this on 10-8-2010 At 08:25 am

I think it's wrong to focus only on the video tag, which is only a tiny fragment of the HTML5 specs. In my opinion, HTML5 can be really useful to developers as it adds many semantic tags that helps structuring your HTML documents. I'll probably won't be using the video and audio tags for the next 4 years or so but I'll surely use the nav, arctivle, aside and other ones that will make my HTML more readable for my co-workers. Also it will help web spiders indexing pages more naturally, as crowlers will probably start ignoring navigationnal content and focus on the core content of pages.

Said this on 10-10-2010 At 12:23 pm

Clement:

I agree, but virtually all of the positive press about HTML5 has been related to the video tag, starting with Apple's decision to not support Flash in the iPad.

Video or otherwise (of course), HTML5 won't really become practical until the installed base of HTML5 browsers gets much, much larger. I don't see that happending that soon.

Jan

Clement
Said this on 10-15-2010 At 05:28 am

You're right, the video tag does get most of the media coverage. The media (are they really knowledgable on the subject?) decided that highlighting this aspect would attract readers, mainly thanks to the hype around apple products and Job's refusal to integrate flash in iOS. I am kind of sad to see most of the attention of the develloper crowd focused on the video tag because of the media while other parts of HTML5 can be implemented right now! If you use a "aside" tag in your HTML it won't break the rendering in IE6 for instance.

Peter
Said this on 10-27-2010 At 05:55 am

Amazing eye-and-mind-opener article. A complement to Jan of course, but I also feel stupid not to have been critical enough to the no-to-flash-mantra of Steve Jobs. Our job is to deliver the best experience, if that goes well without Flash then go ahead! But based on many user engagement objectives, Flash does deliver a superior user experience IF developed by real Flash pros @truimagz. For instance, Flash supports alpha channels in video which can be interlaced seamlessly with content around it making a video sequence (like a moderator) an integral/interactive part of the experience http://active.tutsplus.com/tutorials/screencasts/g... . I don't know the specs of HTML5, so pls correct me if wrong, video is supposed to play in a box ... right?

Said this on 10-28-2010 At 12:07 pm
Thanks! I'm the wrong guy to ask about HTML5 capabilities, but I'm guessing that they go beyond video in a box. Not sure that HTML5 can simulate the experience that you're pointing too though.

Thanks for your thoughts!

Jan
AC
Said this on 11-19-2010 At 11:45 am

Thank you for this excellent summary! As someone who is in the middle of a year-long Flash development project, I can now just send your link to all the people in my world who are freaking out over iPhones and iPads. You are absolutely correct that most of the Flash detractors are ill-informed about browser/device/video technologies. 

Said this on 11-19-2010 At 12:06 pm

AC:

Glad to help; thanks for taking the time to write.

Jan

Said this on 11-20-2010 At 12:16 pm

People really need to read this article, both programmers and web owners. There was some excellent information posted in this blog article. Thanks.

Jan Ozer
Said this on 11-20-2010 At 12:24 pm
Thanks for the kind words.

Jan
MrVideo
Said this on 12-7-2010 At 03:24 pm

HLS may not be a full-fledged DRM solution however you can get pretty close.  Decryption keys should be unique for every video title and always delivered to the iOS client over a secure connection.  BTW, the "client" in this case is the native iOS player client which directy handles keys, decryption and playout.  Decryption key's for video playout are not handled by the browser and are therefore never exposed like they can be in a traditional web-based solution once they're transmitted.  Do you get playout protection with HLS... no.  So if this this is a non-starter for you then you don't really have a choice but to move to a full-blown DRM solution.  However, the security model can be made very tight and a no brainer if all you are securing is SD content.  Certainly worth a deeper dive when comparing the pains of implementing DRM and driving your customers crazy.  Content rights holders want DRM (i.e. studios), not customers.  There are good reasons for the consumer backlash against DRM and both complexity and infexibilty are among the biggest concerns.  Is HLS the Fort-Knox of security... no.  But you need to ask yourself if you really need that given the trade-offs you and your customers have to make with DRM.  Is HLS "enough" security... yes depending on what your protecting.  SD content in my opionion is a definite "yes".  Anything that minimizes that barrier for consumers should be seen as a good thing which makes HLS for iOS delivery a very viable option and one that should be seriously considered before moving to a full-fledged DRM solution.

Said this on 12-7-2010 At 04:47 pm

Hey Mr. Video:

Thanks for your post - I thought HLS was iDevice only - does it currently work on Windows/Mac? If so, which browsers?

Please let me know - this sounds interesting.

Thanks.

Said this on 11-29-2012 At 11:40 am
Hi,You can make flash animations as silmpe as a ball bouncing across the screen and as complicated as a timed, multi-layer animation using Adobe Flash software.
Post a Comment
* Your Name:
* Your Email:
(not publicly displayed)
Reply Notification:
Approval Notification:
Website:
* Security Image:
Security Image Generate new
Copy the numbers and letters from the security image:
* Message: