Fraud science used to promote Flash performance over web standards
March 11th, 2010
Daniel Eran Dilger
A report purporting to vindicate the performance of Adobe’s Flash plugin in comparison to open standards broke through the weak editorial barriers of the tech community yesterday. It’s wrong, here’s why.
The report was created by Jan Ozer, a proponent of Flash who makes his living selling books and seminars about Adobe’s technologies. The original article is even interrupted by an advertisement promoting Ozer’s “Streaming Production and Flash Delivery Workshop.”
After noting Ozer’s bias, one site commenting on it wrote, “we don’t think that [his bias] has any effect on the test outcomes [his report presented].”
The problem wasn’t that Ozer faked data to promote Flash; some of his findings actually indicate that even the early beta implementations of HTML5 beat the latest version of Flash in video playback tests. The real issue is that Ozer framed the debate around an absurd premise to shift the conversation from real issues to contrived garbage.
A press release of fake science
Coverage of Ozer’s press release uncritically reported his findings that certain browsers were no better (or at least not much better) at rendering video from YouTube via Google’s experimental HTML5/H.264 site than via the standard Flash version of YouTube.
Ozer detailed only the reported “CPU Utilization” for his test Mac running Safari, Chrome, and Firefox browsers, and a PC running the same three browsers in addition to Internet Explorer. He compared the performance of Flash 10 with the latest Flash 10.1, and contrasted HTML5 playback on browsers that supported that as an alternative to Flash, not too subtly suggesting that HTML5 and H.264 were riddled with problems that inspire fear, uncertainty and doubt, while Flash simply works everywhere.
However, his results made no comment on the visual quality of Google’s Flash vs raw H.264 implementations. Previous tests I performed indicate that Google’s beta version of YouTube running HTML5 delivers raw H.264 video with remarkably better picture equality compared to the HD version of its Flash video for the same file. You can see for yourself by viewing anything on YouTube in “HD quality” via both Flash and HTML5.
Additionally, Ozer seemed to gloss over the fact that his tests really say next to nothing about the efficiency and performance of the Flash runtime compared to the use of open standards, because he wasn’t testing Flash content rendering, but really only the playback of video data delivered via a Flash wrapper.
To deliver video, Flash really isn’t doing anything special. That’s why browsers supporting HTML5 can do this themselves without needing something like Flash (or its doppelgänger, Microsoft’s Silverlight).
HTML5 savvy browsers like Safari and Chrome can also animate content and even (with a little more work) do the kinds of fancy interactive apps and games that Flash was originally targeted toward, all using open web specifications.
The Flash problem
Flash is promoted by Adobe as being a great way to create everything from simple website navigation and interactive content to full-blown Rich Internet Apps. Using Flash is an alternative to using open web standards to build these types of content.
The problem is that when content creators built stuff using Flash, they’re locking up their code in a form that can only be rendered by Adobe’s sanctioned Flash Player plugin. Nobody else can create their own legitimate implementation of Adobe’s Flash Player because Flash isn’t an open specification. It’s a proprietary technology fully owned by Adobe.
That’s a problem for Apple because it wants web-based content to play back well on everything from the Mac to its iPhone platform. If content is created in Flash, that means Apple has to wait powerlessly for Adobe to fix the situation in its Flash Player plugin, something that Adobe (and Macromedia before it) consistently failed to deliver for the Mac platform over the last decade. Apple gave up on Flash in the mobile realm in part to hasten the development of open alternatives.
Flash content also forces Mozilla, Opera, and the other WebKit developers outside of Apple to similarly sit back and idly support Adobe’s poorly performing Flash platform in preference to independently optimizing the rendering of open web standards that were designed to scale better from desktops to mobile devices.
Adobe’s current mobile strategy has literally emerged just over the last year or two, largely in panicked response to the iPhone. Prior to that, Adobe was pushing the joke that is Flash Lite on mobile platforms, and a different version of Flash on PC desktops.
Ten Myth of Apple’s iPad: 2. iPad needs Adobe Flash
An Adobe Flash developer on why the iPad can’t use Flash
Ideological fraud science
Much like Microsoft and its new Windows Phone 7 initiative, Adobe is hoping everyone will forget that it has done an abysmal job in deploying appropriate mobile technology over the past decade, and has its fingers crossed that everyone will abandon the much better options that have become available over the past few years and instead turn back to subservient dependance upon refreshed version of 1990s monoculture instead.
Much of the ignorant tech media is actually cheering on this absurdity, which is a bit like right wingers hailing more Reganomic deregulation even as the economy fell into ruin due to misguided efforts at putting financial institutions (rather than the law) in charge of regulating themselves in the first place.
In both cases, ideologues are quick to leap upon the most ridiculous fraud science in order to support what they’ve been told they should shill. Ozer’s “report” on Flash conveniently ignores the real problems (which include both replacing the open web with a closed plugin architecture owned by Adobe, and Adobe’s terrible performance in building and delivering this).
Instead, he creates a strawman problem (suggesting that Apple is accusing Flash of being really bad at simply delivering H.264 video in comparison to open HTML5, and then attempting to show that’s not the case at all) while launching a conspiracy theory (that Apple is out to get Flash for malicious reasons) and a dramatic morality play (that Apple ought to instead work to invest its efforts into making Flash play slightly better, so Ozer can keep writing books about Flash for his captive audience of Flash creators and users).
The truth is that Flash is irrelevant in the future
What Ozer should do instead is present the plain truth that Flash is a terrible platform for creating web content because it violates everything the web was designed to do: openly share content using openly documented specifications that any vendor can implement in competitive ways that advance the state of the art in hyperlinked, multimedia communications. Flash smothers the web with closed binaries that require Adobe’s interest to play back.
Additionally, Ozer should stop presenting half-truths comparing Google’s currently experimental version of H.264 playback with its refined existing implementation of Flash. Ozer fails to admit that Flash isn’t primarily a video distribution system, nor that video playback is really where Flash really exhibits its “CPU hog” problem. Flash is an interactive content platform that rivals the open web. Trying to subtly suggest there is not really a problem with Flash is the opposite of being honest.
By presenting trickery in numbers, Ozer is playing the same role as climate change doubters: creating a distraction that lasts just long enough to turn the conversation away from meaningful changes and toward a false controversy that invents blame where none exists. In Ozer’s case, he deflects real criticism of the terrible performance of Flash (particularly on mobile devices) in order to shift the conversation to one that demonizes Apple for not rescuing Adobe from its own terrible implementation of its Flash platform.
Rather than encouraging developers to use open standards for creating interactive web content, and imploring Adobe to drop its dead end Flash runtime acquired at great cost from Macromedia and instead focus on creating tools for modern and open web standards, Ozer attempts to instead suggest that Apple is a bad company for not focusing most of its efforts into shoring up the performance of a fatally flawed web-alternative so that Adobe can serve as the sole beneficiary of all web development going forward, without actually doing anything but tainting the web with a proprietary binary trap.
Shame on you, Mr. Ozer. … and all of you in the tech media who gobbled up his fraud science while remarking how delicious it was.