Zuckerberg: Betting On HTML5 Was Facebook's Biggest Mistake 290
An anonymous reader writes "Speaking yesterday at TechCrunch Disrupt, Facebook CEO Mark Zuckerberg acknowledged that the company's stock performance was disappointing. He also made an interesting remark about Facebook's development efforts over the past couple of years: 'The biggest mistake we made as a company was betting too much on HTML5 as opposed to native. It just wasn't ready.' According to Mashable, 'the benefits of cross-platform development weren't enough to outweigh the downsides of HTML5, which pulls in data much more slowly than native code, and is much less stable. ... Now, Zuckerberg says, Facebook is focused on continuing to improve the native mobile experience on iOS, as well as bringing a native app to Android.'"
Facebook never bet on HTML5 (Score:3, Informative)
http://www.readwriteweb.com/mobile/2011/09/how-facebook-mobile-was-design.php [readwriteweb.com]
TL;DR:
If you think that's an HTML5 approach, I have some lean agile behavior-driven coaching hours for which I'd like to bill you.
Re:I don't give a Zuck! (Score:5, Informative)
The problem is more complex than that: Mobile devices lack CPU grunt to do things which are easier to do on a desktop systems.
Because of this the mobile OS builders concentrate what little CPU they do have to make sure their apps run the best as they can at the cost of anything else you may wish to run on top of that. In fact I think they even cripple Javascript on iOS to make sure the OS keeps ticking nicely, for example native scroll events take precendence over Javascript scroll events. I think the main reason that flash was killed in iOS was because it was a closed source CPU hog that they couldn't cripple.
The only thing that will change this for mobile development is more CPU power, which is difficult if we don't want to have personal hand warmers in our pockets.
I don't have a problem with JS for application GUI development as long as there is enough juice to run it.
I suppose that *is* a problem, but really the big thing that Facebook has screwed up in mobile is not having the infrastructure (server side) to push all content as updates to the app. Instead, each time a user wants to browse their wall, they have to download the whole flogging thing again. The absolute biggest threat to mobile experience is the actual content download itself, it requires the user to stand around and wait, and it eats battery like crazy. Twitter got this right, partly because that's the entire model of their service, but if you look at how well their app runs on mobile you kind of get tired of even tolerating Facebook at all.
Re:I'd second that. He's spot on with this. (Score:5, Informative)
From the actual quote [tobie.me], it does not sound like Zuckerberg is really down on HTML5 overall. I think, rather, he is saying that the company invested too much time trying to optimize the HTML5 client for mobile clients, when the company was ultimately able to get better performance with less effort by developing native apps.
Re:I don't give a Zuck! (Score:5, Informative)
Just because HTML5 might have wrinkles to iron out doesn't mean that it's a failed endeavor. Rather, it means that the browsers, companies behind said browsers, and the users have created a massive cluster of epic proportions.
So, basically, blame everyone but the people who wrote the spec? Sounds like the consortium's made up of entire middle-managers.
HTML5 is the poster-child for designed-by-committee, slow-as-molasses processes that are out-paced by everyone else because, in the real world, things actually need to get done this decade, and the rest of us can't wait. HTML5 has been in development for eight years, and their current target is another two years before it becomes a Candidate Recommendation. Bearing in mind that they've already missed all their previous targets, Ian Hickson estimated that they'd have the requisite two, independent working implementations in 2022. That's eighteen years from start of development, 10 years from now.
By the time the spec is completed, devices will have been forced to roll their own solutions, simply because the spec isn't done. Now, they might have some inter-operable features, if that aspect of the spec had been fully codified before they had to implement it, but that's precisely the situation we had in the Netscape v. IE browser wars - each had a somewhat common base, but were independently adding new features to try an improve the browser. The features they added were mutually incompatible because there was no common standard - and we're staring straight down that road again. It's a very clear example of perfect being the enemy of good.
Re:HTML 5 Java (Score:5, Informative)
"HTML5 is roughly equivalent to Java as far as a multi-platform programming language and development platform."
No, not in the slightest. Not even close
"The only successful approach I've ever encountered to using a virtual machine was employed by the Digitalk VM which cached successive VM invocations so that you ran at native 'raw iron" machine speeds after encountering the performance hit the first and only time an pseudo-instruction was executed in a method.".
When did you last read anything about the JVM? 1995?
"The lethal performance problems that WordPerfect encountered trying to implement their suite of office products in Java still apply."
No, no they don't. That was the best part of a decade before Hotspot even came along, which was basically a complete rewrite.
You could've typed your post about 15 years ago, and you might've had a point. Now however, your post makes absolutely no sense, and shows an understanding that only someone who had literally been living under a rock for 15 years would have. Java has changed a lot since 1997, and your criticism is nonsensical in the context of those changes.
Re:Correction... (Score:5, Informative)
Re:Correction... (Score:5, Informative)
If he lives in California and the divorce proceedings are started there that prenup is worthless after 5 years.
Under California Law, she's not entitled to property he owned prior to getting married (gee, I wonder why he waited until the day *after* the IPO to get married!?). She'd be entitled to any gains the stock made after they were married, but she's probably in for a long wait before the stock rises to meet the IPO price again.
MySpace was shit, even compared to Facebook (Score:5, Informative)
You badly, badly misremember the awfulness of MySpace. Please don't make me post this more than once, the memories are painful...
Embedded Flash objects in the page. Lots of them. All set to auto-play when the page loads. Facebook doesn't allow embedding arbitrary content, and doesn't allow auto-playing video on your page either.
Incredibly atrocious CSS, like text that ballooned to 40pt on hover or that was in incredibly unreadable fonts, or covered up / replaced navigation links on the page... Facebook doesn't allow custom styling.
I'm not sure if this is the fault of ColdFusion or just of MySpace programmers being incredibly shitty, but every 5-10 navigations on MySpace would usually result in a server error. Sometimes, you'd get a server error when the server tried to serve the error page! Facebook has had occasional stability issues, and PHP is lame (but then, apparently very little of their backend is still PHP), but it's rock-solid by comparison.
Back when MySpace was hemmorhaging users to Facebook, there was a limit on the number of pictures you could host on MySpace. Considering that one of the main uses of Facebook for some people seems to be "host every single picture my phone can take" you can see why this appeals.
Strange though it may be to think of Facebook and security together, they beat the pants off MySpace, which has such glamorous characteristics as being the first site to host an in-the-wild XSS worm (because it was trivial to inject script into your page, and somebody figured out how to exploit that).