Fragmentation
Building web apps is not getting easier. The fragmentation of operating systems and browsers is getting worse, not better.
Here's a chart of the past thirty days of activity at AVC.com:
No OS/browser combo has more than 17% share. And there are five with more than 10% share. iPhone is about 6% and iPad is about 4%. If you go down to the next ten combos, you find a number of Android and Blackberry combinations.
Mobile OS/browser combos in total add up to about 15% of all visits and that number is up from less than 5% a year ago.
Add in the need to build mobile apps for iOS, Android, and possibly Blackberry and you've got quite a difficult environment for developers these days.
Comments (Archived):
html 5! We have stopped all development for iphone and Android apps. At the end of the day software drives the type of devise, not the other way around.
It’s a great point, Fred. My developers are spending too much time browser testing when they could be building out new features from our ever-expanding iteration list.I think we’re fortunate at http://gorankem.com to have a fairly tech-savvy crowd so we don’t really waste our time catering to IE 6/7 browsers. Is it really worth slowing us down to please 2% of our visitors?Unless you’re a widely popular site that caters to everyone (i.e. CNN, Yahoo, etc), I don’t think there’s many that should waste their time on browser testing when they could be working towards changing the world. That’s just silly!On a related note, it was great (and amusing) to see a movement like http://www.ie6nomore.com/ build some steam earlier this year 🙂
I should have added that especially when you’re a bootstrapped startup and only have so much time/resources, your time should be spent making a stronger offerring. Like you said, it’s a shame that we have to please all of that fragmentation. Is there any easy solution?
it’s going to be hard to drop ie6 (unfortunately) apparently there are a lot of third party applications that were custom built on ie6, and there is very little in the way of easily migrating.That’s a shame.
Some of the sites I support still see around 12% of visits coming from IE6 (mostly from larger companies I suspect). It’s frustrating to support a 9 year old browser, but hard to ignore those visits when they are potential customers.
By total coincidence I started this thread on Hacker News asking whether the time was right to go exclusively HTML5 on mobile: http://news.ycombinator.com…
It depends. If you want great user experience (like in games, like in flash/silverlight style visualizations, offline usage) you can’t go the html5 route, it’s not enough, you don’t have that control of the visual/audio space.Even html5 is not really supported in mobile phones, for example the contentEditable and designMode is not working rightly in iOS or Android or Windows Phone 7, and that’s necessary for WYSIWYG editing.For everything else html5 is a good idea.
I partially agree. For some really cycle-intensive use cases native will be the right choice for a long time yet.But the full HTML5 spec already includes offline support and powerful local caching – and the combination of the video and canvas tags with WebGL is already spawning a ton of really impressive real-time 3D demos – for example the Google Aquarium.
Do you think that (at least to some extent) the fragmentation of OS and browser combinations is a good thing and a sign of a competitive market? I love seeing the emergence of new viable players because it forces product innovation. Point taken, though – it certainly does pose a problem for developers.
All the more reason why I believe that with the advent of 4G networks coming…more and more of an apps functionality will take place in the clouds, with only the proper UI rendering being done on a phone.
You can’t render a GPS receiver or accelerometer or camera sensor in the cloud. Now lets think about processing in the cloud for 50 million users for image scaling or video encode…yikes! Sure you can do it but offloading that processing to perfectly capable devices is a huge bandwidth and processing saver (maybe a power saver too …but I dunno).Full disclosure – In my day job I work for a company that makes microprocessors – so I am drinking some of the koolaid ;). But we do play on both sides of the client / server compute fence.
Will somebody tell Wired to exhume the browser’s body?
Chris Anderson is getting pretty ridiculous with that stuff. I have no idea on what evidence he bases his conclusions.Here is Pew research that indicates that only 35 percent of adults even download apps. Only two-thirds of those people actually use their apps:http://pewresearch.org/pubs…One in ten didn’t even realize there WERE apps on their phone.I wonder if he’s even seen how complicated it is to create an app for multiple phones. When you consider there are platforms for iPhone, Rim, and Windows, and then you consider that only 25 percent of cell phone users apps at all, it’s just not cost effective for most businesses to have apps.All I can figure is that he toots his app horn so we all download the Wired app and they don’t have to waste as much money on print.
Ah journalists. More important to move volume than to be right.
Business Insider had an interesting chart of the day yesterday to support this same point … Safari is the most important app on the ipad: http://ow.ly/3atsz
Firefox and Chrome so close . Sign of things to come
Would you prefer if it was 75% IE6? How did startups with a big bet on ActiveX do? This is a success of standardization, celebrate it.
Could also just as easily say there’s lots of opportunity. One of my least favorite things is clicking a link in an email on my phone and being routed to either a non-mobile site or the mobile gateway (not the actual article/link I wanted). Someone please fix it and sell the solution everywhere!
Fred,While the graph shows that, yes, people are using different browsers and platforms, you’re omitting one crucial fact – they are all incredibly similar. Of the ten browsers, you have five which use webkit and four Mozilla browsers (which one can assume all use gecko, their rendering engine). So while the browser shell may be different, the viewing and interactive experiences are nearly identical. But if you’re building browser plugins… Well then yes, it’s not easy sailing.
Excellent point.
Yes, not only do we have common rendering and javascript engines, but we also have great frameworks and libraries that take care of many of the browser differences, and graceful degradation when needed.Oddball versions, by the way, are biggest pain in the neck in my opinion. You can choose to support only modern browsers (no IE 6, etc..), but every now and then you get quirks in something like FF 3.6.11, that inexplicably break things. And even though the version may not have been live for very long, people get it installed and stick with it. Thankfully, auto-updates have helped this a bunch.
True, but those are edge cases. The differences are miniscule compared to the differences between a Webkit experience and IE.
Isn’t there a quote from The Matrix which contains the line “the sound of inevitability” ?That’s how I think about HTML 5 – the browser as a unified platform.To me it’s entirely obvious – and it’s just a question of time before it becomes reality.
It’s not quite as fragmented as Fred alludes to. But it’s still getting worse.
I write browser plugins.. and it is pretty tough sailing 🙂 The only saving attribute is that “netscape plugins” run in almost all the non-IE based browsers on Windows, Linux and Mac. Of course you have to compile for 3 operating systems. Browser plugins on devices like iphone,ipad, android etc are another issue all together – I don’t have a good solution there yet. For MeeGo we can use regular netscape plugins…
sldfjsdf
This is a good point. Let’s see a graph that compares THAT data….
Problem? This is opportunity. We’re in the early days of mobile becoming the primary computing platform most people “use”. As in, touch a computer with their hands. Early days, folks.
For developers, the classification of browsers is really between modern browsers and non-modern.There are tools like SproutCore for eg which is an HTML5 app framework that allows you to develop for any modern browsers without plugins. Those with that experience can speak-up if it has indeed made it easier for them.
I’m not sure what to think of it, but what I do know is that there are too many people hoping that HTML 5 is the answer to all of our problems. Truth is, it probably isn’t. There are just too many new devices being released that extended our use of the web in ways that we can’t enhance enough just using HTML. I think we need a platform that lets us create true device agnostic applications. Until we have that, I don’t think there is anything we can do. It’s the price we have to pay to be innovative.
As long as you can access the camera, GPS, touch interface, accelerometer and (perhaps) the microphone from within Javascript you should be okay for quite some time to come….
That’s a good point David – especially for well understood peripherals that have well understand interfaces and are widely used in device designs.The problem is that if I want to access the new piece of hardware that I’ve added to my new device design there is no convenient way to do it from JavaScript …I end up having to modify the browser, write browser plugins, or using something like the QT WebKit Bridge to go from JavaScript to hardware.
I’d say as a one-man startup developer things have gotten much easier. I develop for Chrome, do a quick once over in Firefox, and completely ignore IE unless someone has a problem. All my bugs recently have actually come from Safari’s slightly stricter security model. The decline of IE and the ascendancy of webkit/mozilla has made things far easier for me. And the fact that people are used to the idea of using a different browser has made them more lenient towards small bugs. We don’t have users expecting pixel-perfect designs in their browsers. As long as they can make it work they’re fine with it.
With all due respect, ignoring IE is insane.And I’ve run into Chrome/Safari incompatibilities – despite the fact that they both run WebKit.
Tim – I’m right there with you – our order is similar: Chrome, FF, Qt WebKit (for embedding in apps), Safari (mac,iphone,ipad), IE8+The webkit part is a huge value ..but there’s even fragmentation there..
Its still a good thing prevents browser makers from trying to combine tool bars and ads into it if it was monolithic, developers have to suffer for our sakes to some degree.
Just have a look at http://www.html5rocks.com/ and the html5 games slides by @mahemoff http://prez.mahemoff.com/ki…This is HTML5 and it works now!
Simple solution that I undertook was simply to decide to focus on a “deep” engagement and thus rule out mobile as being a very important platform – that basically gives license to focus on the desktop experience for the first 6 months – 1 year.
By the by, who is viewing A VC in IE?
IE 6 is a leading browser in China .
Hmm. Makes sense. Part of me is still shocked that members of the techcommunity use IE6.
Entrepreneurs with day jobs in corporate America. 🙂
Haha. Ok makes sense now!
…or the opposite – focus on mobile for the first 6 – 12 🙂
Yeah. Absolutely. Either way you need focus – its definitely paid offfor guys like GroupMe, which went so explicitly mobile, they designedtheir product to be used from any phone- ie, not just smart phones.
I would also point out that at least we are talking about only 3 rendering engines. I know, that doesn’t help too much but at least you can expect browsers sharing the same engine to render pages somewhere similar.
It will become like the tea or chocolate sections in whole foods, too many choices to make a selection.
*snicker* my vices too
Fragmentation breeds relevance. I know an awful lot about those segments just from that browser/OS info. Dynamic web FTW!
I will say that I see more device fragmentation than I see render engine fragmentation. And that doesn’t bother me so much- it points to a future where the web of what we see on our device is how we are using the device of the moment. So if you are building something that is a high concentration activity and requires lots of forms (like umm, a tax helper website?) don’t focus on your mobile. If you build something that requires a gps (ummm, something to do with food…) assume you are most likely looking at a mobile.Build out from there.
and thats without touching on the fragmentation within the fragmentation (ex. android). fragmentation is a royal pain for development, but is a very healthy sign for the ecosystem as a whole. as in biology…diversity = fitness
Anyone who believes there is such a thing as a “common rendering engine” in browsers today clearly hasn’t developed a truly cross browser compatible app. The differences alone between IE6, IE7 and IE8 are unbelievably huge (not to mention the incremental versions of each). With their insistence on continuing to build horrible browsers, Microsoft is destroying developer productivity at a break-neck pace. They should be ashamed of themselves.Trying to make JavaScript run across all of these browsers is even more difficult than just getting something to render properly and look pretty. You think HTML5 will fix all this? Not likely.Unfortunately, Fred is spot on with this post and if anything is understating the problem. The good news? With fragmentation and pain comes opportunity. We are seeing lots of cross-browser / cross-platform developer utilities pop up. Most still don’t work (even the companies that do this for a living can’t figure out the massive fragmentation), but they are getting there. As an entrepreneur trying to build real apps across this fragmented landscape – I’m anxious to see some of these startup companies succeed and take advantage of this opportunity.
Perhaps there would be less use or need for IE browser if more developers stop writing compatible plugins.
it is even worse it you start thinking about shipping your content to native apps over different ecosystems …
Thanks for this post, Fred. It’s good to know that there’s someone out there who knows, and cares.The three follow-up questions are: how did it get this way, and how do we deal with it now, and how do we make it better.One fundamental question which isn’t clearly answered in my mind is: what do companies get for creating a new browser? I mean, they don’t charge for it, they don’t advertise on it (except very indirectly via search box defaults and default home page). What’s the motivation?
The browsers are becoming more standards compliant to the point where the biggest differences are now mainly polish like css gradients and rounded corners. Many of those features can be used anyway if they degrade well (like rounded corners). This saves developers a massive amount of time. I think the new time-sink for developers is developing different versions for mobile. Many of us have finally become pros at dealing with IE6 right around the time it’s becoming irrelevant. Sadly, we know more about fixing the IE6 float issues than how to optimize a page to load fast on an iphone.
yup – if HTML5 video tag will work on the user’s device ..or when embedding webkit into an app.
Even further to your main point: I see plenty of fragmentation even within Android itself. I have the new Samsung Galaxy Tab(let) here in front of me and a Galaxy S phone, and can already see where this will be a case of multiple dev tracks and extensive testbeds.”difficult environment for developers these days”– sure, but it’s generating more business for remote development shops, much as a hurricane boosts work for building contractors !
Web apps are a tough business. Mobile apps are a brutal business. Prompted by this post, though, I checked my iTunes account and saw that someone downloaded the Portfolio Armor iPhone app yesterday for $28.99. A few hundred more of those and I could have a positive ROI on the mobile app.
How many types of automobiles are there? For the most part they all work on roads.Protocols and standards are the roads to information. Browser are the vehicles.
This is an issue.
Two words: web standards.I welcome this fragmentation because it hopefully will put a nail in the coffin of walled garden mobile applications.If we had had fragmentation during the browser wars, we wouldn’t have the mess that is IE6.
As long as less people are using Internet Explorer… I’m happy.
A perfect illustration why standards are important.
early standards kill innovation (Ask wireless carriers) ….thus it is important after a certain maturity.
We market to older consumers with an average age of 47 (you know, the one’s with money) and our statistics are RADICALLY different than yours. Internet Explorer is by far the dominant browser and 15 percent of our total traffic comes from old versions of IE. The amount of bad CSS on the Internet is mind-blowing. I can show you websites that have a 0% conversion rate on old IE because the website is unreadable.
if you serve only a segment of the market, you could probably figure out the most popular combos and focus on a few. Tough if you service a wide range like a news outlet!
This just mean there is a need for a “rendering layer” in software development. No more graphical design, logic and DB only but a 4th layer in charge of detecting the right OS, right form factor, right app store -for calling the right libraries. There is nothing on such that cannot be automated here…I think guys at Grapple have “a proprietary” solution to this …at least they claim to. Fragmentation is not slowing down anyhow. but Software development may need to evolve accordingly…by adding a new, important layer.
Fred…worse?? Not IMHO.More browser fragmentation = more competition. More browser fragmentation = more innovation. More browser fragmentation = more need for more developers and thus more jobs. I like it when Gates, Inc. has to work a little harder.
I agree I was going to reply yesterday with a big waaaaa (make a gesture with your fingers playing the violin or the non-PC version when you say that), but thought that was too negative.Seriously development has always been and will always be hard, the parts that get made easier provide the ability to work on new hard stuff. And that’s good. Keeps the riff-raff out.
Difficult? Maybe a bit more work, true.But would we rather live in a world where Apple, or MSFT, or GOOG, or anyone, owned 80% share?THAT would be difficult.
YupThat’s the overwhelming consensus of this community
Our developers want to remake one of our products for Android.I said no just because it’s too much to manage.
I would probably visit more from my iPad but I hate the way Disqus displays comments on it. It defaults to the mobile site which slows down my ‘discussion consumption’ because I have to drill down into each comment tree separately, I’d much rather the desktop view on a device with a screen as large as the iPad.
I agreeI will make that point to the Disqus team
Awesome. Thanks Fred.
If disqus worked better I would always read this blog on a mobile device.
I tend to disagree with the conclusion. Yes, the number of OS/Browser combos are greater than ever, however, they are all measuring themselves over the same benchmark.For example, in the Netscape 4.7 days, it was very hard to do a “rich” application on Netscape and IE 5.x, because Netscape was so behind and broken.Nowadays, HTML4/CSS2 is relatively well supported, and most modern browsers (especially in new mobile/tablet devices) implement a lot of CSS3 and HTML5 and even more (i.e. SVG). So, at the end, there are a strong HTML4/CSS2 base on all browsers, and good start on CSS3/HTML5. Also, most of the browsers are WebKit based which help a lot as well.I personally built lot of Rich Enterprise and Mobile applications using HTML (HTML4 and HTML5) technology, and I can tell you it is becoming much easier than before, and it is the only technology which offers true cross platform.Now, I am not denying that there are a cost, but I think the cost/benefit is getting lower and especially when you consider that there are no alternative.
With limited resources pick the biggest platform best suited for your app and start there. We don’t do apps but we do do eBooks and we’re doing Kindle-only for now because there are Kindle apps for all the platforms. When another platform becomes viable financially we’ll add it.
As long as those mobile OS/browser combos remain using a Webkit based browser then I do not fear for web applications as long as strong standards are adopted. I believe in the one-web approach. A solid application that displays well on all Yahoo A-Grade browsers. I recently launched a major campaign using all HTML5 including video. Despite some minor glitches in some of the video player renderings of controls, all functionality of the application including video works great.Build one application.
Don’t forget the multitude of technologies under the hood of these browsers and platforms.ASP.Net, PHP, Ruby on Rails, ColdFusion, JAVAJavascript, AJAX, Flex, jQueryThe list of interacting technologies never ends.Brett Millerhttp://www.customsoftwareby…
Don’t forget the multitude of technologies under the hood of these browsers and platforms.ASP.Net, PHP, Ruby on Rails, ColdFusion, JAVAJavascript, AJAX, Flex, jQueryThe list of interacting technologies never ends.Brett Millerhttp://www.customsoftwareby…
hey, whats the nice information you posted. God Luck for the future.
Fred made an excellent point, addressing the platform/OS fragmentation/compatibility issue, and correspondingly, software portability. He is factually correct in regards to the platforms/OS: there is indeed an excessive fragmentation in this area.Things are lot better in regards to the Web Browsers compatibility. Major Browsers (IE, Mozilla, Chrome, Safari) with market share of some 98% mostly adhere to the HTML/CSS/JavaScript open standards, thus Web Applications implemented as “thin clients” running in the Browsers demonstrate almost universal portability. In other words, Web Browsers are increasingly becoming the “virtual OS”, making the differences in actual platforms/OS less important (at least in web app domain).As an example, online “Inflation Calculator” was created as educational RIA (Rich Internet Application) project, demonstrating the power of emerging Internet standards: HTML 5, CSS 3 accompanied by jQuery: It’s 100% compatible with latest Firefox (Mozilla) and Chrome/Safari (webkit), capable of running on any desktop/notebook/mobile platforms (it should be presumably compatible w/IE9, either).More on this topic on Examiner: 1). HTML 5, CSS 3 and Inflation Calculator (http://www.examiner.com/onl… 2). How to select web browser and check its capabilities (http://www.examiner.com/onl…
Wow, that’s about the most balanced browser/os analytics I’ve ever seen. Most of my sites currently have about 40-45% Firefox and 25-30% Chrome with IE making up less than 5% and as low as 1.6% on one of my sites. I even have a site that, in two years of being live, has never been viewed in IE6. As far as I’ve heard the “actual” browser share is quite a bit different–IE is still up near 50%, I believe they JUST dropped below. What the whole world uses is mostly irrelevant though, what’s important is what your market uses.
It’s not just fragmentation of the OS that’s causing complications, what about all the different development languages (client and server side), databases, and CMS Systems?There are hundreds of variations if not thousands.Brett Millerhttp://www.customsoftwareby…
Wow! I had no idea.
I think the bigger picture is that it’s getting better.Really the worst offender in the last decade has been Internet Explorer, and we’re finally seeing the end of the pain it caused, reworking everything twice to accomodate Microsoft’s love-hate for the web.We now have Apple’s love-hate for the web, but it’s at least justified with a different form factor, it’s not an artificial requirement when you have to design for the iPhone, a good UI designed specifically for mobile is necessary, not an unreasonable demand.Both the current iPhone browser and the current Android browser are Webkit-based, so it’s really a non-issue, or very-close-to-non-issue on the mobile web front for the most part. There’s also that most developers use jquery, yui, dojo, iui… that kind of thing these days, which abstracts a great deal of the implementation details for specific browsers.I think we’re in a much better place than we were 5 years ago, babysitting Internet Explorer for zero gain or worse.
Thanks dear– this was a great post !!!Also Very informative.