Rich’s April Fools’ post from last week has generated some discussion around how we (and developers in general) handle support for Internet Explorer 6.
For better or worse, the AppsLab team and many others are now choosing to leave IE6 outside of the realm of supported web browsers. While I don’t agree with that choice, I certainly understand it.
The problem with these types of discussions is that they inevitably get tangled with opinions and emotions. Rich and I have a well-documented distaste for Windows, and it’s difficult to take what we say about IE without hearing through that filter. I’ll try to document the facts first before injecting my opinions.
Internet Explorer 6 Facts
Shortly after its release on August 27, 2001 and for several years, IE6 was the dominant browser, surpassing 95% market share at its height. Even today, more than 2,700 days after it debuted, it still commands about 18% of the market.
During those years of domination, web applications were built and tested almost exclusively on IE6 because it was the de facto standard. Many enterprise web apps never tested non-Windows/IE configurations because they just didn’t have any within their organizations.
IE6 even dominated its predecessor, IE7 for several years, primarily due to a lack of enterprise adoption of both IE7 and Vista. As a whole, the IE family, which now includes IE8, has seen its market share decline precipitously over the last 18 months, most recently falling to just under 67%, a number IE hasn’t seen since the late 90s.
By the way, market share statistics come from Net Applications. There are other calculations out there that vary slightly, but usually no more than +/- a few percentage points.
New Browser Facts
Over the last few years, and accelerating over the last year, new browsers have been coming out frequently. Obviously, Mozilla Firefox, now in its third production version, has led the charge and now controls 22% of the market.
New entrants, like Google’s Chrome released September 2, 2008, and major updates to existing browsers like Safari 4, show the willingness of users to try a new browser, even when one comes with their Windows distribution.
Chrome already commands more market share than Opera, which has been making browsers since 1996, and its market penetration is constrained by its lack of OS X and Linux versions (official Google versions anyway).
Safari, bolstered by the market share OS X has taken from Windows, has grown to more than 8%. Having a Windows version definitely helps its adoption too.
Most importantly, these new entrants all support standards that allow developers to come closer to a write-once web app, i.e. you can reasonably expect that a web app will look and behave about the same on Firefox and Chrome/Safari (both WebKit browsers).
Sure, there are exceptions, but for the most part, that’s correct.
Not so with that same web app and IEx.
Insert Opinions Here
I’ve always thought browser market share was a funny concept, considering that multiple browsers can be installed on a machine. But, having been in that situation for many years, I guess I agree that the majority of people use a single browser for most browsing, making it easy enough to make the market share leap.
Granted, Rich and I aren’t fans of Windows or IE. I like choice.
When I returned to Oracle in 2001, everyone ran IE6. I remember distinctly a chat I had in 2002 with David about Netscape 6, which we both used. Compared to IE6, Netscape 6 had a lot of great features, like tabs. I’ve been with the Gecko family ever since, jumping to Firefox permanently in 0.7.
So, I’m definitely biased against IE6.
That said, I would do more to support it if I had the resources.
There are two important pieces here: a) what is support and b) resources.
What is support?
We (AppsLab) do support IE6, in that our web apps (Connect and formerly Mix) run pretty much as designed in that browser. Some stuff looks janky, but the app does what’s expected for the most part.
Until recently, we did make changes for IE6 users to make the experience less jarring.
Why? Because as it is on the ‘tubes, IE6 usage is dropped like a stone. IE currently accounts for only 29% of Connect traffic, of which 9% is IE6. This is definitely down from August 2007, when we began.
Back then, IE traffic was much higher, closer to 40% for all IE versions, and 25% for IE6. As recently as December 2008, IE6 accounted for 14% of the traffic.
As a new app, we needed to support the most common configurations, Win+FF and Win+IE, and we did make allowances for IE6-specific behavior because a much larger contingent of user base used it. Not so anymore.
You could argue this is a chicken-egg problem, since Connect looks better in standards compliant browsers like Firefox, but you could just as easily argue that the overall trend of IE slippage has more to do with choice in browsers and more awareness that other options exist.
For example, our internal IT deploys Firefox on new machines, and the WebKit browsers (Chrome and Safari) account for nearly as much traffic as IE6 now.
Our support for IE6 has declined as its usage has declined. You can read into that any number of ways.
This is the key driver for our IE6 stance. I have to balance investment in new features and bug fixing, just like any product manager.
I can’t spend development time on a browser that accounts for 8% (and falling) of traffic. Instead, I invest in new features and fix bugs that affect all users. Faced with non-critical bugs on the WebKit browsers vs. IE6, I would invest in Chrome/Safari first because they are growing user segments, whereas IE6 is falling.
Note that I say “non-critical”. If IE were completely borked and didn’t work, I would considering investing enough to get it back to its current state, which is pretty much working. But I’d have to take a close look at the effort required.
Despite how it might seem, we haven’t turned our backs on IE6 users completely. When we made the decision to stop fixing IE6 specific issues, we put up a message that lets users know why and where they can find other browsers, including those officially supported by our IT.
At the end of the day, I think we do a good enough job supporting IE6, based on our resources and our user demographics. Our support for it has adjusted as its use has declined.
Even so, we still enjoy making fun of it.