Firefox is somewhere around seventeen or eighteen percent depending on whose measurement you trust, and every quarter it takes another point or so off Internet Explorer. This is the number everyone reports. Safari sits at five or six and gets mentioned last, if at all.
I think the numbers are being counted correctly and read wrongly.
Firefox’s gains are real and they have done something valuable. Because enough people use it, a site that only works in Internet Explorer is now a site with a visible problem, and that has dragged an entire industry toward standards it had been ignoring for the better part of a decade. Tabs, an extension model, and a rendering engine that follows the specification are all things we now expect because Mozilla made not having them embarrassing.
But the desktop share table is a snapshot of a settled fight. The unsettled fight is about which engine ends up embedded in things that are not desktop browsers, and that one is not close.
WebKit is in Safari on the Mac, which is the five percent everyone counts. It is also in Safari on the iPhone, which is the browser people are actually using on a phone for the first time rather than tolerating. It is in Nokia’s browser on the S60 handsets, which ship in enormous numbers. Google shipped a phone platform announcement last autumn with WebKit inside it. Adobe put it in the runtime that lets web code run as a desktop application.
None of that shows up in a desktop share table, and all of it decides what a web developer can assume in three years.
The reason is licensing and shape rather than quality. WebKit is a library with a permissive licence that a manufacturer can embed without publishing their own work and without a negotiation. Gecko is a superb engine attached to an application, and pulling it out to put inside a set-top box or a handset has always been more work than anyone wants to do. That difference was a technical footnote for years. On phones it is the whole thing, because on phones the browser is a component someone else is shipping rather than a product a user chooses.
Safari 3.1 arrived this month with the animation and transition properties, plus video and audio elements that do not need a plugin. Nobody can use those yet in production because the other engines do not have them. That is not the point. The point is that the engine most likely to be embedded in the next generation of devices is the one setting the pace on what the web is allowed to do.
So the reasonable reading of the browser market in early 2008 is that Mozilla won the argument about standards and is winning the desktop, and that a smaller player has quietly taken the position that matters for everything after the desktop.
I would rather have five percent of the desktop and the inside of every interesting device than eighteen percent and the inside of nothing.