The browser wars are here (again)

Whether we like it or not, MSIE is, by far the largest browser. And you will need to check your results in that browser. But which browser should be considered the #2 browser? Netscape based on the version 4 core? Gecko-based browsers? Does the server logs, and fancy log-analysing tools tell us the truth, the whole truth and nothing but the truth?

First off, let me state that what you’re about to read isn’t here to discredit any browser. It’s not here to consent you as the developer in being sloppy when you develop your site. It’s not here to lighten your workload. If anything, it is here to tell you that you actually need to make more of an effort in following standards. It’s here to hopefully make you test your results in more than one or two browsers.

Analysing server logs

Yes, server logs are indeed useful for a number of things. They tell you where your visitors come from, they tell you where they enter your site, where they leave your site. Some log analysis software also tell you how long your visitors stayed, and which path they took through your site. In short, they give you, the site’s owner, a bunch of useful information.

Among the things log analysis software tells you , is which browsers your visitors use. Or does it really? When you view your logs, or thecounter.com’s browser statistics , and find that the most popular version of Netscape that is used, is Netscape 5.0, you should become a just a tad suspicious.

Netscape 5.0 was, as far as I can tell never an officially released product. It was a near-mythical product that was the precursor to what we today know as Mozilla.

What does this tell us? That there is an army of Netscape engineers manically surfing the web with their own, super-secret and private browser? Hardly. But, it gives a tell-tale sign that the methods log analysis software uses to identify browsers are flawed.

Log analysers fail for one simple reason: The User-Agent-string can, and is being faked. The Opera browser has the ability to identify itself as one five different browser. All that’s needed is a couple of mouse-clicks. By default, the Opera browser attempts to identify as MSIE.

Opera isn’t the only browser to do so. Mozilla can also spoof the User-Agent-string with an add-on.

So, my claim is that server logs aren’t always a reliable means to determine which browsers people use. It gives us an indication, but never the complete picture.

Counting downloads

Instead of just using server logs, I’d also suggest that you take the download popularity of a browser into consideration when you choose which browsers to test in.

You might argue that it’s difficult to find reliable figures for the number of times a browser has been downloaded. And I might even agree, to an extent. Why should we trust "57 billion downloads and counting!" on a browser vendors’ site?

But, by checking the figures from a large download site like download.com, that hosts many of the competing browsers, we might get an indication on how popular the different browsers are.

Browser popularity measured at download.com
Browser core Downloads
Gecko 26405
Opera 6 26690
Opera 7 3183
Internet Explorer 6 35677
Internet Explorer 5.5 4091
Internet Explorer 5.01 4463
Netscape 4.x 15115

The above table was sampled on January 19. 2003, and shows the number of times each browser technology was downloaded the week before, as per download.com’s "Last Week" option.

The Gecko-label is the sum of Mozilla, K-Meleon, and Netscape (versions 6.2.3 and 7.0.1). The Opera 6 label is the 6.05 versions of Opera both with and without the JRE. MSIE should be self-explainatory. The Netscape 4.x label is all Netscape versions available that has a version number starting with 4.

I had always taken for granted that Netscape 6+/Mozilla and other browsers based on the Gecko engine was the #2 browser, after MSIE. After having seen these figures, I’m not so sure anymore.

I think it’s hard to ignore the apparent position Opera has gained. It is also equally hard to ignore the fact that there is a fourth major rendering engine emerging. Apple is already a strong brandname, and when they launched their KHTML-based Safari browser, they got much attention. While Safari and Konqueror might be playing catch-up with Opera and Mozilla, I don’t think it will take too long before they too can participate on even terms.

So my statement is this: The browser wars are back. Hopefully, this is a much healthier war than the Microsoft vs. Netscape-battle of the late nineties, since at least three of the four participants in the battle seems to be strongly committed to open standards.

Previous Post
Comments are closed.