User agent spoofing absolutely isn't
useless, and is unfortunately a necessity for Linux users. I could give you a long list of websites (and in some cases quite popular ones) that will redirect Linux users to a page saying their browser is unsupported and to update to a newer version even though they're already using the newest version of Chrome or Firefox. Even Gmail did this to me for a while. The vast majority of these are probably unintentionally slighting Linux users because the Web developer incorrectly parsed the user agent string and didn't account for OSes other than Windows or OS X. Spoofing the user agent to report the OS as OS X or Windows is a simple fix for these stupidly designed web sites, and every Linux user I know does it, at least occasionally, since it's not always possible to simply avoid those websites. Google would tell you I'm using OS X right now instead of Ubuntu since I got tired long ago of switching my user agent string whenever a came across a browser-specific site that actually wasn't.
Fortunately this is gradually becoming less of a problem as more web developers are starting to get a clue and actually designing their sites using something called "standards" rather than using stupid browser-specific hacks that require them to parse the user agent. The slow death of Flash and Silverlight, the gradual adoption of HTML5, an increasing effort by browser developers to conform to accepted standards, and an effort to educate web developers about the perils of using browser detection instead of feature detection (http://goo.gl/GNNV9
) might finally put an end to browser detection, but I doubt it. Stupidity always finds a way to survive.
As far as Web statistics are concerned, if you think they're in any way accurate, you really have no idea how those statistics are derived or how the Internet works. Setting aside spoofed user agents, they're still completely incapable of accurately determining information about individual users due to the nature of the Internet. Some of the stats I've seen are derived from unique IP addresses rather than hits, meaning a thousand users behind a proxy would only be counted as a single user (if a very active one that uses several browsers at once). I strongly suspect this is skewing the stats a bit considering the large number of Linux users on business and academic intranets, such as over 20K of Google's 50K employees (http://goo.gl/EC6xp
). (Btw, I think it's quite amusing that Google employees have to ask permission
to use Windows due to its "'special' security problems".)
Even if the statistics are based on hits, many of these use marketing or tracking services that use third-party cookies or other web bugs to acquire information across a number of websites, which is good in theory but in practice means that anyone using something like Adblock Plus (which in my experience is most people who use Firefox or Chrome) won't be counted or would be counted incorrectly. If Google or Facebook would release their browser statistics, then we might
possibly get something approaching reality, but even then it's only an educated guess.
And before you post another long reply trying to explain how things really work... I have a Master's degree in CS specializing in IS, have worked in web development for years, currently teach college-level CS, and am a long-time Linux user. I'm certainly not saying I know more about what I'm talking about than you or anyone else, because I don't know anything about you or them. Just saying that I might possibly have a little bit of knowledge in this area. Maybe not, but I'd like to think so.