It's hard to believe that people are so up in arms over Firefox having such a bad week. News flash, folks: it topped the charts for one week because it had two vulnerabilities come up at the same time. Guess what: it had to happen eventually, just as going several weeks with none has happened more than once already. The fact that Microsoft's IE also lucked out at the same time, with a week in which it only had one vulnerability come to light, doesn't make Firefox "less secure" than IE.
What it does do is make the Microsoft "zealots" a lot more annoying.
Firefox is undeniably more secure than IE, for a number of reasons. First and foremost, it's not tied directly into the OS, and thus doesn't grant a direct pass-through from the World Wide Web to your OS kernel processes. Second, it doesn't grant arbitrary remote code execution through ActiveX, one of the worst misfeatures in the history of Microsoft. Third, its vulnerabilities get patched with immediacy, in a few days at most, whereas with IE you're lucky if it gets patched in less than a month, and there are all too common cases of vulnerabilities surviving unpatched for six months or more. In fact, according to Secunia, there are currently 19 unpatched IE 6.x vulnerabilities, dating as far back as 13 March 2003, and an additional ten vulnerabilities that have received only "partial fixes". Meanwhile, those two minor vulnerabilities that everyone has been so excited about for Firefox were patched with the release of Firefox version 1.0.4, mere days after their discovery.
As for Linux . . .
The situation is even more lopsided for Linux vs. Windows than it is for Firefox vs. IE. For one thing, all of IE's vulnerabilities are also Windows vulnerabilities, because IE is integrated directly into core OS functionality in Windows. This allows for rapid and easy privilege escalation for intruders and easy circumvention of security measures meant to stop arbitratry remote code execution. In addition to this, there are the software architecture characteristics of the OSes themselves.
On one hand you've got Linux. Linux, like any unix, is a true multi-user system, with strict privilege separation and sharply defined separation between "userland" and kernel processes, rendering privilege escalation almost impossible outside of social engineering (which does an end-run around technology entirely to take advantage of user stupidity, anyway, and equally affects users of any OS, assuming equal stupidity). There's also the difficulty of arbitrary remote code execution, as Linux, like any unix, doesn't play user-obsequious games with trying to execut every bit of code into which it comes in contact, based on a file extension. This is how image-embedded and macro-based malicious mobile code gets propagated, and it's nearly impossible for it to do so on a unix system: the user basically has to tell it to run with a conscious decision, thus bringing us back to social engineering as the major exploit of the system's weaknes (the human). Even if you get the user to execute some malicious piece of code, it only runs with the permissions of the user which, if you'll remember from my previous mention of strict privilege separation, means that malicious code is very limited in what it can do, thus essentially emasculating what might otherwise be a very dangerous virus. A virus designed to wipe out the boot sector and file allocation table of a drive, for instance, that is only run from a normal user account will simply fail to execute properly. No harm done. All data intact.
On the other hand, we've got Windows. Windows does not have strict privilege segregation. It's not a true multi-user system. It is, in fact, a single-user system with a user profiles application laid over the top of it to make it look like a multi-user system to the credulous user. Privilege escalation is extremely easy in Windows because of this. Additionally, because of the interconnectedness of all the kernel-integrated functionality of a Windows system, even when user privilege separation might stand in the way of privilege escalation, hostile code can simply bypass user permissions entirely and just access system functionality through intermediary applications. The canonical example of this is Internet Explorer, though that is by no means the only example. Furthermore, application and service operation in Windows relies heavily on RPCs (remote procedure calls), even for internal and intra-application interprocess communication. It's so bad that MS SQL Server won't run properly without RPC services running, and much of the feature set of the Windows GUI is crippled if denied access to RPCs (for instance, you can't drag and drop with Windows Explorer without using RPCs). Finally, to round out the comparison, Windows happily executes code willy-nilly based on file extensions, automatically and generally without consulting with the user, and since the user privileges bleed across different user accounts, including the admin account, that means that by simply writing a little clever code and naming it with a convenient file extension a malicious security cracker can create a quick piece of code that will own the box in seconds.
Besides, everybody runs with an administrator-privileged user account in Windows, since you basically can't do ANYthing useful in Windows without admin rights (another distinct disadvantage as compared with unices, but tied in with the fact that Windows is fundamentally a single-user system in any case).
Don't forget about that bit about faster response times on vulnerability patches, as for Firefox vs. Internet Explorer. The same people at MS that are neglecting IE are also neglecting Windows. Meanwhile, the Linux kernel gets patches [i]much faster[/i] than Firefox does, if you can believe that. Whereas Firefox gets its patches in a few days, the Linux kernel gets them in a few hours. I've seen patches for Linux posted in less than two hours after they were reported. I don't remember a Windows patch being released in less than a couple weeks, at the absolute best.
If Linux was the inarguable leader on the desktop and Microsoft only had 2% of that market, but all else was equal, we'd still be complaining about the lack of security of Windows, guaranteed.
Keep in mind that Apache, which runs on Linux systems all over the world, holds sixty-something percent of the web server market and gets hammered on a regular basis with new attacks, and yet it is Microsoft's IIS that falls on its face on a regular basis because of its inferior security of design. Popularity is not the sole arbiter of security issues with software. In fact, the more popular open source software becomes, the more people there are finding and fixing vulnerabilities before a writer of malicious code ever finds them. The common wisdom regarding market penetration leading to security issues is reversed when considering open source software development, and that is one of its greatest strengths.