It is. Let's say Ubuntu takes over Windows and becomes the dominant OS. To do that, it should have a way to easily install applications. So you go to scam/malware/adware page X while looking for porn (I know nobody does that) and it says your system is outdated and at risk. To update it just download this .deb file and install it. Now user Joe using Ubuntu will blindly click on the deb file, put its password when asked and the malware will be installed in your Ubuntu system just as easily as it is in your Windows one. The difference is that right now, almost nobody bothers to package malware as .deb packages as they do with .exe files. I've seen Android phones full of malware, more so than many Desktop PCs running windows. And it's based on Linux and it's supposed to be safer. Last week a friend told me to have a look at his MacBook Air because it acted weird. I don't use Mac very much but I was really surprised to find it full of malware and adware just like my other friend's Windows PCs. There's nothing preventing the average user from ruining their Desktop computers. It's a matter of people wanting to target the average Joe user and it's done. Malware is a social and educational issue, not so much a technical one, so a world ruled by Linux desktops will not do much to make it safer.

Just out of curiosity because I've never had a Mac and I've been using Linux for many years. I understand from a consumer point of view that a Mac is simply more integrated and beautiful so it attracts more standard users than Linux might but for a power user which uses Mac, what are those needs that any Linux desktop should meet for you and that they don't right now?

Everything is fixable by adding patches and complexity and forcing really difficult and over engineered models on top of something that will be very rarely used. Now I have to disagree with you. As a developer, breaking API compatibility is sometimes a relieve, allows you to advance faster and design better solutions when the requirements have changed significantly from the original ones. The thing is that 99.9% of computers nowadays have a local graphic card capable of 3D rendering and that 99.9% of people don't care about network transparency so breaking an API designed around a model which is not needed anymore sounds like a really good idea to me. Specially if the major players like KDE and GNOME groups are on board and will support the new graphic system.

"However, we have an option here of pushing X out of the hotpath between clients and the hardware and making it a compatibility option." (Wayland FAQ) That X can be patched for the Nth time and add more bullshit on top of it just to be able to cope with modern systems doesn't mean it should be done. X is centered around network transparency which is not used in 99.9% of the times in modern systems so wayland is doing what's right: simplify all this mess so doing composition and hardware accelerated operations easy in 99.9% of the systems in which the GPU is in fact in the same machine of the display and leave all the old X code as a compatibility option just used by this remaining 0.1% of people. It's about simplification and maintainability.