1. You download virusBin in Linux, you cannot run it- it does not have execute permission by default.
2. OK so either graphically or at the terminal (chmod u+x) you make virusBin executable.... (I'm not sure why you did this...) If you run it, all of its access outside of /home/$USER is read-only, the system and other users are safe, maybe your data is not.
So if you really want to mess up your machine 1. Download virusBin 2. Give virus Bin execute permission 3. Assuming you are the systems admin, you can now run it as root... sudo ./virusBin..
To get a user to do all that it is going to take a LOT of social engineering.
The question is, how did the executable get there in the first place? Most of the explots that really hit windows are bugs in browsers / addins.
A lot are also spread via 'naughty things', to view this really degrading porn of some celebrety you must first do this survey and execute this code. Or even bundled in with pirate versions, over a few beers a buddy who works for a small design software firm was laughing at how many people had become infected by a virus which was spread by just one torrent of their product (not by him!).
Or like 419 scam emails, really obvious things, that weed out the people with an ounce of sense to leave only those dumb enough to run it. People like my Dad, who typed in the admin password because he was running as regular user :( Then threw a hissy fit when I suggest he not know the admin password in future.
Sure administrator as default is dumb, really dumb. But most of these attack vectors would still be open on linux, in fact we've seen some for the bsd based OSX last year.
"The downside is the user is prompted in a manner that allows them to override this."
Exactly right. Since both of those prompts are "click OK to continue" types it makes the barrier they present is much lower than Linux where you have to change a file permission, then type your password again (and probably from a terminal to use sudo). I would say the barrier is so low they seldom work from what I have seen.
My family members who run Linux don't even know how to use sudo... they just use the Ubuntu software center/synaptic.
"in fact we've seen some for the bsd based OSX last year" I have not used OSX in 4 years so I can't comment meaningfully on this. My gut tells me Apple probably takes a lower barrier approach like MS, but I don't know.
I think that, once smaller developers start getting their apps signed (I'm not sure how hard it is to do this) then this will be a wonderful tool, and already probably is nice for non-power users. I currently leave it disabled because enough of the stuff I use isn't signed, but in the future it might be a nice step forwards for security.
Most things I've seen spreading are already infected executables of pirate code. Or exploits.
Now we can talk about the fact running as Admin is stupid, and UAC is just a way to allow bad habbits to continue. But, just look how complex (and sometimes elegant) the average entry in pwn2own web browsers are. Side stepping ASLR, breaking the sandbox, privlidge escalation.
All of those things can and are possible on any operating system, this includes linux.
Virtually all desktops are either single user.
I exploit a bug in any of your networked client software to run locally. Since it's running under X11, I can now log all of your keystrokes. If you ever run sudo and type in your password, I have root. On the other-hand I don't really care about running as root, since most of what a botnet is used for can be done with non-root permissions anyway.
Then to type your password when the usual dialog appears.
What if there is a security bug that permits escalation? Windows defaults to limited permissions now also BTW.