For answers there is a system that detects certain things in new users posts and if it looks like it's poor quality, they're forced through a quick how to answer guide. This invariably fails at stopping people posting "I have the same problem". But it's a nice thought.
I have a lot higher tolerance for colloquial, even off-topic comments. This is mainly because they're only very rarely an issue and when they are, it's simple to clean up. Comments self-compact and are small compared to answers. They're also not available until people have been on a Stack Exchange site for a while. It's hard to get to 50 points if you don't do anything worthwhile.
The Ask-a-Question page has links to the FAQs and it even does a fairly good job at spotting duplicates... But people ignore it.
As a great philosopher once said: Website users are just like cats.
I'm a web developer and I've been dealing with "users" and how they interact with websites for about 8 years at a professional level.
Is it ever possible to predict what a user will do? Almost certainly not. Even if you put arrows, warnings and bells up around something, they're going to stumble in and start ripping up your curtains.
First of all, they don't need documentation because cats can't read. They don't need to because every web-surfing kitty these days has innate knowledge of how a website works. You click here, oh there's a text box, text in, post, done, catnip mouse plz.
The more warnings and FAQs you put up, the cooler they look when they nonchalantly stroll past them. They're the centre of the Internet at that point in time. FAQs read them, not the other way around.
If you try to over-complicate things and try to herd your cats, they'll give up, walk away and have a nap. They'll probably accuse you of being an elitist dog-sympathiser too. That is how defamation lawsuits start.
The Stack Exchange system understands all this. We have good tools to try and spring as many helpers up as is automatically feasible and when they do post something awful, flags are raised for moderators and the trusted users.
At that point we can gently say "No kitty, that's a bad kitty", and point them in a better direction. But by this point, we've got a brand new user inside the system. They're logged in to some degree and they're learning about how it works.
The hope from that point is we turn them from a one-post-wonder into a regular user That's a reason that if you see somebody doing something wrong you should interact with them instead of just flagging it. Help them out, even if it's yet another I get this too.
We do appreciate the flags (and we'll comment if somebody else hasn't) but good, fast communication is the way to get people involved here.