What I found is that if a company has a problem and you sell yourself as the solution, you can ask for the moon. This highly depends on the company. If the company is simply looking for bodies, avoid :)

Even for the exact same job, I've seen pay difference of beyond 50% (even within the same team!). I helped this guy raise his salary from 85k(?) to 130k simply by showing him a piece of paper. (Don't ask)

Oh, another tip, discuss salary with your co-workers, only good can come out of it.

I think this is totally dependent on the position itself. If a company has a good base and needs one super duper person to fill a specific need, they may be likely to pay that one person really well. If a company has a bunch of crap that needs fixing (legacy codebase, lots of context switching between front end, back end, sysadmin, etc.) then that might be the person they want to throw the cash at. Seems to me its a good idea to focus on something and get experience, start to generalize as you gain experience, and then start to specialize again as you find a niche/see what the market demands.

In my somewhat anecdotal exposure to salary data over the years, it seems the best thing for straight cash is to specialize in moderately out-of-date technologies. Big, stable companies with lots of money need people to fix and maintain their legacy systems and the ranks of skilled/interested/breathing tend to thin out faster than the demand declines.

At some level of leadership it seems to me that having broad experience is better, but in my experience first-level leadership roles usually go to specialists and second-level roles to go first-level leads etc.