Emacs's keyboard macros are one of its weaker areas, IMO. C-g cancels recording, but if you accidentally invoke the wrong command and end up in a minibuffer prompt, C-g is the reflexive way to get out of it - terminating the record. So you need to enter your command sequence with zero errors and no use of C-g for it to be of any use. I find multiple cursors more usable in practice, since you can see the results of various navigations simultaneously (at least for those instances that are visible on screen).
I use emacs because it functions the same on Windows, Linux, OS X and Solaris, all the platforms I use across work and home. I use it in the terminal rather than the GUI, because then it functions the same whether over ssh or locally - and no, TRAMP is not quite good enough. And finally, I use emacs because I've had it with learning a different IDE every couple of years. At this point, I want to invest my muscle memory and macro library in something with some longevity. I want something where getting my settings and modules all installed is no more effort than a git clone.
If I could pay someone for something that did all this for me, I would, but the solution would look a lot like emacs: highly scriptable, with a consistent terminal interface across all 4 platforms I use. Ideally it would use a much stronger, faster version of Lisp, although its dynamic scope lookup is actually convenient for layering new behaviour on top of existing code.
(On switching buffers, I use helm + projectile, which does incremental filtering search over file names. For navigating projects, I find helm-git-grep is often more useful than trying to remember a file name - this is an incremental regex search across the whole project, using git grep. No other IDE I use is quite as quick to switch buffers as my emacs config.)