This is basically a question of, "What do you wish could be automated in font production?"
Frankly, I'd like to automate almost everything. Type design should be about expressing an idea regarding shapes, proportions and spacing graphical signs in a medium-specific way. It shouldn't be about spending hours and hours doing mostly repetitive production work.
But we're hampered not by tools but by digital font formats that were invented to ease translation of existing designs from previous technologies, rather than creation of new designs. Hence, we're still working with glyph architecture definition that is completely ignorant of what is being defined, i.e. there is no way to label the parts of a structure such that they can be independently manipulated across glyphs with shared features. The TrueType instruction set is as close as we've come to this, in that one is able to identify common stem weights and alignments through CVTs and adjust them, which is why I wish it were more of a paradigm for outline manipulation in tools. But that's only a move in the right direction: the long jump is to a glyph description language capable of expressing shapes at discrete levels of complexity, such that large glyph sets could be generated from small sets of modular elements and rules.
John: I expressed the very same longing just a few weeks ago on Twitter, using a method similar to how one hints TrueType for controlling features font wide. It is indeed an interesting idea.
All very sensible suggestions: too much of the font production process is hampered by accommodation to legacy, and the unfortunate belief that Postscript offers any real advantage other that more elegant outline rendering,