Apple probably doing dozens of things right and dozens of things wrong when it comes to developers.
I would like to see more support and focus on very small firms that have good ideas ... perhaps something like the SBIR grant structure.
Really small firms (1 to 5 people) have severe cash flow problems in the early phases and most of them never survive. One can imagine small competitive grants based on proposals, code samples and early versions of the software. Fifty grants on the order of $20k each would probably result in at least ten or twenty interesting products that survive. I would propose the grant review be done by an external group of volunteer experts from the developer and user community. (say academics for the developer community to minimize conflict of interest).
There are many other areas, but in the spirit of keeping this short, I'll quit here.