the sockets actually tend to break internally. micro usb is so small that the contact fingers of the female end just don't have the longevity that the larger connector standards do. honestly, if you're not space constrained (and keyboards are NOT space constrained for the most part :\) then the standard USB A socket is just the best choice bar none.
the existence of mini and micro USB 2.0 jacks is entirely due to device makers wanting increasingly thinner profiles, and not for any reason that one might consider to be mechanically sound. the mini USB 3 jack helps quite a bit by making the connector significantly wider and giving it a reinforcing pinch seam, but the connectors and cables are more expensive, and the connector takes up about 3x as much board space.
which bring us to micro lightningbolt. it's actually quite a nice mechanical design. the male end doesn't rely on any outer chassis; it's just a two sided conductor carrier. the female end can then exert a much larger amount of pressure with longer, mechanically stronger mating fingers. the connectors are going to be SMT, frankly, as we're getting closer and closer to just using flexible circuit substrates, and those won't offer enough structural integrity to be worth thru-holing. so, the best thing you can do is provide as much give _within_ the female jack as possible. this minimizes the pressure that the jack exerts against the circuit substrate, which is the next point of possible failure.
anyway, the bottom line unfortunately is that things are going to get worse on the connector front before they get better. device manufacturers are pushing for really really thin connector standards, basically to the point of absurdity, and the connector manufacturers are just kind of shrugging and doing the best they can within target pricepoint.
i do see a couple of really good trends that will hopefully end this garbage. the first is that short-range 2.4ghz bluetooth transceivers and txrx stacks are getting extremely reliable, and BT 4.0 is finally becoming flexible enough to be useful. once wireless datacom is sufficient, the only wires you need are for power. this suddenly makes the problem much easier. 5vdc @ 500ma is the defacto power supply standard, and this is weeny enough that you could put it on a 4th conductor on a TRRRS headphone jack, or take the existing micro usb standard and make the current tiny complex conductor block a single +v and the entire metal chassis a gnd that is potted down to a huge single ground pad on the substrate.
ps, wireless power would be more realistic and much more fun if it were transferred with tiny centrifuges that spun your device until a small onboard MEMs transducer produced enough electrical charge for the battery. as a plus, you could flash the screen at the spin rate to display pictures of cats. seems like a win win win to me.