Solar and wind power both have four common pitfalls: production is unreliable, power can't easily be stored, distribution is a pain, and all of those make the job of reliable distribution slightly harder. The last issue is one of least obvious and important.
Power lines are not just conductors. They're conductors lying side by side for miles and miles, which makes them pretty significant capacitors. To get transmission going on a newly-connected cable, you have to "charge the line" and overcome it's natural capacitance, so switching connections in and out is not trivial. They also are significant inductors. Any change in current is resisted, so sudden changes in current would be impossible if we didn't do some sort of power matching. The idea of power matching is that you can design a system so that at exactly one frequency the impedance from capacitance and inductance cancel out. This is why all power is at a constant frequency (60 Hz in the U.S., 50 Hz most other places).
Most power is produced by some sort of turbine, which means that to keep power at exactly 60 Hz (say), you have large pieces of machinery rotating at a multiple of 60 cycles per second. You can't just shut these down and spin these up at will. Any time you increase the load or lighten the load, you have to either disconnect turbines (which means waste from connecting them again later), or sink significant amounts of power into speeding or slowing them to keep power at your matched frequency. The fact that this works even remotely well is amazing. The fact that we do it reliably and precisely even more so. (It is so precise, in fact, that most wall clocks used to just rely on the 60 Hz cycle of power for timekeeping.) This is also why the Obama administration proposed lightening the rules on frequency accuracy in a recent pilot program: transmission efficiency would drop, but the savings from not having to juggle around turbine speeds might make up for it.
So now we get to heart of the problem: the physics of distribution is not tolerant of change. More than anything, the grid needs a steady, constant demand. Introducing irregular sources of power and sudden drops in demand can put just as much strain on the system as sudden spikes in demand. And the prohibitive capacity/size of capacitors and inefficiency of batteries means that all but a negligible amount of power must be consumed as it's produced.
Right now, they try to use fuel sources that have long lag times (coal, hydropower) all the time, and sources that require less lag (natural gas) to help ease the transition between peaks. They're also trying to sculpt demand, which is basically all a "smart" thermostat does when it refuses to turn on the AC at peak hours. That's about as smart as we've gotten on a large scale. In general, production cannot change until distribution changes. Distriubtion cannot change until storage changes. Storage cannot change until we solve hard (perhaps intractably hard) problems in physics and chemistry.