> Pumping CO2 into the ground sounds like it works great on paper, but I think we're fooling ourselves that it won't reenter the atmosphere.
Sequestration alone shouldn't be the solution, but if we can find a thermodynamically feasible way to store CO2 for at least thousands of years while getting away from new CO2 emission, it won't matter that the solution isn't fully permanent. Some proposed solutions are expected to be stable on long enough timescales that the CO2 can (under proper geologic conditions) turn into rock. Other solutions rely on keeping it in oxygen-poor environments where it won't oxidize (which is what happened with carbon that became fossil fuels in the first place; clearly that carbon was not permanently sequestered, since we're burning it now!).
As long as it's thermodynamically feasible to store carbon (read: our net carbon emission goes down and it doesn't cost too much in alternative energy sources), a relatively short (on geological timescales at least) storage solution is more than sufficient (again, as long as it's coupled with critically necessary reductions in new CO2 emission).
[edit] There are, to be clear, many geological formations that are more than stable enough to store volatile materials on longer timescales than are necessary (look at how effectively salt domes trap oil through a combination of buoyancy and impermeability of certain types of rock). The deep, possibly inescapable problem is the thermodynamic difficulty of extracting CO2 from the atmosphere (where its concentration is low, and hence extraction is inefficient) vs. the relative ease of releasing energy from highly-concentrated carbon stores like fossil fuels. It's like trying to unmix your cream from your coffee; it's just much much easier to avoid pouring the cream in in the first place. As others have mentioned, we can rely on the fact that nature already gives us cheap distributed solar farms with built in sequestration abilities (also known as "plants"); non-biological methods don't come with the built in solar energy converter, and so we need to consider the harrowing reality of thermodynamics when trying to make them scalable.